Arthur Rock

1. Snapshot and testable edge thesis

Arthur Rock moved from 1950s investment banking through Davis & Rock, a Rock–Kramlich arrangement, later Rock-branded entities, personal positions, syndicates, and long board roles. Those are different economic actors, not one continuous fund. [researcher inference · strong] Vehicle analysis

Testable thesis. [researcher inference · moderate] In U.S. technology-company formation from roughly 1957 through the 1980s, Rock's repeatable edge was the combination of trusted access to proven teams and capital, longitudinal people assessment for intellectual honesty, and concentrated financing/governance work that accelerated capital formation and negative-information flow. HBR, diligence and control passages, printed pp. 63–67 · Intel memorandum, PDF pp. 1–4 · Something Ventured part 2, printed pp. 10–13 and 28–29

Mechanism and scope. [researcher inference · moderate] Warm access should improve the opportunity set and financing speed; repeated observation should reduce avoidable people error; a small active portfolio should enable recruiting, accounting, escalation, and stop decisions. The mechanism should be strongest when first capital is scarce, trusted references are informative, and a director can change company systems. CHM Evening, printed p. 15 · HBS, printed pp. 8–16

Evidence status. [researcher inference · moderate] Fairchild, SDS, Intel, and Apple support distinct parts of the mechanism; Diasonics, Ridge, ROLM, Compaq, and Rock's admission that roughly half of selected investments disappointed show large residual error. Eight decision labs · Failure table

Denominator boundary. [researcher inference · strong] The public set contains 26 named families—16 capital/ownership, four intermediation, two board/acquirer controls, two passes, and two association/failure controls. All 26 were searched/researched, but 0/26 has a complete vehicle, entry, security, ownership, follow-on, exit/write-off, and cash-flow chain; career coverage is not computable. DEALS reconciliation

Predictions. [researcher inference · moderate] Unseen records should show repeat referrer/founder channels, repeated-contact diligence, few active companies, fast syndication, board-level recruiting/accounting interventions, and better ordinary-case outcomes when those mechanisms were present. Falsifiers. [researcher inference · strong] The thesis fails if full decisions and cash flows show no incremental selection or outcome after controlling for access, founders, partners, vehicle, vintage, and outliers; if board work was nominal; or if results vanish without SDS/other dominant companies. Portfolio limits

2. Core philosophy

Rock's own language is stable: ideas are plentiful, execution and good management are scarce; diligence means listening over time; intellectual honesty means seeing conditions rather than desired outcomes; leaders must know what they cannot do. [investor-stated · strong] HBR, opening and diligence passages · Steve Jobs Archive, paragraphs 1–5

The synthesis is more specific than a people-first maxim. [researcher inference · strong] Select for demonstrated execution and reality testing; structure finance around development risk; create independent channels for bad news; concentrate only enough to perform real governance; redesign management as complexity changes; and make follow-ons or exits separate decisions. Written-corpus synthesis · Spoken-corpus synthesis

The philosophy is probabilistic, not deterministic. Rock's own funnel and luck admission, the ROLM/Compaq misses, and Diasonics prevent a deterministic great-people rule from becoming a causal law. [researcher inference · strong] HBR, opening funnel, luck passage, and final paragraph · CHM Pioneers, 01:09:42–01:10:32

3. Sourcing

4. Picking

Rock's first screen was not résumé prestige alone. He reviewed prior work and mistakes, listened through multiple meetings, tested whether answers stayed consistent, asked whom founders admired and would recruit, and separated company-building motive from wealth-first motive. [investor-stated · strong] HBR, résumé, listening, and “Belief in the business” passages

ROLM supplies the critical false-negative test: a perceived experience gap prompted a pass, yet Rock later hired Ken Oshman at Echelon. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32 A usable people screen must state what evidence reverses the judgment. [researcher inference · moderate] ROLM lab

5. Deal and portfolio mechanics

Precise economic contribution is not computable. SDS is directionally first in the D&R narrative, but removing it leaves incompatible partnership, company-scale, and personal-ownership measures. [researcher inference · strong] Pareto stress test

6. Vehicle and incentive fit

[researcher inference · strong] The strategy and the vehicle cannot be separated: small pools made intensive governance plausible, finite terms created distribution pressure, relationship capital reduced fundraising friction, and key-person discretion made succession fragile. Only governing agreements and cash-flow ledgers can test the strength of those causal claims. Current LP comparison

7. Team attribution and work with founders

Rock's documented value-add mechanisms were arranging capital, drafting/organizing financing, recruiting, interviewing, board/audit participation, accounting discipline, offering preparation, CEO oversight, and deciding whether to continue funding. [researcher inference · strong] Something Ventured part 2, printed pp. 10–13 and 28–29

8. Views on the craft

• Power law and humility. [researcher inference · strong] Rock never published a formal power-law theory in the processed writings; the D&R/SDS narrative and missing denominator nevertheless demand outlier testing rather than fame-weighted storytelling. Pareto analysis
• Fund size and service. [investor-stated · strong] More capital and more boards can reduce time per company; the limiting resource is attention, not only dollars. Founding Funders, “How many boards” exchange
• Access versus selection. [researcher inference · strong] Trusted networks are real advantages but do not prove that Rock picked better than investors seeing the same deals. Talks, trusted-access counterargument
• Governance. [investor-stated · strong] Boards should recruit, test information, oversee CEOs, preserve audit independence, and approve adaptation without pretending to run engineering. Levitt letter, PDF p. 1 · Moore–Rock, pp. 19–31
• Exits. [researcher inference · provisional] Long company association and resistance to quick-exit promoters coexist with finite-fund distributions, acquisitions, and conflict-driven resignation; no universal hold-forever doctrine survives. Silicon Genesis, quick-money and Xerox passages
• Great investing. [researcher inference · moderate] The strongest supported definition is repeat access plus judgment plus governed execution, with honest recognition of luck and stop decisions—not prediction alone. HBR, full body

9. Signature bets and decision process

Use the no-hindsight labs rather than the later fame of each company. Every decision-replay question and evidence boundary in this table is [researcher inference · moderate] based on the linked frozen-date packet.

10. Mental models and maxims

The reconciled field guide contains nine supportable models; each link opens its full provenance, procedure, cases, exceptions, and modern translation:

1. Test people longitudinally for intellectual honesty, self-knowledge, and builder motive — [investor-stated · strong] HBR diligence passages.
2. Let earned reputation and trusted intermediaries compress uncertainty—never erase it — [researcher inference · strong] Moore–Rock, pp. 19–24.
3. Match the financing architecture to the development risk and align every capital provider — [contemporaneous record · strong] Intel memo, pp. 1–4.
4. Keep the portfolio small enough that governance is real work — [investor-stated · strong] HBS, pp. 8–9, 15–16.
5. Make bad news travel faster than reassuring narrative — [investor-stated · strong] HBR finance/control passages.
6. Redesign management and information paths when complexity changes — [investor-stated · moderate] HBR three-stage passage.
7. Update on observable evidence without pretending to time what you cannot know — [researcher inference · moderate] Founding Funders, closing valuation exchange.
8. Make each follow-on earn a new decision; an advocate who disagrees funds the continuation — [investor-stated · moderate] Something Ventured part 2, pp. 28–29.
9. Build companies for duration, but separate company quality from vehicle, liquidity, and conflict decisions — [researcher inference · strong] Something Ventured part 2, pp. 24–25.

Plausible but excluded models and the proof missing for each are recorded in Evidence gaps and candidate models.

11. Evolution and contradictions

Every unresolved-tension cell below is [researcher inference · moderate]; each tests the labelled continuity claim in the same row rather than asserting an investor-stated doctrine.

12. Failures and limits

• Diasonics — [investor-stated · strong]: later-described aggressive sales booking and revenue-recognition problems, competitor financing, CEO weakness, and inadequate board attention make this the strongest monitoring failure; original thesis/terms and later process adoption remain missing. Bancroft, pp. 63–68
• ROLM and Compaq — [investor-stated · moderate]: passes establish false-negative risk; hindsight success cannot reconstruct the proposals or prove irrationality. Pioneers, 01:09:42–01:10:32
• Ridge — [retrospective witness · strong]: founder/VC sale and governance disagreement, later H&Q leadership, and reduced Rock activity show that syndication can diffuse accountability; participant testimony says Ridge did not go bankrupt. Sell, pp. 13, 15, 20
• ELXSi/Xynetics/Qantel — [researcher inference · moderate]: company-level financing stress and continuation evidence exist, but Rock's security, follow-on, stop authority, and cash flow are incomplete. Failure postmortems
• Selection/survivorship — [researcher inference · strong]: 26 public names and 13 unnamed Rock–Kramlich companies do not form a complete funded denominator; visible filings and oral histories overweight famous outcomes. Denominator
• Era dependence — [researcher inference · strong]: low competition, federal semiconductor demand, company-formation scarcity, and different security/exit regimes limit transfer. Historical context

13. Historical operating environment

Fairchild emerged before the 1958 SBIC regime, in a semiconductor market shaped by federal R&D and procurement; Rock did not create the technical team or demand environment. [researcher inference · strong] CRS, “Background” and federal-policy sections · Small Business Investment Act, §§102 and 301 / pp. 2, 6–7

D&R operated among private LPs, SBICs, ARD-style investment companies, corporate/family capital, banks, and underwriters before pension capital became a dominant venture source. [researcher inference · strong] Federal Reserve Staff Study 168, printed pp. 7–15 The 1979 prudent-man clarification later widened the institutional-capital environment but cannot explain Rock's earlier access. [contemporaneous record · strong] Federal Register, printed pp. 37222–37225

Intel's 1968 packet shows what a contemporary could know: founders, product scope, expected development period, uses, proposed security, incentives, board, and conflicts—not later microprocessor dominance. [contemporaneous record · strong] Intel memo, pp. 1–4 Apple's 1978 packet similarly exposed shipments, dealers, product roadmap, staffing, manufacturing, risks, and capital needs—not its later IPO value. [contemporaneous record · strong] Apple memo, pp. 1–34

Later fund studies document selection, survivorship, persistence, and vintage effects, but they do not estimate D&R or Rock-personal returns. [researcher inference · strong] Kaplan and Schoar, pp. 1–7, 14–16 · Gompers and Lerner, pp. 1–10

14. Modern VC translation (as of 2026-08-19)

All rows are analyst translations, not Rock-attributed current advice.

15. Comparison with completed peers

Reid Dennis is the closest same-region, overlapping-era syndication/governance peer; Laurance Rockefeller is the family-capital and Venrock-as-Apple-co-investor contrast; Georges Doriot is the public closed-end/permanent-capital and operator-development contrast. The first four columns summarize ordinary documented facts at their cited links; each bounded conclusion is separately labelled. Legal owners, periods, and return methods prohibit a performance ranking. [researcher inference · strong]

16. The strongest case against greatness

Skeptical thesis. [researcher inference · strong] Rock's reputation may be the product of privileged access to extraordinary founders, scarce early capital, favorable semiconductor/computer regimes, team and vehicle work attributed to one famous financier, and a few visible outliers—without a complete personal or vehicle return record proving repeatable outperformance.

1. Access before selection. [researcher inference · strong] Kleiner, Noyce/Moore, Markkula, portfolio executives, and known capital providers prefiltered the opportunity set. People map
2. Founder/team causality. [researcher inference · strong] Fairchild's eight, Palevsky, Singleton/Kozmetsky, Noyce/Moore/Grove/Vadasz, Jobs/Wozniak/Markkula, co-investors, and employees owned much of the operating work. DEALS people map
3. Missing denominator/economics. [researcher inference · strong] 0/26 is fully reconstructed; the 13 Rock–Kramlich companies are unnamed; no personal return ledger exists. Reconciliation
4. Outlier dependence. [researcher inference · moderate] SDS directionally drives D&R's story, but conflicting reports prevent a clean ex-outlier test: reported: $3.0 million invested and $100 million returned (HBS profile, “Venture Capital” and results paragraphs, 2001); reported: $3.4 million invested and nearly $77 million returned (Mallaby, Rock chapter, printed pp. 17–39, 2022); reported: $3.5 million raised (Kenney and von Burg, printed p. 142, 2000). No multiple is calculated.
5. Residual error. [researcher inference · strong] Roughly half of selected investments disappointed; Diasonics, ROLM, Compaq, Ridge, and unnamed ordinary cases defeat a uniformly superior-judgment story. HBR · Failures
6. Era effects. [researcher inference · strong] Federal demand, young technical markets, low venture competition, changing public markets, and later institutional-capital expansion supplied conditions not created by Rock. Historical context
7. Vehicle and succession. [researcher inference · moderate] Finite terms, opaque later entities, Kramlich's control critique, and no visible succession system can make a strong personal craft non-transferable. Kramlich, pp. 4–7
8. Retrospective narrative. [researcher inference · strong] Repeated Rock interviews are correlated retellings; repetition proves stated consistency, not causality or efficacy. Source-map archive assessment

Evidence favors the edge thesis if original memos, reference records, votes, intervention logs, and full cash flows connect Rock-specific actions to better ordinary-case outcomes after access, team, vehicle, era, and outliers are controlled. Evidence favors the skeptic if roles are mostly nominal, selection resembles peers seeing the same deals, ordinary-case economics are weak, or results concentrate in exceptional founders and vintages. [researcher inference · strong] Profile falsifiers

17. Analyst study guide, glossary, and go deeper

Recommended reading path

1. Start with the biography and write the edge thesis, alternative explanation, prediction, and falsifier in your own words.
2. Read the 1987 HBR analysis, then the 1968 Intel written control; mark where people-first judgment becomes structure rather than planlessness.
3. Read the spoken synthesis and separate Rock, founders, partners, editors, and later narrators.
4. Complete all eight decision labs before reading each reveal; include the Diasonics failure and ROLM miss.
5. Use the nine-model field guide to turn the historical record into IC procedures, then re-read the skeptical case.

Glossary

Unless a row carries a stronger source label, each definition is [researcher inference · moderate] and applies only within this package.

Reusable sourcing checklist

• Who introduced whom, when, and with what incentive?
• Is the source a founder, financier, operator, co-investor, LP, filing, or later narrator?
• What did the trusted intermediary directly observe, and what is hearsay?
• Which peers saw the same opportunity, and what did they decide?
• Does a relationship expand access, compress diligence, or both?

Reusable diligence and IC checklist

• Freeze the decision date; list known, unknown, disputed, and later-reveal facts.
• Test prior work, mistakes, self-knowledge, recruiting, and response to contradiction.
• Independently test product performance, customer demand, revenue quality, cash, and collectibility.
• State the thesis, alternative, kill criteria, ownership/security, and next-check milestone.
• Name source, sponsor, diligence owner, approver, board owner, and follow-on decider.
• Record what evidence would reverse both invest and pass decisions.

Reusable governance and portfolio checklist

• Does each board seat have time, named work, and escalation authority?
• Can the CFO/auditor/product/customer evidence bypass management filtering?
• Are assurance providers independent from the work they assess?
• Has management/information design changed with complexity rather than a revenue slogan?
• Is each follow-on a new opportunity-cost decision with an explicit advocate and kill case?
• Are company quality, fund liquidity, conflict, and exit price decided separately?
• Can portfolio results be shown with and without the dominant outlier?

Ranked primary and secondary sources

1. Intel financing memorandum, PDF pp. 1–4 — best ex-ante primary control for people, technology, financing, incentives, governance, and conflicts. [researcher inference · strong]
2. HBR, 1987, printed pp. 63–67 — best direct method statement, including funnel, controls, failures, and luck. [researcher inference · strong]
3. Bancroft oral history, printed pp. 1–68 — deepest chronology, direct philosophy, vehicles, boards, and Diasonics; quote restrictions and retrospection apply. [researcher inference · strong]
4. Something Ventured Rock interviews — richest operational mechanisms and stop-funding logic; session-date conflict remains. [researcher inference · strong]
5. Moore–Rock panel, printed pp. 19–31 — best paired attribution control for Intel plan, operating work, options, and memory decision. [researcher inference · strong]
6. Markkula oral history, printed pp. 21–29, 43 — strongest Apple sourcing/underwriting countervoice. [researcher inference · strong]
7. Apple filing, SEC FOIA PDF pp. 33, 36–38, 57 — primary allocation and board-date control. [researcher inference · strong]
8. Kramlich oral history, printed pp. 4–7 — essential team, economics-boundary, control, and succession countervoice. [researcher inference · strong]

The source map ranks the decisive missing records: Fairchild contract/prospect ledger, D&R and Rock–Kramlich agreements/cash flows, Intel closing/minutes, Apple purchaser/minutes/transfers, Diasonics audit/board/position records, ROLM/Compaq proposals, the controlled SHI body, and Stanford V0311.

Pareto 80/20 — what drove the record

The public record does not support a literal 80/20 calculation. No complete portfolio, security-level cash-flow ledger, or Rock-personal realization schedule is public. The rankings below therefore separate three different questions and use only directional language. [researcher inference · strong] Source-map denominator and retrieval boundary

Economic contribution — directional, not a personal-return ranking

Removing SDS makes the ranking much less defensible: Teledyne, Intel, and Apple use incompatible measures—partnership value, company scale, and documented personal ownership—and none has a complete Rock cash-flow series. [researcher inference · strong] Economic concentration excluding SDS is therefore not computable. D&R evidence conflict

Historical importance — ranked independently of economics

Decision-learning value — ranked for replayability

Evidence boundary and searched denominator

Scope and rules

• A candidate lead enters when a named source associates Rock or a defined vehicle with financing, beneficial ownership, board/governance, or an explicit pass/failure. A candidate is not automatically an investment. [researcher inference · strong] Source-map inclusion rule
• One company family is counted once; renamed entities, follow-ons, public offerings, acquisitions, and later board service do not create new investments. [researcher inference · strong] Source-map role boundary
• Rock personally, Hayden, Stone, Davis & Rock, Rock–Kramlich, Arthur Rock & Associates, Arthur Rock & Co., syndicates, trusts, and portfolio-company boards are separate actors. Unknown vehicle, security, check, ownership, follow-on, exit, or cash flow remains unknown. [researcher inference · strong] Profile vehicle map
• Company sale price, IPO valuation, public-market value, or a vehicle-level result is not Rock's realized return. A numeric multiple or IRR is used only in the required reported: form with source, locator, and year. [researcher inference · strong] Source-map return conflicts

Reconciliation

The searched public set currently contains 26 named company/opportunity families. The count is a discovery denominator, not a funded-portfolio or return denominator. [researcher inference · strong] Source-map candidate ledger · T006.4 retrieval boundary

Coverage fraction is not computable. Rock's 1987 essay reports about 300 plausible plans per year, one or two investments, and roughly half of selected investments disappointing; Kramlich recalls an unenumerated 13-company pool. Those statements are incompatible with treating 26 public names as a complete career denominator. [investor-stated · strong] HBR, printed pp. 63–67 · [retrospective witness · moderate] Kramlich, printed pp. 4–7

Search boundary

The sweep searched all 26 named cases through 43 priority DEALS/control source families, plus SEC EDGAR, company histories, court opinions, CHM/Stanford/SHI oral histories, period press, acquisition records, bankruptcy routes, and exact Rock/vehicle/company/pass/failure/ownership combinations. [researcher inference · strong] It did not recover a complete D&R portfolio, Rock–Kramlich company list, fund agreements, capital calls, distribution ledgers, board minutes, or person-level tax/realization records. Source-map archive exhaustion

Historical market, regulatory, and base-rate context

Vehicles, economics, and incentive boundaries

Dimension-matched comparison with a current LP-backed venture fund

This comparison is an analyst control as of 2026-08-19, not a claim that historical vehicles should have copied current documents. [researcher inference · strong] NVCA model-document scope and tailoring boundary

People, capital, and influence map

Comprehensive discovered-deal ledger

Decision labs

Decision lab 1: Fairchild — structuring a company around eight engineers

• Freeze — 1957 decision: Reconstructed from later participant testimony because no public decision memo was located: eight engineers wanted to leave Shockley, initially sought another employer, and had technical capability but no independent company or conventional venture pool. [retrospective witness · strong] Last, printed p. 4
• Known: The team was intact; numerous prospects had declined; Rock and Coyle were placement intermediaries, not operators. [retrospective witness · strong] Last, printed p. 4 · Bancroft, printed pp. 25–31
• Unknown or disputed: Customer pipeline, unit economics, founder hierarchy, legal structure, and the exact prospect count—20–30, about 35, or 45—were not recoverable as one decision-date packet. [researcher inference · strong] Source conflict ledger
• Alternatives and dissent: Join an established employer, continue searching for a corporate sponsor, form a new company, or abandon the group transaction; prior declines are evidence of dissent, not proof the opportunity was bad. [researcher inference · strong] Last, printed p. 4
• Terms, milestones, follow-on, and governance: A later contract reconstruction supplies the proposed founder/employee shares, Fairchild Camera loan/options, and voting trust; authorized versus drawn capital and milestone releases remain unclear. [documented behavior · moderate] CHM, PDF p. 9
• Decision exercise — invest/arrange or pass: Choose among the four alternatives and name the vehicle. [researcher inference · strong] Fairchild packet
• Decision exercise — kill criteria: Define technical-team fracture, sponsor-control, customer, and capital-availability failures. [researcher inference · strong] Fairchild packet
• Decision exercise — ownership / terms: Allocate founder/employee incentives, sponsor options, voting control, board rights, and maximum capital. [researcher inference · strong] Fairchild packet
• Decision exercise — next-check evidence: Require an executed agreement, customer/market evidence, hiring plan, and capital-draw milestones. [researcher inference · strong] Fairchild packet
• Reveal — decision and outcome: Rock/Coyle proposed a new company and Fairchild Camera financed it; founders and employees built the semiconductor company. [researcher inference · strong] Bancroft, printed pp. 25–31
• Post-reveal lesson: The supported mechanism is team-preserving transaction design, not a Rock-personal investment return or sole-founder narrative. [researcher inference · strong] Fairchild ledger row

Decision lab 2: Scientific Data Systems — underwriting a new computer company

• Freeze — about 1961: Later testimony, not a contemporaneous memo, says Palevsky proposed a scientific-computer company while D&R had finite pooled capital. [investor-stated · moderate] Bancroft, printed pp. 38–45
• Known: Palevsky had relevant operating knowledge; computing required product, manufacturing, sales, and service capability; D&R could supply startup capital and governance. [retrospective witness · moderate] SDS panel, printed pp. 8–14
• Unknown or disputed: Decision-date market study, security, ownership, customer commitments, and follow-on plan are missing; cost recollections use about $250k/$280k/$300k. [researcher inference · provisional] SDS ledger row
• Alternatives and dissent: Invest through D&R, syndicate, require customer milestones, or pass; no contemporaneous dissent record was located. [researcher inference · moderate] Source-map retrieval boundary
• Terms, milestones, follow-on, and governance: Rock's chair role is later documented, but entry ownership, reserve budget, product/customer gates, and underwriting allocations are unknown. [researcher inference · insufficient public record] SEC News Digest, printed p. 1
• Decision exercise — invest or pass: Choose a capital amount/vehicle and explain why syndication is or is not required. [researcher inference · strong] SDS packet
• Decision exercise — kill criteria: Define founder, product, manufacturing, service, and customer failures. [researcher inference · strong] SDS packet
• Decision exercise — ownership / terms: Set security, ownership, board, protective rights, reserves, and founder incentives. [researcher inference · strong] SDS packet
• Decision exercise — next-check evidence: Require product validation, reference customers, gross-margin/service evidence, and a closing ledger. [researcher inference · strong] SDS packet
• Reveal — decision and outcome: D&R invested and Rock chaired; SDS later completed a 1964 public/secondary offering and Xerox acquired it in 1969. [documented behavior · strong] SEC News Digest, printed p. 1 · SDS panel, printed pp. 8–14
• Post-reveal lesson: Board-and-financing support accompanied Palevsky's operating execution; no audited cash flows show how much D&R or Rock captured. [researcher inference · strong] SDS ledger row

Decision lab 3: Intel — structuring a semiconductor-memory company

• Freeze — 1968: Noyce and Moore brought established semiconductor records but proposed a new memory company. The surviving memo is an ex-ante financing packet, though unsigned and only later attributed to Rock. [contemporaneous record · strong] Memo, PDF pp. 1–4
• Known: Proposed $2.5m convertibles, founder-price common, employee options, uses of proceeds, board design, and disclosed conflicts. [contemporaneous record · strong] Memo, PDF pp. 1–4
• Unknown or disputed: Memory-product adoption, closing subscriptions, exact memo version, and whether Rock's expected $300k debenture allocation closed. [researcher inference · strong] Intel evidence boundary
• Alternatives and dissent: Founder self-funding only, smaller staged round, strategic/corporate capital, wider syndicate, or pass; the memo's conflict disclosure is the visible internal challenge. [researcher inference · strong] Memo, PDF p. 1
• Terms, milestones, follow-on, and governance: Proposed security/conversion, board, option pool, and expenditure plan are visible; product/customer milestones and follow-on reserves are not fully specified. [contemporaneous record · strong] Memo, PDF pp. 2–4
• Decision exercise — invest or pass: Choose common, debenture, both, or pass and name the subscribing entity. [researcher inference · strong] Intel packet
• Decision exercise — kill criteria: Define technical yield, memory-demand, founder-team, conflict, and financing failures. [researcher inference · strong] Intel packet
• Decision exercise — ownership / terms: Set conversion, founder common, option pool, board, consent rights, and syndicate allocation. [researcher inference · strong] Intel packet
• Decision exercise — next-check evidence: Demand signed subscriptions, product/customer tests, hiring, and cash-use evidence. [researcher inference · strong] Intel packet
• Reveal — decision and outcome: Intel history reports Noyce/Moore at $245k each, Rock at $10k common, and Rock organizing the $2.5m syndicate; management built the company. [documented behavior · strong] Intel history, PDF p. 4
• Post-reveal lesson: Separate the proposed $300k allocation from the closed $10k common receipt and separate financing design from operating credit. [researcher inference · strong] Intel ledger row

Decision lab 4: Apple — underwriting a consumer-computer company through Markkula

• Freeze — January 1978 packet: Apple's preliminary confidential memorandum, not later Rock recollection, is the evidence set. It describes 1977 shipments/revenue, over 180 dealers, product roadmap, market forecasts, staffing, manufacturing, and capital needs. [contemporaneous record · strong] Apple offering memorandum, printed pp. 1–34
• Known: Apple II was shipping; Markkula chaired/marketed; the company planned dealer-led distribution and additional products. [contemporaneous record · strong] Offering memorandum, printed pp. 1, 15–25
• Unknown or disputed: Forecast accuracy, manufacturing scalability, consumer adoption, channel service, later governance, and liquidity. The preliminary cover leaves price/proceeds blank. [contemporaneous record · strong] Offering memorandum, printed pp. ii–iii, 1, 19–24
• Alternatives and dissent: The packet itself proposes equity, profits, and long-term debt; risk factors identify short history, case supply, cash-flow pressure, and inexperienced management. [contemporaneous record · strong] Offering memorandum, printed pp. iii, 1
• Terms, milestones, follow-on, and governance: Preliminary maximum 150,000 pre-split shares, sophisticated-investor restrictions, management/staff plan, manufacturing plan, and financial plan; exact investor allocation is reveal evidence. [contemporaneous record · strong] Offering memorandum, printed pp. ii, 25, 32, 34
• Decision exercise — invest or pass: Decide using only the packet and name personal versus pooled capital. [researcher inference · strong] Apple packet
• Decision exercise — kill criteria: Define shipment, margin, supply, channel, management, and cash failures. [researcher inference · strong] Apple packet
• Decision exercise — ownership / terms: Set price, ownership, board/information rights, financing mix, and dilution protections. [researcher inference · strong] Apple packet
• Decision exercise — next-check evidence: Require audited sales/cash, supplier capacity, dealer productivity, service quality, and hiring progress. [researcher inference · strong] Apple packet
• Reveal — decision and outcome: The later filing proves Rock's 640k split-adjusted shares, 1.3%, and October 1980 board appointment, while preserving $0.09/$0.09375 presentation. [contemporaneous record · strong] Apple filing, PDF pp. 33, 36–38, 57
• Post-reveal lesson: The packet supported a product/channel/capital decision; it does not prove Rock was the original VC or realized a particular return. [researcher inference · strong] Apple ledger row

Decision lab 5: Diasonics — testing revenue quality and board attention

• Freeze — about 1977–78, sparse packet: No public decision-date memo, capitalization table, forecast, or entry terms were located. The lab therefore exposes the missing record rather than importing later evidence. [researcher inference · insufficient public record] Diasonics retrieval route
• Known: Public later sources identify a medical-imaging company and Rock in founding/governance; that is the maximum baseline permitted before the exercise. [investor-stated · moderate] Bancroft, printed pp. 63–68
• Unknown or disputed: Product evidence, customers, revenue policy, financing arrangements, management capability, security/check/ownership, co-investors, reserves, board information, and dissent are all missing at decision date. [researcher inference · insufficient public record] Diasonics ledger row
• Alternatives and dissent: Pass; require an independent market/accounting review; stage capital against product/cash milestones; or invest with enhanced audit/governance rights. These are analyst-created alternatives, not documented Rock choices. [researcher inference · provisional] Diasonics packet
• Terms, milestones, follow-on, and governance: None are reconstructed ex ante; do not infer them from the later retrospective record. [researcher inference · insufficient public record] Diasonics ledger row
• Decision exercise — invest, stage, or pass: State whether this evidence deficit itself blocks the decision. [researcher inference · strong] Diasonics packet
• Decision exercise — kill criteria: Define product, customer, cash-collection, accounting, management, audit, and attention failures without assuming they occurred. [researcher inference · strong] Diasonics packet
• Decision exercise — ownership / terms: Specify security, board/audit rights, financing restrictions, reserves, and stop authority. [researcher inference · strong] Diasonics packet
• Decision exercise — next-check evidence: Demand contracts, collections, accounting policy, receivables aging, product comparison, CEO references, and board materials. [researcher inference · strong] Diasonics packet
• Reveal — decision and outcome: Rock later admitted inadequate monitoring, revenue-recognition problems, aggressive competitor financing, CEO weakness, and partial audit-chair responsibility; the court records IPO/loss allegations but no merits finding. [investor-stated · strong] Bancroft, printed pp. 63–68 · [contemporaneous record · strong] Court, 599 F. Supp. at 449–50
• Post-reveal lesson: Missing ex-ante records prevent proving the original thesis; the supported lesson is the monitoring/control failure Rock himself later described. [researcher inference · strong] Failure postmortem

Decision lab 6: ROLM — evaluating an inexperienced team

• Freeze — around 1969, sparse retrospective packet: No proposal, terms, market study, or contemporaneous memo is public; only Rock's later account supports a concern about Oshman's experience/business sense. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32
• Known: A founder/team opportunity was presented and Rock later remembered experience as the decisive screen. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32
• Unknown or disputed: Product, market, customers, capitalization, vehicle authority, ownership, milestones, competing offers, and contemporaneous dissent. [researcher inference · insufficient public record] ROLM ledger row
• Alternatives and dissent: Pass, stage capital, recruit an experienced operator, or syndicate with a domain expert; these mitigations are analyst-created. [researcher inference · provisional] ROLM packet
• Terms, milestones, follow-on, and governance: No public terms exist; any proposed vesting, board or milestone protection is hypothetical. [researcher inference · insufficient public record] ROLM ledger row
• Decision exercise — invest or pass: Decide whether the evidence deficit and experience gap are fatal. [researcher inference · moderate] ROLM packet
• Decision exercise — kill criteria: Define which specific experience, customer, product, or team gap ends the case. [researcher inference · moderate] ROLM packet
• Decision exercise — ownership / terms: Design vesting, board, staged capital, and operator-recruiting protections. [researcher inference · provisional] ROLM packet
• Decision exercise — next-check evidence: Name evidence that would reverse the pass rather than simply confirm it. [researcher inference · moderate] ROLM packet
• Reveal — decision and outcome: Rock passed; ROLM later succeeded; he later recruited Oshman at Echelon. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32
• Post-reveal lesson: A people screen needs falsifiable subcriteria and a documented updating path; hindsight success does not prove the original pass irrational. [researcher inference · moderate] ROLM ledger row

Decision lab 7: Qantel — deciding whether to continue support

• Freeze — 1970s follow-on, sparse retrospective packet: The public record says Rock/co-investors backed Qantel but exposes no decision-date forecast or follow-on memorandum. [contemporaneous record · moderate] TIME, 1984, page 2
• Known: Small-business-computer company, continued support need, and reported Teledyne/co-investor participation. [contemporaneous record · moderate] TIME, 1984, page 2
• Unknown or disputed: Entry security, Rock vehicle/check/ownership, sales quality, margins, retention, cash runway, reserve budget, board rights, and contemporaneous dissent. [researcher inference · insufficient public record] Qantel ledger row
• Alternatives and dissent: Continue pro rata, stage, recapitalize/syndicate, seek a buyer, or stop; the archive does not identify which were presented. [researcher inference · provisional] Qantel packet
• Terms, milestones, follow-on, and governance: No public term sheet or reserve ledger; any milestone package is analyst-created. [researcher inference · insufficient public record] Qantel ledger row
• Decision exercise — continue or stop: Compare follow-on need with portfolio opportunity cost. [researcher inference · moderate] Qantel packet
• Decision exercise — kill criteria: Define cash, margin, retention, channel, management, and legal/conduct failures. [researcher inference · moderate] Qantel packet
• Decision exercise — ownership / terms: Set price, seniority, board/control, reserve cap, and buyer process. [researcher inference · provisional] Qantel packet
• Decision exercise — next-check evidence: Demand cohort/sales quality, gross margin, cash conversion, litigation exposure, and financing plan. [researcher inference · moderate] Qantel packet
• Reveal — decision and outcome: TIME reports continued Rock engagement, $49m sales, and a 1980 $34m company sale; Sarofim describes investors ahead but below norm. [contemporaneous record · moderate] TIME, 1984, page 2
• Post-reveal lesson: This is an ordinary positive company outcome, not a Rock return; corporate litigation is a conduct control, not a finding against Rock. [researcher inference · strong] First Circuit, lines 57–76

Decision lab 8: Ridge Computers — underwriting a prototype and co-investor board

• Freeze — late 1981: Founders had invested $500k, worked without salary for about 18 months, and built a RISC prototype before the first VC close. Rock's actual decision is withheld until reveal. [retrospective witness · strong] Sell, printed pp. 13, 15
• Known: Working prototype, founder commitment, technical architecture/implementation available for review, and need for outside capital. [retrospective witness · strong] Sell, printed pp. 13, 15
• Unknown or disputed: Customer demand, manufacturing economics, security/check, H&Q split, board rights, follow-on dilution, reserve needs, and dissent. [researcher inference · insufficient public record] Ridge ledger row
• Alternatives and dissent: Continue founder-funding, raise from Rock alone, syndicate, seek strategic capital, or pass; later testimony does not recover a contemporaneous dissent memo. [researcher inference · moderate] Sell, printed pp. 13–15
• Terms, milestones, follow-on, and governance: Founders later say they retained majority through early rounds, but first-round price/security and board allocation are absent. [retrospective witness · moderate] Sell, printed p. 20
• Decision exercise — invest or pass: Decide whether prototype/founder commitment overcome market/manufacturing uncertainty. [researcher inference · strong] Ridge packet
• Decision exercise — kill criteria: Define architecture, customer, manufacturing, team, and syndicate failures. [researcher inference · strong] Ridge packet
• Decision exercise — ownership / terms: Set founder majority, security, Rock/H&Q allocation, board deadlock, dilution, and sale rights. [researcher inference · strong] Ridge packet
• Decision exercise — next-check evidence: Require customer tests, bill of materials, manufacturing plan, hiring, cash runway, and governance agreement. [researcher inference · strong] Ridge packet
• Reveal — decision and outcome: Rock invested, brought H&Q, and split the round; later H&Q led, VCs held majority, Rock was less active, and the board replaced management after sale disagreements. [retrospective witness · strong] Sell, printed pp. 15, 20
• Post-reveal lesson: Sell expressly says Ridge did not go bankrupt; the supported failure is governance/exit disagreement with unknown cash flows. [retrospective witness · strong] Sell, printed p. 20

Portfolio forensics — economic concentration without a cash-flow ledger

Observed exposure map

Concentration, hits, misses, and bias

• Mallaby reports that D&R's narrative result was concentrated in SDS, while HBS, Mallaby, and Kenney/von Burg disagree on vehicle inputs/outcomes. That makes “SDS first” directional, not an auditable attribution. [researcher inference · moderate] Mallaby, printed pp. 17–39 · vehicle boundary
• Removing SDS leaves Teledyne, Intel, and Apple on incompatible partnership/company/personal measures; a second-place economic ranking cannot be validated. [researcher inference · strong] Pareto stress test
• The observed set includes 16 capital/ownership cases, four intermediation/support cases, two board/acquirer controls, two passes, and two association/failure controls; 0/26 has a full chain. [researcher inference · strong] Denominator accounting
• HBR says roughly half of selected investments disappointed, and Kramlich recalls 13 unnamed companies with periods of distress; the public named corpus therefore overweights famous outcomes and cannot support a hit rate. [researcher inference · strong] HBR, printed pp. 63–67 · Kramlich, printed pp. 4–7

Failures, misses, and anti-portfolio

Apple litigation must remain split: the 1984-filed case ended in 1987 summary judgment for defendants; the 1993 UPI report describes different complaints whose docket/disposition were not located. [contemporaneous record · strong] 1987 judgment · [contemporaneous record · moderate] UPI, 1993-08-04

Uncertainty and retrieval ledger

1. Retrieve D&R formation documents, full portfolio, capital accounts, LP letters, carry/fee terms, distributions, and SDS/Teledyne position records; reconcile the conflicting reported capitalization and outcomes above without computing a new return. [researcher inference · strong] Source-map priority 3
2. Obtain Rock–Kramlich agreements and enumerate all 13 companies; map the reported vehicle economics above to actual cash flows. [researcher inference · strong] Kramlich, printed pp. 4–7
3. Retrieve Intel subscription/closing records and board minutes; determine whether the memo's expected $300k Rock debenture allocation closed. [researcher inference · strong] Memo, pp. 1–4
4. Retrieve Apple purchaser agreements, transfer/sale records, board minutes and distinct 1993 complaint dockets; do not infer realized return from 640k shares. [researcher inference · strong] Apple filing, SEC FOIA PDF pp. 33, 36–38, 57
5. Retrieve Diasonics prospectus, insider-sale table, audit/board minutes, financing agreements, settlement/disposition records, and Rock capital account. [researcher inference · strong] Diasonics court opinion, 599 F. Supp. at 449–50
6. Resolve Astrodata, Benrus, Xynetics, ELXSi, Rational, Daisy, and Qantel securities/vehicles/exits; seek primary controls for Anadex and General Capacitor before admission. [researcher inference · strong] Source-map archive exhaustion
7. Retrieve contemporaneous ROLM/Compaq proposals and pass notes; retrospective admissions alone cannot reconstruct alternatives, terms, or kill criteria. [researcher inference · strong] Pioneers note

Bottom line

Across the observed public record, Rock repeatedly appears as a financing organizer, syndicator, board participant, and recruiter around founder/operator teams. [researcher inference · strong] This is a recurring role pattern, not a comparative edge, causal efficacy estimate, or personal-return claim. The same record shows large residual error: unenumerated ordinary investments, a major monitoring failure, explicit misses, vehicle/key-person discontinuities, and no public personal-return ledger. [researcher inference · strong] Bancroft, printed pp. 25–68 · HBR, printed pp. 63–67 · Pioneers, 01:09:42–01:10:32

Coverage and evidence base

Arthur Rock's live source map contains 85 unique records: 31 ★★★, 41 ★★, and 13 ★. The 72 priority records have one exclusive primary destination: 21 TALKS, 8 WRITINGS, and 43 DEALS/controls. Later TALKS and DEALS discoveries expanded the map without changing the locked eight-family WRITINGS denominator; cross-use is allowed, but no source is double-counted in that routing arithmetic. [researcher inference · strong] Source map, mechanical coverage

The WRITINGS denominator contains seven Rock-attributed works and one ex-ante written control. Seven bodies were available in full; the Done Deals chapter is metadata-only because its legitimate public route requires controlled borrowing. All eight have exact-template notes; seven receive body-level analysis and the eighth receives an access-and-authorship analysis only. [researcher inference · strong] Source-map routing and access controls

The corpus is not eight independent Rock voices. The two UC essays are joint works; the Intel scan lacks an internal byline; Done Deals is edited and unread; and the 2025 letter is retrospective. The most defensible denominator is therefore 8 written families / 7 full bodies / 1 metadata-only body, not eight sole-authored method statements. [researcher inference · strong] Intel authorship testimony, CHM printed pp. 21–22 · Done Deals access metadata · SFGate May byline · SFGate August byline

Three lower-priority collective records complete the public attributed-writing inventory but do not enter the priority denominator: a 1999 campaign-finance letter in which Rock is one signatory, a 2003 California-recall op-ed presented as signed by five people, and a 2022 SIJ work-authorization letter whose official PDF route returned HTTP 403 here. None proves that Rock drafted the prose. [researcher inference · strong] 1999 Congressional Record, printed S13233–S13234 · 2003 Los Angeles Times signers · 2022 USCIS FOIA route, Rock listed at printed p. 14

The HBR access boundary deserves precision. The visible page shell stops after the opening, but the canonical HBR HTML delivered the complete 18,736-character essay in its official structured articleBody; the note uses phrase anchors because that representation has no page breaks. This is full, legitimate metered access—not an unofficial mirror or a purchased PDF. [contemporaneous record · strong] HBR canonical article, title, byline, opening, and structured body · HBR Store, item 87612, four-page reprint metadata

The Done Deals boundary is the opposite. Internet Archive marks the item access-restricted and its text/image derivatives private or encrypted; Open Library offers controlled borrowing, and public catalogs expose only metadata and the chapter's pp. 139–148 span. No unauthorized mirror was used. [contemporaneous record · strong] Internet Archive metadata, access-restricted-item · Open Library controlled edition · BiggerBooks table of contents

Source-by-source analysis

1955 — health reinsurance through private carriers and a public start-up fund

Representative Charles A. Wolverton inserted a March 4, 1955 letter from Arthur Rock, identified as secretary of the New York Young Republican Club, into the Congressional Record on March 21. Rock transmitted the club board's renewed support for the administration's health-reinsurance proposal. This is a signed organizational communication before Rock's documented venture career, not a personal investment memorandum. [contemporaneous record · strong] Congressional Record, 84th Congress, 1st session, Appendix A1929

The letter's mechanism is institutional rather than entrepreneurial: a $100 million public start-up fund would encourage private carriers to extend coverage to people then excluded, after which premiums were expected to support the program. The club framed this as expanding benefits through existing private organizations rather than nationalized provision. [contemporaneous record · strong] Those are the letter's policy claims; it supplies no actuarial evidence, take-up forecast, default model, administrative-cost estimate, or comparison with alternatives. [researcher inference · insufficient public record] Congressional Record Appendix A1929, entire letter

Its only safe intellectual implication is narrow: Rock was willing in 1955 to endorse temporary public risk-bearing when it was designed to mobilize private provision. It would be overreach to call that a venture-capital doctrine or the cause of later financing structures; no later writing makes that connection. [researcher inference · moderate] 1955 letter, Appendix A1929 · Bancroft oral history, printed pp. 7–8, New York Young Republicans chronology

Decision use is correspondingly bounded: distinguish a temporary risk-sharing mechanism from permanent public operation, and ask who bears losses after the start-up period. The failure mode is assuming that a premium-funded steady state will emerge merely because the launch subsidy exists. [researcher inference · moderate] 1955 letter, Appendix A1929

1968 — Intel's financing memorandum as an ex-ante written control

The surviving four-page memorandum has no signature, letterhead, internal date, or byline and refers to “Mr. Rock” in the third person. Later participant testimony is meaningful but not identical to intrinsic proof: Rock told Gordon Moore that he wrote a three-page, double-spaced Intel memo, and Moore replied that Intel's only business plan was the one Arthur wrote. The three-page recollection and four-page scan leave exact version identity unresolved. [retrospective witness · moderate] Moore–Rock panel, printed pp. 21–22 / PDF pp. 20–21, lines 487–506 · surviving memorandum, PDF pp. 1–4

The memo begins with people and technical scope. It names Robert Noyce, Gordon Moore, and Rock; summarizes the founders' Fairchild experience; proposes large-scale arrays using MOS and bipolar technologies; and permits adjacent advanced solid-state work only where prospective return justified the expenditure. It predicts no significant first-year sales, focuses that year on development, and allows that profit might begin in year two. These were underwriting assumptions, not later outcomes. [contemporaneous record · strong] Intel memorandum, PDF pp. 1–2

The capitalization and proposed financing are unusually explicit. Noyce and Moore each held 245,000 common shares and Rock 10,000 at $1; the company sought $2.5 million of ten-year, 6% convertible debentures at a $5 conversion price. Full conversion would leave founders and debenture investors with 50% each before later dilution. The instrument waived interest for three years, subordinated the debt, delayed callability, introduced a fifth-year sinking fund, restricted dividends, protected conversion economics, and granted registration rights. [contemporaneous record · strong] Intel memorandum, PDF pp. 2–3

The reasoning chain is coherent: protect early development cash, give outside capital a senior claim plus upside, align employee option pricing with the conversion price, and add a board spanning founders, investors, and operating networks. It also discloses potential Rock conflicts through Teledyne and Intersil. Yet the memo states only that Rock was expected to buy $300,000 of debentures; that is not a closing receipt, and it must remain separate from his documented 10,000 common shares. [contemporaneous record · strong] Intel memorandum, PDF pp. 1–4

The document is richer than a “people over plans” slogan. It combines founder judgment with technical scope, capital structure, cash preservation, governance, conflicts, incentives, and a milestone-linked acquisition earnout. A modern reader should not infer that Rock ignored plans; the stronger interpretation is that he distrusted false precision while still demanding an internally consistent financing-and-execution architecture. [researcher inference · strong] Intel memorandum, PDF pp. 1–4 · HBR, financial-projection and people passages

1987 — Strategy vs. Tactics from a Venture Capitalist

Rock's central argument is that execution, management, and honest day-to-day control matter more than novelty alone. His self-reported funnel is severe: about 300 plausible plans per year, one or two investments, and roughly half of selected investments failing to meet expectations. This is important denominator evidence and a direct admission of residual error; it is not an audited deal ledger or proof that people-selection caused the survivors. [investor-stated · strong] HBR, opening through “strategy is easy,” printed pp. 63–67

The diligence procedure starts with résumés and prior mistakes, then moves to long meetings in which Rock says he mostly listens. He asks whom candidates admire, what errors taught them, whom they will recruit, whether they want an active partner, and whether company-building or personal wealth dominates their motivation. [investor-stated · strong] The mechanism is repeated behavioral evidence, but the essay provides neither declined-candidate outcomes nor a false-positive/false-negative ledger. [researcher inference · insufficient public record] HBR, phrase anchors “The first place I look is the résumés,” “I don't talk much,” and “Belief in the business”

Rock makes financial control part of the people thesis. A tough accounting function and strong CFO should resist premature shipments, low-quality orders, doubtful collectibility, quota-driven revenue recognition, excess inventory, and expansion ahead of evidence. Anonymous computer-company stories illustrate how reassuring internal reports can conceal product delays, weak selling, and re-engineering costs. [investor-stated · strong] HBR, finance-function and anonymous-company passages

He also proposes a three-stage management model: the founder initially does nearly everything, then manages direct functional heads, then adds organizational layers once a company reaches roughly $100–$200 million of sales. The useful rule is that information paths and delegation must change as complexity grows; the dollar threshold is a 1987 estimate, not a timeless scaling constant. [investor-stated · moderate] HBR, phrase anchors “a company's growth has three stages” and “$100 million to $200 million”

Diasonics and Daisy are named counterexamples: both initially prospered and later required reorganization after growth outstripped management and controls. Teledyne's Henry Singleton and SDS's Max Palevsky serve as positive but stylistically different leaders. Apple is used to argue that technical expertise is usually useful while John Sculley was, by Rock's own account, a fortunate exception. These are selected retrospective illustrations, not ex-ante case reconstructions. [investor-stated · moderate] HBR, Singleton/Palevsky, Sculley, and “Consider the experience of Diasonics and Daisy” passages

The closing market critique links too much venture money to too few capable managers. Rock preferred founders to approach him; he worried that large funds created companies and pulled talent from incumbents. That is a dated statement about capital-supply and labor-market incentives, not proof that inbound sourcing dominates across eras. [investor-stated · strong] HBR, “too much money chasing too few good managers” and “vulture capitalists” passages

The essay contains its own strongest rebuttal: Rock admits both substantial failure among selected investments and the role of luck, then ends with rhetoric suggesting great people cannot lose. The evidence supports a probabilistic people-and-control screen, not managerial determinism. [researcher inference · strong] HBR, opening funnel, luck passage, and final paragraph

2000 — Done Deals chapter: a priority source with no readable body

Public records establish a Gupta-edited first-person volume, an “Arthur Rock & Co.” entry beginning at printed p. 139, and the next chapter beginning at p. 149. [contemporaneous record · strong] They do not expose the Rock chapter's prose or reveal whether it is a transcript, edited interview, subject-written chapter, or blended narrative. [researcher inference · insufficient public record] Internet Archive controlled-book route · BiggerBooks table of contents

Accordingly, this analysis extracts no argument, example, decision rule, deal claim, quotation, or return figure from the chapter. Its importance is prospective: a 2000 first-person account could test what Rock chose to emphasize between the HBR essay and later oral histories, but that hypothesis cannot substitute for the pages. [researcher inference · strong] Open Library edition · Google Books bibliographic description

The exact legitimate next step is controlled borrowing, a library copy, or purchase, followed by page-by-page separation of Gupta framing from Rock-attributed language. Secondary excerpts and unauthorized mirrors cannot authenticate omissions, transitions, or editorial intervention. [researcher inference · strong] Internet Archive metadata and restricted derivatives

2000 — letter to Arthur Levitt on auditor independence

Rock wrote one day after SEC chairman Arthur Levitt presented an auditor-independence proposal. He advocated separating auditing from consulting, with a tax-services exception, and offered an issuer-level fallback: prohibit companies from buying consulting from their own auditors. This is contemporaneous participation in a live policy debate. [investor-stated · strong] Rock letter, PDF p. 1, both body paragraphs · SEC, Levitt remarks, June 27, 2000

His stated mechanism is conflict prevention. Rock reports that while chairing audit committees at Intel, Apple, and Teledyne he insisted that the outside auditor not consult for the issuer other than on tax matters. That is strong evidence of his claimed governance practice, but not independent proof of titles, dates, implementation, or outcomes. [investor-stated · strong] Rock letter, PDF p. 1, first body paragraph

The letter's strength is clarity; its limitation is analytical thinness. It provides no named failure, empirical comparison, fee threshold, preapproval process, monitoring system, switching-cost analysis, or remedy. The final SEC rule restricted specified non-audit services rather than adopting a total ban, so the historical outcome cannot be presented as simple adoption of Rock's position or evidence that he caused it. [contemporaneous record · strong] SEC final rule, summary and discussion of alternatives

May 2006 — UC accountability without disabling talent competition

Gordon Moore, Rock, and William K. Coblentz jointly argued that UC compensation and disclosure problems required correction but should not trigger remedies that impaired the university's ability to recruit and retain strong faculty and executives. [investor-stated · moderate] The joint byline supports a shared public position; it does not reveal which author drafted each sentence. [researcher inference · insufficient public record] SFGate, title, byline, opening, and paragraphs beginning “There is an obvious need”

The reasoning balances two objectives. Public accountability and better administration are accepted; competitive compensation and leadership discretion are treated as necessary inputs to academic quality. The authors point to UC's research standing, alumni, faculty, and economic contribution, then predict that strict bureaucratic rules could damage recruitment. Several numerical and causal claims are unlinked in the op-ed, so they remain coauthors' evidence, not independently validated base rates. [investor-stated · moderate] SFGate, paragraphs beginning “We agree that UC,” “We need to remember,” and “To remain at the top”

The operational rule is not “ignore governance failures.” It is to fix a demonstrated control defect at the narrowest level that restores accountability without destroying a scarce-capability acquisition system. The missing evidence is whether the proposed reforms actually improved disclosure or whether looser discretion perpetuated the problem. [researcher inference · moderate] SFGate, “11-point plan” and closing paragraphs

August 2006 — disclosure at decision, privacy during negotiation

Warren Hellman, Rock, and Moore opposed AB775's requirement that UC conduct senior-management hiring and compensation discussions in public. They accepted public disclosure and open final decisions while arguing that interviews, negotiation, and sensitive candidate information needed protected deliberation. The position is jointly attributed; sentence-level drafting remains unknown. [investor-stated · moderate] SFGate, opening through paragraphs beginning “We recognize” and “Still, Yee's legislation”

Their causal mechanism is a candidate-market effect: public negotiations could deter privacy-sensitive people, reveal UC's offers to competitors, narrow the pool, and weaken internal promotion. Executive recruiter Eunice Azzani supplies a named contemporaneous practitioner view, but the essay gives no comparison group, hiring series, or observed effect under the proposed rule. [contemporaneous record · moderate] SFGate, paragraphs beginning “As graduates,” “We agree with Eunice Azzani,” and “Not only would AB775”

The strongest counterargument appeared alongside it: Assemblyman Leland Yee cited audit findings and argued that public deliberation was needed because compensation exceptions and approvals had occurred outside expected processes. [contemporaneous record · strong] The dispute is therefore not transparency versus secrecy in the abstract; it is where in the hiring process confidentiality should end and enforceable accountability begin. [researcher inference · strong] Yee, companion SFGate op-ed, paragraphs on audit exceptions and open meetings

2025 — leaders, listening, execution, and self-knowledge

Rock's five-paragraph letter compresses the people thesis into a procedure: choose good people, listen carefully, spend enough time to test first impressions, prefer people who know both their capabilities and limits, and value execution, drive, intelligence, integrity, and intellectual honesty. [investor-stated · strong] Steve Jobs Archive, paragraphs 1–5

The longitudinal element matters. A polished first meeting is treated as weak evidence; repeated observation tests whether conduct and self-description converge. [investor-stated · strong] That is a stronger rule than charisma-based founder selection, but the short letter supplies no cases, reference process, thresholds, negative examples, or denominator. [researcher inference · insufficient public record] Steve Jobs Archive, paragraphs 2–5

The vocabulary is remarkably stable against 1987: execution over ideas, careful listening, integrity, and reality-based self-assessment recur after 38 years. Stability of articulation is not proof of stable application or selection skill; the letter is a retrospective self-assessment written decades after the signature deals. [researcher inference · strong] 2025 letter, paragraphs 1–5 · HBR, diligence, execution, and intellectual-honesty passages

Cross-source synthesis

What is foundational, occasional, and absent

The strongest supported written thesis is therefore narrower than “bet on people.” Rock's accessible corpus says to select for execution and reality-testing, build independent channels for negative information, and change the management system as complexity grows; the Intel control adds stage-specific financing and governance. It does not show that these rules generated superior returns, that Rock applied them uniformly, or that people quality dominated market timing, access, terms, team contribution, and luck. [researcher inference · strong] HBR, funnel, control, management-stage, and luck passages · Intel memo, PDF pp. 1–4

The 1955 and 2006 civic writings add an institutional theme but should not be laundered into investment behavior. They balance private capability with public or board accountability: private carriers with a public start-up fund; discretion in talent recruitment with disclosure and final accountability. That resemblance is analytically interesting, yet no source says it shaped Rock's venture method. [researcher inference · moderate] 1955 letter, Appendix A1929 · May 2006 UC essay · August 2006 UC essay

Intellectual chronology, evolution, and contradictions

Four tensions should remain unresolved rather than averaged away:

1. People over projections versus a detailed plan. HBR attacks false five-year precision, while the Intel memo forecasts the development period and specifies financing, capitalization, governance, incentives, and risk protections. The most coherent reconciliation is planning by decision horizon rather than planlessness, but that is an analyst reconstruction. [researcher inference · moderate] HBR, projections passage · Intel memo, PDF pp. 1–4
2. Execution primacy versus residual error and luck. The public doctrine elevates people, yet HBR says roughly half of selected investments disappointed and many success stories owed something to luck. The record establishes neither necessity nor sufficiency for people quality; it does establish that Rock's stated screen left substantial residual error. [researcher inference · strong] HBR, opening funnel and luck passage
3. Active partnership versus protected independence. HBR favors founders who want an investor involved in decisions; the Levitt letter insists that an auditor remain free of consulting conflicts. These are compatible only if the role boundary is explicit: advice and governance are not independent assurance. [researcher inference · strong] HBR, active-investor passage · Levitt letter, PDF p. 1
4. Transparency versus confidential deliberation. The UC essays support disclosure and final public decisions while resisting public negotiation. The companion Yee essay argues that secrecy enabled control failures. The unresolved design question is which decisions, facts, and conflicts require disclosure at which stage. [researcher inference · strong] Rock/Hellman/Moore essay · Yee counterposition

Decision playbook and modern translation

Historical operating rules

Modern VC translation (as of 2026-08-19)

These translations are analyst-created. None should be back-attributed to Rock, and none establishes historical investment efficacy. Their purpose is to convert bounded written evidence into testable present-day questions while preserving changed technology, instrument, regulation, and institutional context. [researcher inference · strong] HBR, failure and luck boundaries · NIST AI RMF Playbook, voluntary and adaptable scope · NVCA documents, tailoring disclaimer

Best writings and unresolved record

Read these five first

1. HBR, 1987 — Strategy vs. Tactics from a Venture Capitalist. Best general method statement: sourcing preference, diligence questions, finance controls, scaling model, named failures, a selection funnel, and luck in one short primary essay. Its own denominator caveat makes it more useful than a maxim collection. [researcher inference · strong] HBR, complete official article body
2. Intel financing memorandum, 1968. Best ex-ante control: people, technology, capitalization, instrument, cash preservation, governance, incentives, conflicts, and planned investors before outcome. Read with the authorship/version caveat. [researcher inference · strong] Intel memo, PDF pp. 1–4 · Moore–Rock authorship discussion, printed pp. 21–22
3. Levitt letter, 2000. Clearest written governance rule and a rare contemporaneous intervention in public policy; its brevity also exposes what is missing from a slogan-like control principle. [researcher inference · strong] Rock letter, PDF p. 1
4. Leaders letter, 2025. Best compressed restatement of the people thesis and a useful test of rhetorical continuity against HBR. Read it as retrospective doctrine, not a skill audit. [researcher inference · strong] Steve Jobs Archive, paragraphs 1–5
5. UC transparency and talent essay, August 2006. Best explicit tradeoff between private deliberation and public accountability, with a same-day published counterargument. It is joint voice, not Rock-alone doctrine. [researcher inference · moderate] Rock/Hellman/Moore essay · Yee counterposition

Evidence that could change the synthesis

• Retrieve Done Deals. Borrow or inspect a lawful copy at printed pp. 139–148; record the interview date, editorial transitions, exact Rock-attributed language, and whether it changes the 1987 method. [researcher inference · strong] Open Library controlled edition
• Identify the Intel document version. Locate a dated original, transmittal, distribution list, or Rock file copy that links the surviving four-page scan to the three-page/two-and-a-half-page recollections; retrieve signed subscriptions and closing evidence. [researcher inference · strong] Moore–Rock panel, printed pp. 21–22 · Intel memo
• Inspect the 1996/1998 Levitt correspondence. Columbia's finding aid lists two incoming and two outgoing Rock/Intel/Arthur Rock & Co. items, but no body or subject is public. [researcher inference · strong] Arthur Levitt papers finding aid, printed p. 167
• Test claimed practice against records. Audit-committee charters, minutes, auditor engagements, non-audit fees, CFO hiring, shipment controls, and board decisions could confirm or contradict the HBR and Levitt claims. [researcher inference · strong] Intel 2003 proxy, “Directors Emeriti” · Levitt letter, PDF p. 1
• Recover a decision denominator. The HBR funnel lacks company names, rejected plans, ordinary outcomes, loss definitions, and capital-weighted results. Without those, the people thesis remains a plausible method statement rather than a validated edge. [researcher inference · strong] HBR, opening funnel
• Resolve joint-author drafting and outcomes. Editorial correspondence for the 2006 UC essays and subsequent hiring, compensation, and transparency evidence could show which mechanisms the authors owned and whether their predictions held. [researcher inference · moderate] May essay · August essay

No retrievable public priority writing remains unprocessed. The outstanding items require controlled borrowing, archive access, or nonpublic company records; these are specific evidence limits, not a claim that the missing materials do not exist. [researcher inference · strong] Source map, direct-archive exhaustion and highest-value retrievals

Coverage and evidence base

The TALKS corpus contains 21 deduplicated priority families, each represented by one source note: 19 full public source bodies and 2 controlled institutional barriers. Fourteen families contain Rock speech or a cataloged Rock appearance; seven are participant or colleague controls. Of the 19 public bodies, 18 expose complete human-edited text layers. The CHM Pioneers of Venture Capital record exposes the full hosted video plus machine-generated timed text, which was used only after speaker turns were controlled against moderator prompts and surrounding context. Contemporaneous record · strong for inventory and access; researcher inference · strong for family boundaries. Source map

No raw transcript, video, audio, WARC, or machine-text dump is stored in the repository. Stanford Law was read through a disclosed public text renderer because the non-paywalled official page returned automated HTTP 403. Silicon Genesis was recovered from a fixed Common Crawl record because the original Stanford URL is dead. The two Something Ventured sessions remain one family and preserve their repository-date conflict. No local ASR was used; any genuinely missing-media transcription must use the OpenAI API. Researcher inference · strong. Stanford Law note · Silicon Genesis note · Something Ventured note · Pioneers note

Source-by-source guide

Arthur Rock — Bancroft Library oral history (2008–2009)

The most complete direct account separates Rock's nontechnical people screen, Fairchild financing, the Davis & Rock partnership, Intel and Apple governance, and Diasonics as a major failure. Rock says intellectual honesty is tested over repeated contact, describes Intel as an exception underwritten through known operators, and admits that board status did not substitute for attention at Diasonics. Investor-stated · strong for Rock's account; researcher inference · insufficient public record for portfolio-wide application, cash flows, or efficacy. Bancroft transcript, printed pp. 17–18, 25–68 · note

Arthur Rock — HBS Entrepreneurs oral history (2001)

Rock presents a concentrated, involvement-heavy practice: spend time with prospective leaders, look for internal consistency and intellectual honesty, and limit company count so governance remains possible. He also distinguishes Teledyne's operating judgment from his sounding-board role and says Apple involvement was smaller than at Intel, SDS, or Teledyne. The HBS introduction's reported D&R economics are editorial framing without a reproduced ledger. Investor-stated · strong for stated method and role limits; researcher inference · insufficient public record for returns or selection efficacy. HBS transcript, printed pp. 4–16 · note

Arthur Rock — Science History Institute oral history 0265 (2002)

The archive catalogs a 29-page transcript and three audio files totaling 01:45:00, but access requires a patron name/email because of an interviewee-stipulated privacy condition. No body, quote, or substantive Rock claim is used. Contemporaneous record · strong for catalog metadata; researcher inference · insufficient public record for content. SHI record and request route · route note

An Evening with Arthur Rock — CHM/John Markoff (2007)

Rock frames his edge as evaluating drive, intelligence, and intellectual honesty rather than forecasting technology. He differentiates no board role at Fairchild from prolonged Intel and Teledyne governance, assigns Intel's memory decision to Moore and Grove, and says Markkula and observable customer demand changed his Apple judgment. Investor-stated · strong for those distinctions; researcher inference · provisional for causal value-add. CHM transcript, printed pp. 9–15 · note

Gordon Moore and Arthur Rock — CHM panel (2014)

This paired record is the strongest live attribution control. Rock describes recruiting, accounting, and board mechanisms; Moore supplies technical and operating ownership. Rock says he wrote the short Intel memo, while Moore confirms that Intel's only business plan was Rock's; their page-count memories still differ from the surviving four-page scan. Moore assigns the memory exit to product economics and management, not Rock. Investor-stated · strong for Rock's statements; retrospective witness · strong for Moore's corrections; researcher inference · insufficient public record for exact scanned-version identity. CHM panel, printed pp. 9–12, 19–31 · note

Arthur Rock — Something Ventured raw interviews (2008–2009 date conflict)

Across two sessions, Rock turns abstract value-add into mechanisms: repeated listening, board service, recruiting, staff-meeting attendance, conservative accounting, audit structure, public-offering preparation, CEO oversight, and willingness to stop financing. He limits his own credit for Intel's memory exit to approval and makes Markkula the key Apple underwriting subject. Smithsonian dates the family 2008-09-04; Stanford calls part 1 approximately 2009 and part 2 2009-11-18, while part 2 carries an internal 2009 header. Investor-stated · strong for the claims; contemporaneous record · strong for repository metadata; researcher inference · strong for keeping the dates unresolved. Part 1, printed pp. 8–15 and 29–34 · part 2, printed pp. 10–29 · note

Jay Last — CHM oral history (2007)

Last says the scientists first sought employment together and that Rock and Bud Coyle proposed forming a company. His roughly 20–30 rejected prospects counter Rock's later 35 and 45 variants. This is founder testimony that constrains financier-only mythology without resolving the exact list or sponsor terms. Retrospective witness · strong. Last transcript, printed pp. 3–5 · note

Mike Markkula — CHM oral history (2012)

Markkula separates Apple's sequence: Valentine referred him; Markkula committed before Rock; Venrock/Peter Crisp and Rock later joined; Rock then contributed as a director. His account also says Rock had reportedly passed earlier, which remains in tension with Rock's simpler Markkula-called-me narrative. Retrospective witness · strong for Markkula's own actions; retrospective witness · moderate for what Rock previously decided. Markkula transcript, printed pp. 21–29 and 43 · note

Herman Fialkov — Science History Institute oral history (2009–2010)

Fialkov supplies a limited-partner-side account of D&R solicitation, reports, term, and a stock distribution. It is useful evidence that Rock raised from prior relationships and that LP experience differed from a founder-only narrative, but it is not a partnership statement or audited cash-flow schedule. Retrospective witness · moderate. Fialkov oral history, printed pp. 88–91 and 103–106 · note

Arthur Rock and Eugene Kleiner — Forbes Founding Funders (2000)

Rock calls Fairchild the galvanizing event, recalls a maximum D&R check pattern rather than a formal concentration rule, and argues that large board loads reduce company service. Kleiner separately explains builder motivation and the industry's institutionalization. Agreement between two participants in one edited Q&A is not independent corroboration. Investor-stated · strong for each speaker's positions; researcher inference · moderate for synthesized doctrine. Forbes Q&A, exchanges beginning “galvanizing event,” “typical deal,” and “How many boards” · note

Silicon Valley — PBS American Experience (2013)

Five edited Rock clips sit beside Moore, Last, historians, and narration. Rock supplies his inbound-letter, sponsor-search, Sherman Fairchild, and Sputnik recollections; Moore and Last keep the company proposal, team intent, and sponsor decision multi-actor. The broadcast preserves a 45-company count and a $1.3 million narration, both in tension with other sources. Investor-stated · strong for Rock's aired words; retrospective witness · strong for Moore/Last; researcher inference · provisional for exact count and financing amount. PBS official program transcript, sequences beginning “letter landed,” “working the phones,” and “Sherman Fairchild” · note

Fairchild Semiconductor 50th Anniversary — Stanford V0311 (2007)

Stanford catalogs two DVDs of the October 4 panel with Julius Blank, Jay Last, Gordon Moore, Arthur Rock, moderator Leslie Berlin, and introducer John Hennessy. It is on-site and reproduction-controlled; no public body was inspected. The route is distinct from CHM's October 5 Legacy panel, where Rock is not a speaker. Contemporaneous record · strong for metadata; researcher inference · insufficient public record for content. Stanford finding aid · route note

Gordon Moore — Fortune interview (2012)

Moore credits Kleiner's inquiry and Rock/Coyle's startup proposal, recalls roughly 30 rejected prospects, and separates founder-price Intel purchases from Rock's smaller equity purchase and the debenture syndicate. He also owns the technical and memory-exit reasoning. Retrospective witness · strong for Moore's account; researcher inference · insufficient public record for exact cash flows. Fortune, “Mutiny and serendipity” and “Inside Intel” · note

The Legacy of Fairchild — CHM panel (2007)

Moore, Corrigan, Sanders, and Kvamme discuss Fairchild's founding and spinout legacy; Rock is not a speaker. Moore again says the group sought employment and Rock plus a Hayden Stone colleague proposed a company. The source is a witness control, not another Rock interview or independent source family for Moore's repeated memory. Retrospective witness · strong for Moore's account; researcher inference · strong for the speaker boundary. Legacy panel, printed pp. 2–4 · note

Reid Dennis — Something Ventured raw transcript, part 2 (2008)

Dennis praises Rock's long Intel committee involvement and company-building horizon. This supports durable governance effort but does not identify each committee action, establish causality, or turn Dennis into an independent observer of every Rock deal. Retrospective witness · moderate. Dennis transcript, printed pp. 29–31 · note

Dick Kramlich — CHM oral history, part 1 (2015)

Kramlich distinguishes the post-D&R Rock–Kramlich period, fundraiser roles, Apple access, and the later split. His “one-third partner” phrase does not by itself establish carry, voting, profit-sharing, or ownership rights. The account therefore supports team and relationship attribution while leaving legal/economic terms unresolved. Retrospective witness · strong for Kramlich's account; researcher inference · insufficient public record for rights and economics. Kramlich transcript, printed pp. 4–7 · note

Arthur Rock — Stanford Law Legal Matters (2022–2023)

Rock discusses Intel's legal design, broad options and vesting, governance, Apple/Jobs, luck, and intellectual honesty. Grundfest's prompts and Driscoll's editing are framing, not Rock evidence; the late recollection cannot prove historical implementation. Investor-stated · strong for Rock's positions; researcher inference · provisional for generalization. Stanford Law Q&A, questions on Intel, Apple, luck, and education · public renderer and note

Arthur Rock — Stanford Silicon Genesis (2002)

Rock says Intel capital was committed before his short plan circulated, defines his Intel contribution as financing/incorporation/governance rather than science, explains why SDS was sold and why Xerox's later strategy failed, and ties Apple to Markkula plus visible demand. He also criticizes quick-exit arithmetic. The original Stanford page is dead; claims are tied to a fixed Common Crawl record and unique paragraph anchors. Investor-stated · strong for Rock's recovered account; researcher inference · moderate for cross-deal synthesis. Common Crawl index and WARC coordinates · note

Scientific Data Systems alumni panel — CHM (2007)

Rock and SDS participants jointly revisit Max Palevsky, financing, capitalization, board roles, competition, and growth funding. The multi-speaker record distinguishes Rock's financing/governance memory from operators' product and market experience; speaker agreement inside one panel is still one event family. Investor-stated · strong for Rock; retrospective witness · strong for SDS participants; researcher inference · moderate for integrated role allocation. SDS panel, printed pp. 8–14, 23–24, 30, and 35–36 · note

Triumph of the Nerds — PBS (1996)

The earliest Rock moving-image testimony in the mapped public corpus is a short edited clip about meeting Jobs: appearance and diet violated the prevailing venture norm, while Jobs was articulate. It is a contemporaneous-to-1996 broadcast record but retrospective about Apple; the underlying full interview was not found on the inspected PBS route. Investor-stated · moderate. PBS Part I transcript, Jobs/Rock sequence · note

Pioneers of Venture Capital — CHM (2002)

In moderator-controlled turns, Rock recounts Fairchild financing and ownership structure, distinguishes SDS/Teledyne from Intel as his first large outcomes, explains the SDS sale and Xerox post-acquisition change, describes the Intel raise and founder commitment, and admits passing on ROLM and Compaq. The miss is especially diagnostic: he rejected the original ROLM plan for weak experience/business sense, then later hired Ken Oshman at Echelon. Investor-stated · moderate because the host-generated text required speaker/context control and contains recognition errors. CHM catalog, Rock turns 00:22:31–00:26:50, 00:42:17–00:44:49, 00:57:37–01:00:28, and 01:09:42–01:10:32 · note

The catalog states a 01:41:06 media extent, while the first machine-text segment starts at 00:00:00 and the last starts at 01:54:01; the latter is not a verified runtime. That discrepancy and the transcript's unlabeled speakers prevent treating raw machine output as self-authenticating quotation. Contemporaneous record · strong for catalog/asset metadata; researcher inference · strong for the limitation. CHM catalog

Recurring frameworks and strongest counterarguments

People-first selection is a screen, not an outcome guarantee

Rock repeatedly favors prolonged observation, listening, internal consistency, drive, and intellectual honesty over technical or forecast precision. The record also supplies falsifiers: Diasonics escaped adequate monitoring, ROLM was rejected for thin experience, Compaq was missed, and Rock's written funnel says roughly half of selected investments disappointed. Investor-stated · strong for the screen; researcher inference · strong that residual error remains large; researcher inference · insufficient public record for incremental predictive power. Bancroft, printed pp. 63–68 · Pioneers, 01:09:42–01:10:32 · written-corpus HBR analysis

Trusted access and demonstrated execution are distinct from selection skill

Fairchild arrived through Kleiner and Hayden Stone; Intel through Noyce and Moore's record; Apple through Markkula, with Valentine and Venrock also in the sequence; D&R capital came largely from people who knew prior work. These channels create valuable information and scarce access, but they do not prove that Rock selected better than comparable investors from the same network. Researcher inference · strong. Last, printed pp. 3–5 · Moore–Rock, printed pp. 19–22 · Markkula, printed pp. 21–29

Board help is mechanism-specific and bounded

The spoken record supports recruiting, CEO succession, accounting discipline, audit structure, staff-meeting participation, offering preparation, and long committee work. It also records no Fairchild board seat, management-owned Intel product decisions, limited Apple involvement relative to other companies, and failed monitoring at Diasonics. A board title is therefore neither operating control nor causal proof. Researcher inference · strong. Something Ventured part 2, printed pp. 10–13 · Moore–Rock, printed pp. 21–31 · Bancroft, printed pp. 63–65

Vehicle, capital, and team boundaries change the meaning of concentration

Hayden Stone placement work, Davis & Rock partnership capital, the Rock–Kramlich period, Arthur Rock & Associates/Co., personal Apple shares, syndicate capital, and portfolio-company board service are separate. Rock's preference for few companies may reflect a service model, a small vehicle, era-specific capital supply, personal choice, or all four. Kramlich and Fialkov make partner/LP roles visible without supplying governing agreements. Researcher inference · strong for separation; researcher inference · insufficient public record for exact decision rights and economics. Fialkov, printed pp. 88–91 · Kramlich, printed pp. 4–7 · source-map boundary

Company-building and liquidity are separate decisions

Rock says the SDS sale premium made rejection difficult, while Xerox's later strategy damaged the business; D&R's term forced stock distribution; Intel and Teledyne involved decades of governance. These examples reject both “never sell” and “exit quickly” as universal rules. Investor-stated · strong for the episodes; researcher inference · provisional for a general exit doctrine because complete exit, reserve, and opportunity-cost records are absent. Silicon Genesis, paragraph beginning “Xerox came along” · Something Ventured part 2, printed pp. 24–25

Repetition is not independence, truth, or efficacy

Fairchild, Intel, and Apple stories recur across many Rock interviews, often from the same speaker and later institutional retellings. Moore, Last, Markkula, Fialkov, Dennis, Kramlich, and SDS participants add distinct perspectives, but shared event memory and editorial reuse remain correlated. Researcher inference · strong. source map · chronology conflicts

Intellectual evolution and unresolved contradictions

Researcher inference · strong: the stated philosophy is unusually stable—people, listening, company building, long governance, and few commitments—but the evidence becomes more bounded over time as misses, team roles, vehicle constraints, and luck enter the record. Stability of rhetoric is not evidence of stable execution or excess returns. Bancroft, printed pp. 63–68 · Pioneers, 01:05:01–01:10:32 · Stanford Law Q&A

Biography, deals, vehicles, and attribution leads

Five sources to study first

1. Bancroft oral history — Researcher inference · strong. Deepest direct chronology, method, vehicle, board, and failure record.
2. Something Ventured Rock interviews — Researcher inference · strong. Most detailed operating mechanisms, stop-financing logic, and role boundaries across two raw sessions.
3. Moore–Rock CHM panel — Researcher inference · strong. Strongest paired participant control for Intel financing, memo authorship, product decisions, and governance.
4. Pioneers of Venture Capital — Researcher inference · moderate. Best same-stage comparison with peer investors and rare direct admissions about misses.
5. Markkula oral history — Researcher inference · strong. Strongest Apple sequence and attribution counterweight to Rock's own compressed story.

Researcher inference · strong. Last is the next Fairchild control; the SDS alumni panel is the next company-specific operating control; Kramlich is the next vehicle/team control.

Evidence still missing

• The SHI transcript/audio requires its stipulated patron request; Stanford V0311 requires on-site inspection or authorized reproduction. SHI route note · V0311 route note
• Pioneers needs independent human correction of machine wording and complete speaker segmentation; the host-generated text is useful for locators, not a self-authenticating transcript. Pioneers note
• Fairchild contract/correspondence, D&R governing and LP records, Rock–Kramlich agreements, Intel subscription/board records, Apple purchaser/board records, Diasonics board/accounting files, and pass memos remain necessary to adjudicate oral-history claims. Source-map retrieval priorities
• No complete Rock opportunity, investment, ownership, follow-on, exit/write-off, or cash-flow denominator surfaced. Spoken recall cannot support personal IRR, a literal Pareto contribution, or an ex-outlier return test. source-map denominator
• The public broadcast and Q&A routes do not expose the complete underlying Rock interviews for PBS Triumph or Silicon Valley; absence from inspected public routes is not proof the footage no longer exists. Triumph note · Silicon Valley note

Researcher inference · strong: the spoken record supports a bounded system—relationship-based sourcing, prolonged people assessment, concentrated ownership attention, specific governance mechanisms, and willingness to stop or sell—alongside access advantages, misses, monitoring failure, vehicle constraints, market luck, and team/company agency. It does not prove Rock-exclusive credit, uniform application, a complete personal investment system, or audited outperformance. Bancroft, printed pp. 17–18 and 63–68 · Moore–Rock, printed pp. 19–31 · Pioneers, 01:05:01–01:10:32 · coverage denominator

Test people longitudinally for intellectual honesty, self-knowledge, and builder motive

• Rule — Do not treat a polished first meeting, a novel idea, or a precise forecast as the decision. Spend enough time listening to test whether the founder's account stays consistent, whether bad news changes the account, whether the founder knows the limits of their competence, and whether the objective is building a company rather than engineering a quick exit. [investor-stated · strong] HBR, diligence and motivation passages, printed pp. 63–67
• Evidence status — [investor-stated · strong] as a selection rule repeated from the 1987 HBR essay through Rock's 2025 leaders letter; [documented behavior · moderate] in named cases because most application accounts are retrospective and the rejected-plan denominator is unavailable. HBR, printed pp. 63–67 Rock leaders letter, paragraphs 1–5
• Provenance and origin — The earliest accessible general articulation is Rock's 1987 essay: “Good execution and good management—in a word, good people—are rare.” HBR, opening after the funnel, printed pp. 63–67 His 2008–2009 raw interview gives the operational maxim: “I think you learn a lot more by listening than by asking questions.” [investor-stated · strong] Something Ventured part 1, printed p. 9, block 09
• Mechanism and reasoning — Repeated observation creates opportunities for rehearsed claims to collide with references, operating facts, mistakes, and changed evidence. Rock's intended signal is not charm but reality contact: the ability to acknowledge error, recruit around personal limits, and change direction without surrendering intensity. [investor-stated · strong] HBS oral history, “Investment Philosophy,” printed pp. 15–16 Rock leaders letter, paragraphs 2–5
• Historical and vehicle context — Rock says sparse competition let him spend much longer with candidates than later investors could, and his small practice did not need to deploy a large institutional pool on a fixed cadence. Those conditions aided the method; they are not proof of its accuracy. [investor-stated · moderate] CHM Markoff interview, printed p. 15 HBS oral history, printed pp. 15–16
• When to use — Use before founder selection, key-hire approval, and a major follow-on when the team's judgment is central and observations can be repeated. Do not use it as permission to ignore product evidence, market structure, price, security, or conflicts; Rock's own Intel memo addressed all of those. [researcher inference · strong] Intel memorandum, PDF pp. 1–4
• How to apply it — (1) Compare the résumé with references and actual shipped work. (2) Ask for a consequential mistake, its detection path, and the resulting change. (3) Revisit the same market, product, and hiring claims after new evidence. (4) Ask whom the founder admires and which complementary executives are needed. (5) Kill or pause when contradictions are material, bad news is rationalized, references fail, or the founder cannot name a competence boundary. (6) The next check is unlocked by consistent reference evidence plus demonstrated correction of one real operating miss. [researcher inference · strong] HBR, résumé, listening, mistakes, and recruiting passages
• Examples —
  • Intel — documented application, retrospective: Rock says his unusual confidence came from Noyce and Moore's demonstrated Fairchild execution rather than forecast precision; Intel became a major company, but later product decisions belonged to management. [investor-stated · strong] HBS oral history, printed pp. 9–11 Moore–Rock panel, printed pp. 21–31
  • Apple — documented application, retrospective and bounded: Rock says he underwrote Markkula, not the two founders he initially disliked; visible show demand changed the case. Apple succeeded, but Markkula's account shows that he performed the technical evaluation and company-building work that preceded Rock's participation. [investor-stated · strong] Something Ventured part 1, printed pp. 32–33 Markkula oral history, printed pp. 21–23
  • ROLM — documented failure/exception: Rock says he passed because Ken Oshman lacked experience and business sense; ROLM later succeeded, and Rock later hired Oshman at Echelon. The screen produced a false negative and later evidence changed his person judgment. [investor-stated · moderate] Pioneers panel, 01:09:42–01:10:32
  • Diasonics — researcher-applied counterexample: Rock later faulted his own monitoring while aggressive sales booking and operating problems developed. Selecting people did not remove the need to verify information after investment. [researcher inference · strong] Bancroft oral history, printed pp. 63–65
• Evidence — The direct doctrine is preserved in the HBR note, 2025 letter note, and Something Ventured note; the case boundary and anti-portfolio are in the deal ledger. Independent underlying works include Rock's 1987 essay, two institutional oral histories, the late letter, and Markkula's participant testimony. [researcher inference · strong]
• Limits, failure modes, and exception ledger — The HBR funnel says roughly half of selected investments disappointed, so the rule is neither sufficient nor validated against rejected founders. ROLM and Compaq expose false-negative risk; Apple exposes first-impression error; Markkula's testimony disputes any claim that Rock performed all relevant evaluation. Attempted routes that remain missing are the HBR company's named 300-plan annual denominator, rejected-plan files, reference notes, investment-committee records, and the controlled Done Deals chapter at printed pp. 139–148. [researcher inference · strong] HBR, opening funnel Done Deals controlled route
• Evolution and contradictions — The vocabulary is stable from 1987 to 2025, but the later letter compresses away finance controls, company stages, and admitted luck. Rock's 2000 view that some Internet prices were “ridiculous” while he would not forecast a crash demonstrates useful epistemic restraint, yet it does not prove that the people screen protected him in that period. [researcher inference · strong] Forbes, closing market exchange Rock leaders letter, paragraphs 1–5
• Modern VC translation (as of 2026-08-19) — What transfers is longitudinal evidence and explicit knowledge-boundary testing; what fails is the assumption that a modern competitive process permits months of access. Observable signals are consistent references, changed views after disconfirming data, reproducible product claims, and complementary recruiting. Likely misuse is “back the person” as an excuse to waive market, safety, price, or governance work. Analyst-created hypothetical: require an AI founder to reproduce a critical evaluation live, explain a failed result, and identify the expert who must own the next risk gate; NIST's current playbook organizes evidence around Govern, Map, Measure, and Manage rather than pitch quality. [researcher inference · moderate] NIST AI RMF Playbook, functions and tailoring guidance, updated 2026-06-10

Let earned reputation and trusted intermediaries compress uncertainty—never erase it

• Rule — Give demonstrated operating history and trusted relationship referrals more weight than an untested plan, but identify exactly whose reputation supports which claim and preserve independent diligence on technology, market, terms, and conflicts. [researcher inference · moderate] Bancroft oral history, printed pp. 45–48 Intel memorandum, PDF pp. 1–4
• Evidence status — [investor-stated · strong] for Intel's reputation-led syndication; [documented behavior · strong] across Fairchild, SDS, Intel, and Apple sourcing routes; [researcher inference · moderate] as a general rule because Rock never publishes a reference-weighting formula. Bancroft oral history, printed pp. 45–48 Stanford Silicon Genesis, paragraph beginning “What did I do?”
• Provenance and origin — The earliest observed use is 1957: Eugene Kleiner's family connection reached Rock, who recognized the intact Shockley group as a financing opportunity and searched for a sponsor. This is behavior reconstructed from participant testimony, not a contemporaneous Rock rule. [retrospective witness · strong] Jay Last oral history, printed pp. 3–4 Moore interview, “Mutiny and serendipity”
• Mechanism and reasoning — A trusted intermediary reduces identity and character uncertainty; a demonstrated team record supplies evidence of joint execution; a financing reputation can coordinate otherwise fragmented capital. The chain breaks if reputation is borrowed across domains, if the referee lacks direct knowledge, or if familiarity suppresses adverse evidence. [researcher inference · strong] Moore–Rock panel, printed pp. 20–22
• Historical and vehicle context — Early venture syndication depended on wealthy individuals and personal networks because institutions were not available to Davis & Rock. Rock's ability to call intended Intel investors in roughly a day and a half was an access advantage, not simply selection skill. [investor-stated · strong] CHM Markoff interview, printed pp. 7–8 Moore–Rock panel, printed pp. 5–13, 20–22
• When to use — Use when the referee observed the team in a comparable setting or when founders have a directly relevant shared operating record. Do not use for a new market, a changed team, a conflicted referee, or a claim that can be tested directly. [researcher inference · strong] Moore–Rock panel, printed pp. 20–22
• How to apply it — (1) Draw the referral chain. (2) Record each person's direct observation and incentives. (3) Separate individual from team evidence. (4) Test whether the prior task matches the new company's hardest task. (5) Seek one disconfirming reference and one direct artifact. (6) Kill when attribution is vague, references recycle the same story, or the claimed record belongs to a different person. (7) Unlock the next decision only after the direct evidence agrees with, or explains divergence from, reputation. [researcher inference · strong] People/capital map
• Examples —
  • Fairchild — documented behavior, researcher-reconstructed rule: Kleiner's letter surfaced a cohesive group; Rock and Coyle reframed placement as company formation; repeated sponsor rejection persisted until Sherman Fairchild acted. The technical work belonged to the eight and the capital decision to Fairchild Camera. [retrospective witness · strong] Last, printed p. 4 PBS, “Sherman Fairchild” sequence
  • Intel — documented application, retrospective: Noyce and Moore's Fairchild record let Rock syndicate rapidly; the short memo still specified technology scope, capital, terms, governance, and conflicts. Reputation compressed diligence but did not eliminate transaction design. [investor-stated · strong] Bancroft, printed pp. 45–48 Intel memo, PDF pp. 1–4
  • Apple — documented application, retrospective: Rock used prior Intel knowledge of Markkula plus show-floor demand to overcome his negative first impression of Jobs and Wozniak. Markkula's independent testimony shows that the trusted intermediary also performed material technical and organizational work. [investor-stated · strong] CHM Markoff interview, printed pp. 13–14 Markkula, printed pp. 21–23
  • ROLM — documented failure/exception: Rock's experience screen rejected Oshman, whose subsequent performance caused Rock to revise his view and recruit him later. Prior experience can be underweight evidence of learning capacity. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32
• Evidence — The model reconciles the Fairchild participant notes, Moore–Rock panel note, Markkula note, and source/role map in investments.md. These are independent witness families rather than repetitions of a single Rock retrospective. [researcher inference · strong]
• Limits, failure modes, and exception ledger — Access can masquerade as judgment; famous teams create survivorship bias; repeated accounts may share the same mythology. Fairchild prospect counts conflict; Intel subscription closings remain missing; Apple sourcing is often wrongly collapsed onto Rock; ROLM is the direct false-negative. Attempted but unresolved routes include Hayden Stone correspondence, a complete Fairchild prospect list, signed Intel subscriptions, Apple investment-committee or board minutes, and the request-only Science History Institute Rock interview. [researcher inference · strong] SHI catalog, request boundary
• Evolution and contradictions — The 1957 brokerage model became a 1968 reputation-led capital network and later a repeated people doctrine. Yet institutional capital and faster competition reduced time for relationship testing, while the Intel and Apple records show that trusted people did not make product, market, and security analysis unnecessary. [researcher inference · strong] HBS oral history, printed pp. 15–16 Apple offering memorandum, printed pp. 1–24
• Modern VC translation (as of 2026-08-19) — The transferable unit is a dated, claim-specific reputation graph; the non-transferable part is assuming a famous founder or warm introduction proves a new domain. Signals are referees with direct comparable observation, a team that has executed together, and primary artifacts matching the story. Likely misuse is circular reference checking inside one network. Analyst-created hypothetical: for a repeat founder entering regulated AI, credit prior recruiting and shipping evidence but require an independent safety-domain reference and reproduced system evidence before pricing the round. [researcher inference · moderate] NIST AI RMF Core, lifecycle evidence and governance

Match the financing architecture to the development risk and align every capital provider

• Rule — Translate the venture's next development horizon into a security, cash schedule, incentives, protective rights, governance, and syndicate that preserve operating runway while making dilution, downside, conflicts, and the next financing decision explicit. [researcher inference · moderate] Intel memorandum, PDF pp. 1–4
• Evidence status — [documented behavior · strong] for Intel's proposed 1968 architecture; [researcher inference · moderate] as a repeatable Rock model because other transactions lack comparable term records. Intel memorandum, PDF pp. 1–4
• Provenance and origin — The earliest full written instance is the 1968 Intel memorandum: ten-year 6% convertible debentures, three years of waived interest, subordination, conversion and anti-dilution protection, dividend limits, registration rights, price-aligned employee plans, and a mixed board. Its bounded cash forecast contradicts the caricature that Rock ignored plans. [contemporaneous record · strong] Intel memorandum, PDF pp. 2–4
• Mechanism and reasoning — Deferred cash yield extends development runway; conversion shares upside; protections constrain value transfer; founder common and employee ownership align builders; a syndicate diversifies funding capacity; governance and disclosure make conflicts visible. These mechanisms work only if subscriptions close and decision rights are enforceable. [researcher inference · strong] Intel memo, PDF pp. 1–4
• Historical and vehicle context — Intel was a pre-revenue semiconductor development program structured for financing by founders and a proposed private syndicate before modern preferred-stock templates. Rock's $10,000 common holding is documented; the memo's expected $300,000 debenture purchase is not closing evidence. [contemporaneous record · strong] Intel memo, PDF pp. 2–3 Moore interview, Intel-capital exchange
• When to use — Use when technical or regulatory milestones precede revenue, early cash must remain in the company, and several parties bring different capital or governance needs. Do not copy Intel's instrument where the milestone is unobservable, enforcement costs exceed protection, or terms impair hiring and iteration. [researcher inference · strong] Intel decision lab
• How to apply it — (1) Name the next value-changing evidence and cash required to reach it. (2) Model founder, employee, and investor ownership under success, delay, and another round. (3) Select cash-pay versus deferred consideration. (4) specify information, voting, board, conflict, and registration rights. (5) Identify each subscribing entity and closing evidence. (6) Kill if runway does not reach a test, incentives reward premature optics, conflicts lack a remedy, or “expected” capital is treated as committed. (7) Set the next check at independently verified milestone completion. [researcher inference · strong] Intel decision lab
• Examples —
  • Intel — documented application: The memo connects a development-year cash plan to convertible debt, founder common, employee pricing, investor protections, a syndicate, and board composition. It is a proposed packet rather than a complete closing ledger. [contemporaneous record · strong] Intel memo, PDF pp. 1–4
  • Fairchild — researcher-applied historical illustration: The later contract reconstruction combines founder and employee shares, a corporate loan/options, and a voting trust to preserve the team while supplying sponsor capital. It is not evidence that Rock held or applied the 1968 Intel rule in 1957. [documented behavior · moderate] CHM contract reconstruction, PDF p. 9
  • Ridge — researcher-applied illustration: Founders self-funded a prototype before Rock invested and brought H&Q; missing security and board-deadlock terms later matter because VC control, management change, and sale disagreement emerged. [retrospective witness · strong] Sell, printed pp. 13–15, 20
  • ELXSi — researcher-applied failure case: A plan sized for $40–50 million of sales met lower recalled sales and a closed financing market; no Rock-specific terms or follow-on decision survive, so staged financing is an analyst counterfactual, not a claimed historical remedy. [retrospective witness · moderate] Thomas, printed pp. 18–19, 23
• Evidence — The ex-ante core is the Intel memo note; transaction comparisons appear in the Intel, Fairchild, Ridge, and ELXSi labs. The cited underlying works include one contemporaneous financing document, a contract-history reconstruction, and independent participant accounts. [researcher inference · strong]
• Limits, failure modes, and exception ledger — Only Intel exposes a complete Rock-linked packet, and even it lacks a date, signature, executed subscriptions, closing certificate, and version identity. Fairchild's amount and prospect accounts conflict; Ridge and ELXSi terms are missing. Attempted routes still needed are Intel subscriber records, later cap tables/conversions, the Fairchild agreements, Davis & Rock partnership schedules, and company financing files. The shortfall is transaction-record survival, so the cross-case model remains moderate rather than being padded with inferred terms. [researcher inference · strong] Source retrieval ledger
• Evolution and contradictions — Rock's 1987 attack on five-year forecast precision coexists with Intel's bounded first- and second-year plan. The defensible synthesis is decision-horizon planning, not planlessness. [researcher inference · moderate] HBR, financial-projection passage Intel memo, PDF p. 2
• Modern VC translation (as of 2026-08-19) — The mechanism transfers; the exact debenture does not. Current NVCA forms include preferred-equity, voting, investor-rights, and milestone/time-tranche controls, all requiring tailoring. Signals are a costly technical gate, observable milestone, high early cash need, and meaningful financing risk. Likely misuse is punitive structure that prevents learning or converts weak underwriting into false security. Analyst-created hypothetical: tranche an infrastructure round against independently reproduced capacity and paid-pilot retention, with a pre-agreed stop or reprice review. [researcher inference · strong] NVCA Model Legal Documents, 2025–2026 update and tranched-financing sections

Keep the portfolio small enough that governance is real work

• Rule — Accept only as many investments and boards as the investor can serve through recruiting, finance oversight, difficult decisions, and long-duration attention; concentration is justified by service capacity, not by confidence rhetoric alone. [investor-stated · strong] Forbes, “How many boards” exchange
• Evidence status — [investor-stated · strong] for Rock's preference for a few companies and rejection of a large firm; [documented behavior · moderate] because the complete portfolio and time-allocation denominator remain unavailable. HBS oral history, printed pp. 8–9, 15–16 Forbes, “How many boards” exchange
• Provenance and origin — Davis & Rock's early concentration in SDS and Teledyne is the earliest observed pattern; Rock later says those two absorbed most of his time. The clearest articulation is his 2000 criticism that 17–18 boards leave too little time for each company. [investor-stated · strong] Moore–Rock panel, printed pp. 1–2 Forbes, board-count exchange
• Mechanism and reasoning — Fewer positions preserve attention for hiring, accounting, succession, and strategic adaptation; long involvement compounds context and trust. The hidden assumptions are that board help has value, ownership justifies the time, and concentration does not make one error fatal. [researcher inference · strong] Something Ventured part 2, printed pp. 10–13
• Historical and vehicle context — D&R was a small finite-life partnership, and Rock later used personal or Rock-branded vehicles rather than scaling a large institutional firm. Sparse competition and long boards made service possible; today's larger portfolios, reserves, platform teams, and formal key-person duties change the capacity equation. [researcher inference · strong] Vehicle table
• When to use — Use when the strategy promises investor labor and governance rather than capital alone. Do not use concentration merely because the founder is famous, when reserve needs are unmodeled, when board information rights are weak, or when sector correlation dominates company-specific oversight. [researcher inference · strong] HBS oral history, printed pp. 8–9, 15–16
• How to apply it — (1) List required work per company through the next milestone. (2) Assign a named person and weekly/quarterly capacity. (3) map board, audit, recruiting, reserve, and conflict obligations. (4) Stress a simultaneous-crisis scenario. (5) Kill the new deal if no accountable owner has capacity or if existing companies lose promised attention. (6) Reopen capacity after a governance handoff, exit, or explicit service-level reduction. [researcher inference · strong] People/capital map
• Examples —
  • SDS — documented application, retrospective: Rock says SDS absorbed much of his early partnership time and he served as chair; company operation belonged to Palevsky and the team. [investor-stated · strong] Moore–Rock panel, printed pp. 1–2 SDS panel, printed pp. 8–14
  • Teledyne — documented application, retrospective: Rock says Teledyne also absorbed much of his early partnership time and he served for decades, while assigning operating and acquisition judgment to Singleton. [investor-stated · strong] Moore–Rock panel, printed pp. 1–2 HBS, printed pp. 8–9
  • Intel — documented application, retrospective: Rock describes staff-meeting attendance, recruiting, finance controls, audit work, and public-offering preparation across a long board tenure while assigning product decisions to management. [investor-stated · strong] Something Ventured part 2, printed pp. 10–13
  • Apple — bounded application: Rock's later governance included CEO recruitment and succession, but his involvement was smaller than at Intel, SDS, or Teledyne and collective board attribution remains incomplete. [investor-stated · strong] HBS, printed pp. 11–13 Markkula, printed p. 34
  • Diasonics — documented failure/exception: Rock says he did not monitor closely enough despite chair/audit responsibilities; concentration as a preference did not guarantee attention in practice. [investor-stated · strong] Bancroft, printed pp. 63–68
• Evidence — The doctrine and cases connect the HBS note, Founding Funders note, Something Ventured note, and portfolio forensics. The four underlying sources include Rock interviews, participant testimony, and deal evidence. [researcher inference · strong]
• Limits, failure modes, and exception ledger — The record lacks a complete portfolio, formal board-cap, hours, reserves, ownership-weighted exposure, and contemporaneous service evaluation. Long tenure may reflect inertia or access rather than value. Diasonics is the direct attention failure; Markkula describes selective rather than detailed Intel engagement; East Coast Benrus board work was delegated. Missing routes include calendars, minutes, committee records, fund schedules, board evaluations, and partner work allocation. [researcher inference · strong] Markkula, printed pp. 17–18 Fialkov, printed p. 143
• Evolution and contradictions — Rock's early concentrated practice became an explicit critique of institutional scale by 2000. At the same time, he delegated geography and participated unevenly across companies, showing that “few” is not a complete operating design; named ownership of work matters more than a raw board count. [researcher inference · strong] Forbes, board-count exchange
• Modern VC translation (as of 2026-08-19) — Transfer capacity accounting, not Rock's historical number of boards. Signals are named ownership of recruiting and risk work, crisis slack, explicit handoffs, reserve plans, and LP disclosure of conflicts/key-person exposure. Likely misuse is under-diversifying while providing no actual service. Analyst-created hypothetical: an IC rejects a fifth board seat until it shows who owns two simultaneous financings and one CEO search; ILPA's current framework makes portfolio reporting, conflicts, and key-person governance visible but does not prescribe a universal board cap. [researcher inference · strong] ILPA Principles 3.0, printed pp. 10–22 and glossary p. 41

Make bad news travel faster than reassuring narrative

• Rule — Build an independent finance and assurance system that can challenge shipments, revenue quality, receivables, inventory, product status, and management narrative before capital or reputation commits the board to a false picture. [investor-stated · strong] HBR, finance and anonymous-company passages
• Evidence status — [investor-stated · strong] in HBR and the Levitt letter; [documented behavior · moderate] through Rock's self-reported Intel, Apple, and Teledyne audit practice; [investor-stated · strong] in his Diasonics self-critique. HBR, finance and anonymous-company passages Levitt letter, PDF p. 1
• Provenance and origin — Intel's 1968 memo names an auditor and discloses potential Rock conflicts, but the first accessible full articulation is HBR in 1987: a tough finance function must resist premature shipment, doubtful orders, and comforting internal reports. [contemporaneous record · strong] Intel memo, PDF pp. 1, 3–4 HBR, CFO and anonymous-company passages
• Mechanism and reasoning — Independent measurement shortens the distance between operational fact and board action; separating assurance from consulting removes self-review incentives; visible exceptions allow the board to stage capital and management responses. The system fails if the board ignores it or if independence exists only on paper. [researcher inference · strong] Levitt letter, first body paragraph
• Historical and vehicle context — Rock's examples span private growth companies and later public-company audit committees. Accounting rules and product telemetry differed, but the governance problem was stable: managers and advisers can benefit from reassuring the capital provider. [researcher inference · strong] HBR, printed pp. 63–67
• When to use — Use before a priced round, revenue recognition, a major shipment, forecast-based hiring, and every board/audit review. Do not confuse more reporting with better evidence, or an auditor with product-market validation. [researcher inference · strong] HBR, CFO and product-status passages
• How to apply it — (1) Define product-ready, booked, shipped, accepted, collectible, and retained separately. (2) Give finance and assurance direct board access. (3) reproduce a sample of product and customer evidence. (4) log overrides and conflicts. (5) require audit-committee preapproval of permitted outside services. (6) kill or pause financing when management blocks verification, repeatedly changes definitions, or exceptions outrun cash. (7) unlock the next check after independent reproduction and corrective ownership. [researcher inference · strong] Diasonics lab
• Examples —
  • Intel — documented application, retrospective: Rock describes conservative accounting, audit structure, staff meetings, and offering preparation as board mechanisms; the public record does not isolate outcomes caused by him. [investor-stated · strong] Something Ventured part 2, printed pp. 10–13
  • SDS — documented application, retrospective: Rock and participants connect chair-level concern to accounting honesty and financing continuity; precise decision minutes remain missing. [investor-stated · strong] SDS panel, printed pp. 8, 23–24, 30
  • Diasonics — documented failure: Rock later identifies aggressive sales booking and revenue-recognition problems, competitor financing, competitive product issues, CEO problems, and his own insufficient attention; reorganization followed, but later adoption of a stronger process is not documented. [investor-stated · strong] Bancroft, printed pp. 63–68
  • Daisy — investor-stated boundary case: Rock uses Daisy as a growth/control failure, but no public record proves Rock invested or governed; it supports the general mechanism, not personal application. [investor-stated · moderate] HBR, Diasonics/Daisy passage Ninth Circuit, 1996-09-24 company history
• Evidence — The model joins the HBR note, Levitt letter note, SDS panel note, and failure postmortems. Independent underlying sources include Rock's essay and signed regulator letter, company-participant testimony, court history, and Rock's oral-history self-critique. [researcher inference · strong]
• Limits, failure modes, and exception ledger — HBR's anonymous stories cannot be tied to companies; the Levitt letter is self-report and contains a tax exception; the final SEC rule did not adopt a total ban; Diasonics lacks minutes that date visible warnings and Rock's response. Attempted but missing routes include audit charters, engagements, non-audit fees, CFO records, shipments, receivables aging, board packets, and the 1996/1998 Levitt correspondence listed in Columbia's finding aid. [researcher inference · strong] SEC final rule, summary and alternatives Levitt papers finding aid, printed p. 167
• Evolution and contradictions — The model develops from operating finance control in 1987 to structural auditor independence in 2000. Rock's earlier Diasonics failure makes the later rule credible as learning but does not prove causal process adoption; no record connects the two explicitly. [researcher inference · moderate] Bancroft, printed pp. 63–68 Levitt letter, PDF p. 1
• Modern VC translation (as of 2026-08-19) — The principle survives as continuous, independent evidence; the historical audit boundary must be adapted to modern software and AI lifecycle risks. Signals are hidden evaluations, metric-definition changes, override growth, self-review, and revenue exceptions. Likely misuse is checklist theater or treating every outside service as disqualifying. Grounded current rule example: PCAOB Rules 3524–3526 prescribe preapproval or communication for specified tax and non-audit services. Analyst-created hypothetical: condition a follow-on on independent reproduction of retention and model-risk metrics. [researcher inference · strong] PCAOB ethics and independence rules NIST AI RMF Core, continuous lifecycle governance

Redesign management and information paths when complexity changes

• Rule — Do not ask a founding management system to carry a company through every scale. Change roles, reporting paths, executive depth, and board intervention when the next stage creates decision bottlenecks or hides information. [investor-stated · moderate] HBR, three-stage management passage
• Evidence status — [investor-stated · moderate] as a three-stage management rule in HBR; [documented behavior · moderate] across Apple and Intel succession/governance accounts; [investor-stated · moderate] in named Diasonics and Daisy failures. HBR, three-stage management passage
• Provenance and origin — The earliest accessible articulation is 1987: founder-does-everything, direct functional reports, then a layered organization, with a period estimate of roughly $100–$200 million in sales for the third transition. The causal rule is information-path redesign, not the dollar threshold. [investor-stated · moderate] HBR, phrase anchors “a company's growth has three stages” and “$100 million to $200 million”
• Mechanism and reasoning — Complexity increases coordination load and makes one person's direct observation less complete. Functional owners and later layers can improve decision quality only if accountability stays explicit; premature hierarchy adds latency and status without better evidence. [researcher inference · strong] HBR, stage and bad-news passages
• Historical and vehicle context — Rock's rule arose from capital-intensive hardware companies with manufacturing, sales, service, and public-market transitions. The exact revenue threshold is era- and business-model-specific; the investor's concentrated board capacity enabled recruiting and succession work. [researcher inference · strong] Historical context
• When to use — Use when founder bottlenecks, unowned decisions, inconsistent functional data, repeated rework, or a new regulatory/manufacturing/channel burden appears. Do not professionalize merely because a benchmark says to, and do not assume replacing a founder fixes a weak market. [researcher inference · strong] HBR, management-stage and Diasonics/Daisy passages
• How to apply it — (1) Map recurring decisions and their evidence owners. (2) identify founder bottlenecks versus unique founder advantages. (3) recruit the missing function before adding layers. (4) define board and management decision rights. (5) test a management transition against product, customer, and cash milestones. (6) kill a proposed executive hire if the problem is unnamed or the candidate only adds hierarchy. (7) reopen design after each business-model or scale discontinuity. [researcher inference · strong] HBR management-stage passage
• Examples —
  • Apple — documented application, retrospective: Rock links the case to bringing in experienced management and later board response to CEO performance; Markkula and the collective board share the work, and individual votes remain unavailable. [investor-stated · strong] Stanford Law Q&A, Apple sequence Markkula, printed p. 34
  • Intel — researcher-applied illustration: Rock participated in hiring and governance while Noyce, Moore, Grove, and other employees operated; management rather than Rock drove the memory-business adaptation. [researcher inference · strong] Something Ventured part 2, printed pp. 10–13 Moore–Rock panel, printed pp. 22–31
  • Diasonics — investor-stated failure application: Rock says growth and weak controls preceded management reorganization; missing decision records prevent dating when a transition should have occurred. [investor-stated · strong] Bancroft, printed pp. 63–68
  • Daisy — investor-stated boundary/failure: Rock presents Daisy as initial success followed by a management/control reorganization, but Rock's investment and governance role are unproved. [investor-stated · moderate] HBR, Diasonics/Daisy passage
• Evidence — The rule is developed in the HBR note, constrained by the Markkula note and Moore–Rock note, and tested in the Diasonics and Ridge labs. The underlying works include Rock's method essay, participant histories, and company evidence. [researcher inference · strong]
• Limits, failure modes, and exception ledger — No dataset validates Rock's revenue threshold or compares founder retention, replacement, and layered management outcomes. Sculley is called a lucky nontechnical exception; Diasonics and Daisy lack decision dates; Ridge's management change and sale disagreement show that professionalization can create agency conflict. Missing routes are hiring packets, org charts, board minutes, executive scorecards, and management-change votes. [researcher inference · strong] HBR, Sculley and Diasonics/Daisy passages Sell, printed p. 20
• Evolution and contradictions — Rock's model values strong founders yet rejects the claim that weak management can always be replaced. The tension resolves only probabilistically: build executive depth around a reality-facing founder, but change authority when evidence shows the current system cannot carry the next stage. That reconciliation is [researcher inference · moderate], not Rock's explicit formula. Moore–Rock panel, printed pp. 9–11
• Modern VC translation (as of 2026-08-19) — Transfer decision-right redesign; discard the fixed revenue threshold. Signals are unowned metrics, founder queues, cross-functional failures, hidden exceptions, and repeated execution resets. Likely misuse is importing a “professional” executive before defining the problem or stripping the founder's unique advantage. Analyst-created hypothetical: add a product-risk owner only after mapping which AI evaluation, release, and customer decisions lack accountable evidence; NIST's lifecycle framework supports explicit governance functions without prescribing an org chart. [researcher inference · moderate] NIST AI RMF Core, Govern/Map/Measure/Manage

Update on observable evidence without pretending to time what you cannot know

• Rule — Treat the original view as revisable: identify what evidence would reverse it, update when that evidence arrives, and distinguish a warranted company decision from confidence about market timing or later outcomes. [researcher inference · moderate] HBS oral history, Apple, printed pp. 11–12 Forbes, closing market exchange
• Evidence status — [documented behavior · strong] in Rock's Apple decision change and later ROLM/Oshman reassessment; [investor-stated · strong] in his bounded Intel board credit and 2000 valuation-versus-crash distinction; [researcher inference · moderate] as one unified model. CHM Markoff interview, printed pp. 13–15 Forbes, closing exchange
• Provenance and origin — The earliest deal-level public instance is the 1978 Apple account: Rock's negative founder impression changed after trusted Markkula evidence and visible demand at a computer show. The exact contemporaneous decision memo is missing, so the sequence is retrospective. [investor-stated · strong] HBS oral history, printed pp. 11–12
• Mechanism and reasoning — Precommitting to reversal evidence reduces ego defense; distinguishing operating evidence from macro prediction keeps a board from claiming knowledge it lacks; assigning product decisions to operators preserves attribution and lets the board approve adaptation without inventing it. [researcher inference · strong] Pioneers, Intel memory discussion 01:02:29–01:04:18
• Historical and vehicle context — Rock's long board tenures created repeated decisions after the initial check, while public-market booms created pressure to confuse price discomfort with timing ability. These are different inference horizons and should not be collapsed. [researcher inference · strong] Forbes, board and market exchanges
• When to use — Use at every thesis review, management change, product pivot, follow-on, and exit. Do not update from outcome alone; later success does not prove an earlier pass irrational, and later failure does not prove the initial bet lacked merit. [researcher inference · strong] Decision labs
• How to apply it — (1) Write the decision-date belief and missing facts. (2) list reversal and confirmation evidence before outcome. (3) assign who can produce each fact. (4) update probability and action separately. (5) record whether the board originated, approved, or merely observed the change. (6) kill or change course when a named threshold breaks; do nothing when the evidence is only price excitement or hindsight. (7) set the next review date and artifact. [researcher inference · strong] Decision labs
• Examples —
  • Apple — documented application, retrospective: Show demand and confidence in Markkula reversed Rock's first impression; the subsequent outcome cannot be imported into that evidence packet. [investor-stated · strong] Stanford Silicon Genesis, paragraph beginning “Well, Mike Markkula was” Apple offering memo, printed pp. 1–24
  • Intel memory exit — documented board application, bounded: Rock approved management's decision but assigns the reasoning to Grove and Moore; Moore says Intel retained profitable memory categories, correcting the simplistic claim that it left all memory. [investor-stated · strong] Something Ventured part 2, printed p. 12 Moore–Rock, printed pp. 30–31
  • ROLM/Oshman — documented failure and update: Rock passed on Oshman for experience/business-judgment reasons, then later hired him at Echelon. That later action supports belief revision; it does not prove the original proposal should have been funded. [investor-stated · moderate] Pioneers, 01:09:42–01:10:32
  • Internet valuations — boundary case: Rock called prices ridiculous but declined to predict a crash, separating valuation assessment from a timing forecast. [investor-stated · moderate] Forbes, closing exchange
• Evidence — The model connects the Apple oral histories, Pioneers note, Moore–Rock note, and frozen decision labs. The independent sources include Rock interviews, company participant testimony, and ex-ante Apple materials. [researcher inference · strong]
• Limits, failure modes, and exception ledger — Retrospective narration can recast ordinary adaptation as a rule; no Rock decision journal or probability record survives. Apple can be used to rationalize any exception, ROLM can generate hindsight punishment, and the Intel pivot is often misattributed to the board. Attempted but missing routes include Apple and Intel minutes, ROLM proposal/diligence files, dated thesis updates, and bubble-era deal identities/results. [researcher inference · strong] Stanford Silicon Genesis open questions
• Evolution and contradictions — The Apple change and ROLM reassessment show flexibility, while the people doctrine risks becoming unfalsifiable if every success is post-hoc evidence of a “good person.” Rock's admission of luck and residual failure is the necessary constraint. [researcher inference · strong] HBR, opening funnel and luck passage
• Modern VC translation (as of 2026-08-19) — Transfer precommitted falsifiers, versioned thesis updates, and precise decision attribution. Signals are independently reproduced metric changes, customer behavior, cost curves, regulation, and team learning; likely misuse is chasing noisy dashboards or calling every pivot intellectual honesty. Analyst-created hypothetical: before an AI follow-on, record the eval, retention, cost, and governance thresholds that imply fund, stage, or stop, then version the decision when evidence changes; NIST's lifecycle approach supports continuous measurement and management. [researcher inference · moderate] NIST AI RMF Core

Make each follow-on earn a new decision; an advocate who disagrees funds the continuation

• Rule — Re-underwrite additional capital against current evidence and portfolio opportunity cost. If investors disagree and one insists on continuing, make that advocate supply the incremental capital rather than socializing an unproven rescue. [investor-stated · strong] Something Ventured part 2, printed pp. 28–29, block 09
• Evidence status — [investor-stated · strong] as a general stop-funding rule in Rock's raw interview; [researcher inference · provisional] in company applications because no named vote, reserve ledger, or Rock follow-on amount is public. Something Ventured part 2, printed pp. 28–29, block 09
• Provenance and origin — The only direct articulation located is the 2008–2009 Something Ventured interview. Earlier D&R records show uninvested commitments and distressed companies but do not connect those facts to this explicit rule. [investor-stated · strong] Something Ventured part 2, printed pp. 28–29 Kramlich oral history, printed p. 5
• Mechanism and reasoning — A fresh decision prevents sunk-cost escalation; putting incremental capital behind the strongest advocate reveals conviction and limits forced exposure for dissenters; explicit next-check evidence makes continuation falsifiable. It fails when control rights, signaling, or fiduciary duties make refusal value-destructive. [researcher inference · moderate] Something Ventured part 2, printed pp. 28–29, block 09
• Historical and vehicle context — Public evidence does not reveal D&R reserves, recycling, follow-on rights, or later Rock-vehicle policies. Early syndicates and personal capital could allocate a rescue differently from a modern LP fund with concentration and allocation duties. [researcher inference · insufficient public record] Vehicle comparison
• When to use — Use at bridge rounds, covenant relief, missed milestones, turnaround financing, and portfolio triage. Do not use as a threat that blocks information, or when a small bridge protects a clearly valuable process and the dissent rule would create a holdout windfall. [researcher inference · moderate] Qantel decision lab
• How to apply it — (1) Ignore sunk cost. (2) rebuild the current case, runway, downside, and alternatives. (3) compare with the best unfunded portfolio use. (4) record advocate, dissent, conflicts, and pro-rata capacity. (5) define amount, seniority, control, milestone, and maximum reserve. (6) kill when evidence does not improve, no accountable advocate funds, or a bridge cannot reach a decision. (7) unlock only on a named operational or financing milestone. [researcher inference · strong] Qantel lab
• Examples —
  • Qantel — researcher-applied illustration: Period reporting says Rock and co-investors stayed engaged and achieved a positive but ordinary company-level result; exact follow-ons, votes, and Rock proceeds are unknown. It is a triage case, not a documented application of the advocate-funds rule. [researcher inference · moderate] TIME, 1984, page 2
  • ELXSi — researcher-applied failure case: A lower-than-planned sales base, financing-market closure, merger, and exhausted runway make follow-on re-underwriting central; Rock's own continuation or stop action is absent. [retrospective witness · moderate] Thomas, printed pp. 18–19, 23
  • Diasonics — researcher-applied failure case: Management reorganization followed warning signs, but capital escalation, reserves, and stop votes are missing; the rule cannot be retrofitted as something Rock used. [researcher inference · strong] Bancroft, printed pp. 63–68
  • Davis & Rock — boundary evidence: Rock says the partnership did not deploy all commitments, which is consistent with willingness not to invest but does not prove company-level stop discipline. [investor-stated · strong] Moore–Rock panel, printed pp. 1–2
• Evidence — The direct rule is in the Something Ventured note; candidate applications are bounded in the Qantel lab, failure matrix, and vehicle table. The evidence shortfall is explicitly model-specific rather than filled with inferred decisions. [researcher inference · provisional]
• Limits, failure modes, and exception ledger — No public case links the stated rule to a named Rock vote. Qantel, ELXSi, and Diasonics have incomplete follow-on records; Ridge identifies a co-investor-led control shift but not Rock's funding action. Attempted and missing routes are capital-call and reserve schedules, pro-rata notices, bridge terms, board minutes, side letters, fund allocation policy, and company cash ledgers. This model remains provisional until one named application is documented. [researcher inference · insufficient public record] Uncertainty ledger
• Evolution and contradictions — Long company-building tenure can encourage patience, while re-underwriting requires willingness to stop. The public record states both values but does not show how Rock resolved them in a contested named financing. [researcher inference · moderate] Pioneers, company-building discussion 01:28:21–01:30:22
• Modern VC translation (as of 2026-08-19) — Transfer fresh underwriting, named advocacy, capped reserves, and milestone-linked continuation; do not mechanically force one partner to finance where fund documents allocate decisions collectively. Signals are runway, independently verified milestone progress, financing alternatives, insider conflict, and opportunity cost. Likely misuse is abandoning a strong company because ownership or optics are inconvenient. Analyst-created hypothetical: a bridge closes in two milestone tranches only if the sponsoring partner records the falsifiers and the fund discloses allocation/conflict treatment; modern NVCA documents recognize tranched financing, while ILPA emphasizes allocation and conflict transparency. [researcher inference · strong] NVCA Model Legal Documents ILPA Principles 3.0, printed pp. 10–22

Build companies for duration, but separate company quality from vehicle, liquidity, and conflict decisions

• Rule — Prefer founders and governance built for durable company creation over quick-exit arithmetic, while deciding distributions, sales, board departures, and conflicts on their own legal and economic facts rather than treating patience as an absolute. [researcher inference · moderate] HBR, company-building motive passage Something Ventured part 2, printed pp. 24–25
• Evidence status — [investor-stated · strong] for company-building motive and long board service; [investor-stated · strong] for D&R's term-driven SDS distribution and Rock's Apple/Intel conflict choice; [researcher inference · moderate] as an integrated duration model. HBR, company-building motive passage Something Ventured part 2, printed pp. 24–25
• Provenance and origin — The earliest observed behavior is Rock's long SDS/Teledyne involvement, but the first accessible broad articulation is 1987's preference for builders. By 2000–2002 he explicitly contrasts decades-long Intel/Teledyne boards with promoters seeking liquidity. [investor-stated · strong] Forbes, board and company-builder exchanges Pioneers, 01:28:21–01:30:22
• Mechanism and reasoning — Durable motive favors reinvestment, recruiting, and painful adaptation over terminal-value theater; long governance compounds company context. Yet a finite partnership, fiduciary sale duty, investor liquidity, or portfolio conflict can require action unrelated to company quality. [researcher inference · strong] Stanford Silicon Genesis, SDS sale and Apple conflict paragraphs
• Historical and vehicle context — Rock says D&R's seven-year life forced distribution of SDS stock and that successor-fund design could constrain LP selling. Rock's later personal capital and long public-company boards had different liquidity and governance constraints. [investor-stated · strong] Something Ventured part 2, printed pp. 24–25
• When to use — Use in founder motivation, fund/vehicle design, exit review, and board-conflict analysis. Do not use “long term” to reject a legally compelling premium, trap LPs, ignore concentration, or preserve a conflicted board seat. [researcher inference · strong] Stanford Silicon Genesis, SDS sale and Apple conflict paragraphs
• How to apply it — (1) Separate company thesis from vehicle term and investor liquidity. (2) ask whether the founder's operating choices favor durable capability or merely exit optics. (3) map fiduciary duties, distribution rights, conflicts, and alternative owners. (4) evaluate sale price and strategic risk without later integration hindsight. (5) kill a board role when conflict cannot be managed; distribute or sell when documents/duties require it. (6) document what decision is about the company and what is about the vehicle. [researcher inference · strong] Vehicle table
• Examples —
  • Intel — documented application, retrospective: Rock uses his 30-plus-year board tenure as evidence of company-building duration while assigning operating judgment to Moore, Grove, and the Intel team. [investor-stated · strong] Forbes, board exchange Moore–Rock panel, printed pp. 22–31
  • Teledyne — documented application, retrospective: Rock likewise cites decades on Teledyne's board while assigning operating and acquisition judgment to Singleton. [investor-stated · strong] Forbes, board exchange HBS, printed pp. 8–9
  • SDS/Xerox — documented boundary: Rock says the premium made rejection vulnerable to shareholder litigation and separately faults Xerox's post-acquisition strategy. A justified sale and a poor later integration can both be true. [investor-stated · strong] Stanford Silicon Genesis, paragraph beginning “Xerox came along”
  • Davis & Rock distribution — documented vehicle exception: The finite partnership term required SDS-stock distribution and contributed to the partners' separation; this was a liquidity/vehicle event, not a reversal of the SDS thesis. [investor-stated · strong] Something Ventured part 2, printed pp. 24–25
  • Apple board departure — documented conflict exception: Rock says he left Apple when the PowerPC campaign conflicted with Intel, choosing his longer tenure and greater investment in Intel. The departure does not establish loss of confidence in personal computing. [investor-stated · strong] CHM Markoff interview, printed p. 15
  • Bubble-era promotion — stated counterexample: Rock contrasts builders with promoters and quick-exit motives, but does not disclose the few Internet/telecom deals he joined or their outcomes. [investor-stated · moderate] Pioneers, 01:28:21–01:30:22
• Evidence — The model connects the HBR note, Founding Funders note, Something Ventured note, and vehicle analysis. Independent source families include Rock's essay, multiple oral histories, vehicle testimony, and company records. [researcher inference · strong]
• Limits, failure modes, and exception ledger — Long board tenure is not measured contribution; public records do not expose D&R's agreement, fees/carry, extensions, distribution detail, Rock's personal liquidity, or the outcome of bubble deals. The SDS premium is Rock's recollection, not a fairness record. Missing routes include D&R governing documents and LP statements, acquisition board materials, Intel/Apple conflict minutes, share-sale records, and named bubble investments. [researcher inference · strong] Uncertainty ledger
• Evolution and contradictions — The durable-company ideal survives across decades, while Rock's own cases show three legitimate exceptions: finite-vehicle distribution, a premium sale under fiduciary pressure, and a board conflict. These do not negate patience; they define where company-building motive stops being the only decision variable. [researcher inference · strong] Stanford Silicon Genesis, SDS and Apple passages
• Modern VC translation (as of 2026-08-19) — Transfer explicit time-horizon alignment and separate company/vehicle/conflict memos. Changed conditions include larger LP funds, formal key-person and extension terms, secondary liquidity, and denser conflict sets. Signals are founder reinvestment behavior, durable recruiting, LP term, reserve horizon, concentration, and unmanageable board conflicts. Likely misuse is calling illiquidity a virtue or rejecting beneficial liquidity to preserve identity. Analyst-created hypothetical: an IC supports a founder's long build but separately models fund extension, secondary sale, reserves, and conflict recusal; ILPA's current framework makes term, extension, key-person, conflict, and reporting controls explicit. [researcher inference · strong] ILPA Principles 3.0, printed pp. 10–22 and glossary p. 41

Evidence gaps and candidate models

• Inventory reconciliation — nine retained models. Every philosophy-signal section in all 29 source notes, plus writings.md, talks.md, investments.md, and the nine-heading model index supplied for investmentphilosophy.md was reconciled. Genuine duplicates were merged: listening, execution, integrity, and self-knowledge became the longitudinal people test; network access, team continuity, and reputation became earned-reputation underwriting; CFO truth channels and auditor independence became bad-news architecture; governance intensity and portfolio concentration became a capacity budget; patience, anti-promotion, exits, vehicle liquidity, and conflicts became the duration model. Stage financing, management redesign, belief updating, and follow-on discipline remained separate because their decisions and boundary conditions differ. [researcher inference · strong] Writings cross-source synthesis Talks recurring frameworks Deal decision labs
• Fragments merged or excluded rather than inflated into models. The generic “people” slogan is subsumed by the testable longitudinal screen; persistent sponsor search has only the Fairchild case and remains an example inside earned-reputation sourcing; Rock's $100–$200 million management threshold is a dated parameter inside the stage model, not its own rule; no formal reserve or portfolio-allocation rule survives; and the record supports no universal exit rule, only the company/vehicle/liquidity/conflict separation in the duration model. [researcher inference · strong] HBR, people, management-stage, and market-supply passages Fairchild lab Vehicle and reserve boundary
• Broad employee ownership as a recruiting system — plausible, excluded. Intel records and Moore/Rock testimony support a three-person discussion and an employee-option rationale; Markkula says he copied Intel's culture at Apple. The record does not establish Rock's sole authorship, a general allocation rule, or three Rock-linked application decisions. Needed proof: contemporaneous option plans, grant and vesting schedules, board minutes, and Rock's application in another company. [researcher inference · moderate] Moore–Rock panel, printed pp. 27–29 Markkula, printed pp. 19–20, 33
• Confidential deliberation with transparent final action — plausible, excluded from the venture model set. Two joint 2006 UC essays support a bounded governance tradeoff, but sentence-level Rock authorship is unknown, the application is public-university hiring, and no portfolio-company case or outcome test survives. Needed proof: drafting records, exact bill/history, later hiring data, and Rock-linked board applications. [documented behavior · moderate] August UC essay May UC essay
• Temporary public first-loss capital to mobilize private provision — isolated, excluded. Rock's signed 1955 organizational letter transmits a public-start-up/private-carrier proposal, but it predates his venture career and no later source connects it to his investing. Needed proof: Rock-authored reasoning, later cross-reference, and multiple investment applications. [contemporaneous record · strong] Congressional Record Appendix A1929
• Geographic proximity as a board-capacity rule — plausible, excluded. Fialkov says Rock referred East Coast Benrus board work because he did not want East Coast boards, but one retrospective case cannot establish a general geography model. Needed proof: portfolio geography, travel/capacity records, comparable accept/decline cases, and a stated Rock rule. [retrospective witness · moderate] Fialkov, paragraph oh-t-p1463, printed p. 143
• Immigrant hunger or hardship as an entrepreneurship predictor — excluded. Rock voices the intuition in a framed Stanford Q&A, but the demographic claims belong largely to the interviewer, causality is untested, and the rule risks stereotyping. Needed proof: a defined comparison set, confound controls, and actual Rock decision applications; absent that, it is not a decision-ready model. [investor-stated · provisional] Stanford Law Q&A, “immigrant energy” sequence
• Market-size-first selection — explicitly not attributed to Rock. In the 2002 Pioneers panel, Valentine prioritizes market size and Perkins qualifies complete-team requirements; those are valuable same-session counterarguments to Rock's people emphasis, not Rock models. [retrospective witness · moderate] Pioneers, 01:12:44–01:15:43
• Priority retrievals that could add, merge, or delete models. Lawfully inspect Done Deals pp. 139–148; request the Science History Institute 2002 Rock interview and Stanford Fairchild V0311 panel; locate D&R governing documents and portfolio/reserve schedules; retrieve Intel/Apple/Teledyne/SDS/Diasonics board and audit records; obtain named rejected-plan and reference files; and identify Rock's bubble-era deals. These routes were mapped but remain controlled, request-only, nonpublic, or unidentified—not silently unsearched. [researcher inference · strong] Done Deals controlled route SHI request-only catalog Stanford V0311 finding aid

What the evidence supports

[researcher inference · strong] The public record supports a company-formation and governance thesis more strongly than a personal-alpha thesis. Across Fairchild, Scientific Data Systems, Intel, and Apple, Rock repeatedly appears at the junction of trusted team access, financing architecture, recruiting, accounting discipline, and board escalation. The package also keeps the technical and operating work with Kleiner and the Fairchild eight, Palevsky and the SDS team, Noyce/Moore/Grove and Intel, and Jobs/Wozniak/Markkula and Apple. Profile edge and role map · deal people map

[researcher inference · moderate] The favorable edge claim is therefore narrow: in U.S. technology-company formation from roughly 1957 through the 1980s, trusted access could compress search and capital-formation time; longitudinal people assessment could expose inconsistency; and concentrated board work could improve recruiting and negative-information flow. Fairchild supports team-preserving formation, SDS financing/chair work, Intel explicit ex-ante structure, and Apple belief revision through Markkula and observable demand. None proves comparative selection skill by itself. Profile mechanism, predictions, and falsifiers · eight decision labs

The strongest supported lessons are bounded:

• Preserve an exceptional team while designing the financing — [researcher inference · strong]. Fairchild's engineers supplied the opportunity and built the company; Rock and Coyle reframed the employer search; Fairchild Camera supplied company capital; Kleiner represented the founders. [retrospective witness · strong] Jay Last, printed pp. 3–5 · Fairchild lab
• Let reputation compress uncertainty without replacing terms — [researcher inference · strong]. Intel's four-page memorandum still specifies capitalization, a proposed $2.5 million debenture financing, conversion and investor protections, board composition, and conflicts despite the founders' record. [contemporaneous record · strong] Intel memorandum, PDF pp. 1–4
• Make bad news travel — [researcher inference · strong]. Rock's HBR method and Levitt letter favor finance, audit, and board information paths; Diasonics is the adverse case showing that formal responsibility without attention does not work. [investor-stated · strong] HBR, printed pp. 63–67 · Levitt letter, PDF p. 1 · Diasonics lab
• Update on evidence and preserve attribution — [researcher inference · moderate]. Markkula and customer interest changed Rock's Apple view; Moore and Grove, not the board, owned Intel's product reasoning; later success at ROLM does not reconstruct the rejected proposal. [retrospective witness · strong] Markkula, printed pp. 21–29 and 43 · Moore–Rock, printed pp. 30–31 · Pioneers, 01:02:29–01:10:32

What could falsify the thesis

The favorable thesis predicts that unseen ordinary-case files will show repeated-contact diligence, trusted-referrer access, fast but structured syndication, small capacity-aware portfolios, explicit accounting and recruiting interventions, named reversal evidence, and better ordinary outcomes when those mechanisms were present. [researcher inference · moderate] Predicted evidence

It weakens or fails if complete opportunity and cash-flow records show no selection or outcome advantage after controlling for founders, access, co-investors, vehicle, vintage, and dominant winners; if investment memoranda show shallow people work; if board minutes show nominal participation; or if the economics disappear without SDS and other outliers. [researcher inference · strong] The public denominator cannot decide this: 26 named families were searched, but 0/26 has a complete vehicle-to-cash-flow chain and career coverage remains not computable. Reconciliation · portfolio limits

Four observed controls already prevent a heroic reading:

1. False positives and monitoring failure — [investor-stated · strong]. Diasonics combines revenue-quality problems, weak CEO judgment, and Rock's admission that he did not pay enough attention as audit chair. Bancroft, printed pp. 63–68
2. False negatives — [investor-stated · moderate]. Rock identified ROLM and Compaq as passes; the proposals, alternatives, prices, and terms are absent. Pioneers, 01:09:42–01:10:32
3. Residual error — [investor-stated · strong]. The 1987 funnel says roughly half of selected investments disappointed, so a people screen was never a deterministic law. HBR, opening funnel
4. Team, access, era, and outlier alternatives — [researcher inference · strong]. Extraordinary founders, scarce early capital, federal technology demand, partners and boards, survivor narration, and SDS's directional dominance can explain part of the record without a Rock-specific alpha estimate. Skeptical case

Attribution and denominator audit

The required material attribution cross-check passes only with bounded credit. Fairchild belongs operationally to the eight and financially to Fairchild Camera, with Rock/Coyle as arrangers; SDS to Palevsky/team, D&R/co-investors, Rock as chair/financier, and Hayden Stone as underwriter; Intel to Noyce/Moore/Grove/team for product and operations, and Rock/syndicate for financing, recruiting, and governance; Apple to Jobs/Wozniak/Markkula for product, underwriting bridge, and plan, with separate Venrock/Rock allocations and collective board action; Diasonics and Ridge preserve management, co-investor, audit, and board conflict rather than collapsing failure onto one actor. [researcher inference · strong] Profile map · deal map

Transfer limits for modern VCs

[researcher inference · strong] The transferable object is a procedure, not Rock's historical result: record who sourced and decided; distinguish demonstrated team history from reputation; specify the instrument, conflicts, and decision rights; budget governance capacity; give finance and assurance direct escalation; precommit reversal evidence; and re-underwrite follow-ons. Philosophy translation, as of 2026-08-19 · mental-model field guide

Changed conditions matter. Early syndicates of wealthy individuals, scarce institutional venture capital, founder-price common and convertible debentures, federal electronics demand, and long personal board tenures do not map mechanically onto an LP-backed fund with allocation policies, reserve budgets, formal preferred-stock rights, secondary liquidity, and denser conflicts. [researcher inference · strong] Historical context · dimension-matched comparison

The dated contemporary controls are narrow. NVCA model documents show available legal dimensions, ILPA supplies governance and conflict principles, PCAOB rules supply auditor-independence requirements, and NIST AI RMF supplies lifecycle measurement language. None proves that a Rock rule causes current venture returns. [researcher inference · strong] Likely misuses are substituting prestige for diligence, treating a board seat as efficacy, forcing a personal-capital follow-on rule into collective fund documents, turning a 1987 revenue threshold into a scaling law, or calling illiquidity long-term virtue. Writings translation · model translations

Unresolved evidence

The decisive gaps are named and claim-changing:

• the 1957 Fairchild agreement and complete prospect/rejection ledger;
• D&R, Rock–Kramlich, Arthur Rock & Associates, and Arthur Rock & Co. governing documents, LP statements, allocations, reserves, carry, and distributions;
• Intel subscription closings, board/committee minutes, recruiting files, and conflict records;
• Apple's purchaser schedule, board minutes, transfer/sale records, and the unresolved 1993 complaint disposition;
• Diasonics audit packets, revenue evidence, warning dates, board responses, ownership, and realized losses;
• ROLM and Compaq proposals, diligence, terms, alternatives, and contemporaneous pass records;
• ordinary-case decisions and cash flows sufficient to test access, people assessment, governance, and ex-outlier performance; and
• the request-only Science History Institute Rock interview, reproduction-controlled Stanford V0311 panel, and lawfully accessible Done Deals chapter.

Highest-value source retrievals · deal uncertainty ledger · writing gaps · talk gaps

[researcher inference · strong] These gaps do not justify filler: legitimate routes and barriers are logged. They do prohibit audited personal performance, a clean ex-SDS calculation, a complete career denominator, and a claim that Rock's public rules were uniformly applied.

Central-claim citation audit trail

Ordinary-claim citation sample

Sampling rule: within each of the seven pre-review root documents, enumerate externally checkable claim-bearing blocks with a claim-local link in source order after excluding headings, navigation-only blocks, and the central claim instances audited above; inspect the first eligible block at or after 25% and the first at or after 75% of that document's eligible sequence. This deterministic spread yields exactly two claims per document and 14 claims total.

Twenty-point acceptance evidence

QA disposition

[researcher inference · strong] Acceptance is bounded to the public record: the package passes the twenty structural and evidentiary requirements after reconciling the 43 DEALS/control routes and replacing the retired HBS direct-download URL with the stable official exhibit route. The accepted thesis is formation/governance skill around exceptional teams; Rock-personal alpha, a complete return record, and a universal people-selection law remain unproved.

Source-ranking protocol

Priority is retrieval order, not endorsement: ★★★ must process; ★★ should process; ★ is a lower-priority lead. Reliability A denotes direct testimony or an authoritative record, B a strong institutional or secondary source, and C a discovery lead, interested paraphrase, or unprocessed restricted source. All checkboxes remain empty because SCOUT maps sources; it does not process later-phase evidence.

Direct voice, participant interviews, and filmed testimony

• ★★★ | Arthur Rock oral history, Bancroft Library | https://digicoll.lib.berkeley.edu/record/103610 | interviews 2008–2009 | official university oral-history record and full 76-printed-page transcript; reliability A for Rock's retrospective account | Deepest direct source for biography, vehicle sequence, named companies, Apple and Intel roles, institutional-capital criticism, and Diasonics; printed pp. 63–69 and 75 are priority adverse passages, but recurring stories require contemporary controls.
• ★★★ | Arthur Rock, HBS Entrepreneurs Oral History Collection | https://www.library.hbs.edu/special-collections-and-archives/exhibits/entrepreneurs/arthur-rock | 2001-03 | official HBS 47:12 interview, transcript, and institutional profile; reliability A for attributed speech and B for host claims | Compact direct account across Fairchild, D&R, Teledyne, SDS, Intel, Apple, decisions, and criteria; HBS's reported $3 million to $100 million D&R figure is unaudited and must not become a personal return.
• ★★★ | Arthur Rock oral history, Science History Institute | https://digital.sciencehistory.org/works/ayd1qb3 | 2002-10-09 | official archive record for oral history 0265, 29-page transcript and 01:45:00 audio; reliability A for metadata and Rock's account | Semiconductor-specialist route spanning Hayden Stone through Intel; full transcript/audio require the archive request form, and an impossible catalog education year is a warning not to treat metadata as verified body text.
• ★★★ | An Evening with Arthur Rock | https://archive.computerhistory.org/resources/access/text/2012/05/102658253-05-01-acc.pdf | 2007-05-01 | official CHM John Markoff interview transcript, ref. X4795.2009; reliability A for participant speech | Full direct account of Fairchild, Intel, Apple, competition, firm scale, and company building; printed pp. 3 and 14–15 are priority, while jokes, memory limits, and repeat narration require filings and witnesses.
• ★★★ | Gordon Moore and Arthur Rock Oral History Panel | https://archive.computerhistory.org/resources/access/text/2015/09/102739934-05-01-acc.pdf | 2014-07-09 | official CHM 32-page paired-witness transcript, ref. X7227.2015; reliability A for attributed speech | Best live role-boundary control for D&R, Intel financing, product decisions, and institutional capital; printed pp. 19–21 support the $2.5 million raise and Moore corrects Rock's later-financing memory at pp. 30–31.
• ★★★ | Something Ventured Film Transcripts collection | https://sova.si.edu/components/NMAH.AC.1467 | Smithsonian component dates the Rock transcript family 2008-09-04; Stanford item 1 says approximately 2009 and item 2 says 2009-11-18 | official Smithsonian collection record and two full Stanford-preserved Rock transcripts, collection NMAH.AC.1467; reliability A for custody and attributed speech, with a repository date conflict preserved | One raw-interview family with two sessions: Stanford PURLs https://purl.stanford.edu/jb183tm5048 and https://purl.stanford.edu/gq780fg2129, direct 66-page PDF https://stacks.stanford.edu/file/jb183tm5048/Arthur_Rock_Transcript.pdf and 34-page PDF https://stacks.stanford.edu/file/gq780fg2129/Aurthur_Rock_Transcript_2.pdf; item 2's internal header says “ARTHUR ROCK 2009 - 1-11,” and the official filename misspells “Arthur.”
• ★★★ | Jay T. Last oral history | https://archive.computerhistory.org/resources/access/text/2013/05/102658211-05-01-acc.pdf | 2007-09-15 | official CHM 16-page participant transcript recorded in Beverly Hills, interviewed by Craig Addison, ref. X4158.2008; reliability A for Last's retrospective account | Essential Fairchild founder countervoice: printed p. 4 says roughly 20–30 prospects declined and credits Rock plus Bud Coyle with proposing a company, constraining 35/45 counts and lone-founder stories.
• ★★★ | Mike Markkula oral history | https://archive.computerhistory.org/resources/access/text/2012/08/102746385-05-01-acc.pdf | 2012 | official CHM participant transcript; reliability A for Markkula's account | Printed pp. 21–29 and 43 distinguish Valentine, Markkula, Venrock/Peter Crisp, Rock, financing, and later board contribution; primary Apple filings control shares and dates.
• ★★★ | Herman Fialkov oral history | https://digital.sciencehistory.org/works/8910jv49c | oral history record accessed 2026-08-19 | official Science History Institute participant interview; reliability A for Fialkov's account | D&R limited-partner-side route describing the 1961 solicitation and reported $100,000-to-$2 million stock distribution; it is witness testimony, not an audited partnership statement.
• ★★★ | Arthur Rock and Eugene Kleiner, Forbes ASAP interview and artifacts | https://www.forbes.com/asap/2000/0529/145_print.html | 2000-05-29 | paired direct interview in business press; reliability B | Joint origin account plus route to the surviving Kleiner letter and Rock's crossed-out 1957 prospect list; editorial rhetoric is not evidence and the list is not enumerated in the accessible body.
• ★★★ | Silicon Valley, PBS American Experience | https://www.pbs.org/wgbh/americanexperience/films/silicon/ | 2013 | edited documentary with public transcript and Rock clips; reliability B | Accessible direct-voice route in which Rock gives a 45-company Fairchild count; curated editing prevents treating the clip as a complete interview.
• ★★★ | Fairchild Semiconductor 50th Anniversary videorecording | https://oac.cdlib.org/findaid/ark:/13030/kt8t1nf6wj | 2007-10-04 | Stanford University Archives finding aid, collection V0311, two DVDs; reliability A for metadata | Panel with Julius Blank, Jay Last, Gordon Moore, Arthur Rock, and Leslie Berlin; open for on-site research but not online, with reproduction permission required.
• ★★ | Gordon Moore's Journey | https://fortune.com/2012/09/24/gordon-moores-journey/ | 2012-09-24 | founder interview in business press; reliability B | Moore recalls Noyce and himself at about $250,000 each and Rock at $10,000, an independent witness route for separating founder equity from the $2.5 million debenture syndicate.
• ★★ | The Legacy of Fairchild | https://archive.computerhistory.org/resources/text/Oral_History/Fairchild_at_50/102658284.05.01.acc.pdf | 2007 | official CHM multi-participant panel transcript, ref. X4208.2008; reliability A for attributed speech | Printed p. 3 again says the group sought jobs and Rock plus a Hayden Stone colleague proposed a company; later pages add incentive and spinout context.
• ★★ | Reid Dennis raw Something Ventured transcript, part 2 | https://stacks.stanford.edu/file/mj877gg1809/Reid_Dennis_Transcript_2.pdf | 2008-09-25 | official Stanford/Smithsonian participant transcript; reliability A for Dennis's account | Printed pp. 29–31 praise Rock's long Intel committee work and company-building horizon; useful direct peer and governance witness, not a return audit.
• ★★ | Dick Kramlich oral history, part 1 | https://archive.computerhistory.org/resources/access/text/2016/03/102740064-05-01-acc.pdf | 2015 | official CHM former-colleague oral history; reliability A for Kramlich's account | High-value route for post-Davis Rock–Kramlich vehicles, fundraising, Apple allocation, and partnership dissolution; quantities require scan-level checking because a related Berkeley transcript contains obvious transcription/OCR defects.
• ★★ | Arthur Rock on the Early Venture Capital Decisions That Sparked Decades of Innovation | https://law.stanford.edu/stanford-lawyer/articles/legal-matters-arthur-rock-on-the-early-venture-capital-decisions-that-sparked-decades-of-innovation/ | 2023-01-05 | complete Stanford Lawyer edited Q&A by Sharon Driscoll, with Rock interviewed by Joseph Grundfest; reliability A for attributed speech and B for editorial framing | Full public body is available through the non-paywalled official-page renderer https://r.jina.ai/http://law.stanford.edu/stanford-lawyer/articles/legal-matters-arthur-rock-on-the-early-venture-capital-decisions-that-sparked-decades-of-innovation/ despite direct automated HTTP 403; use question/phrase anchors for Intel legal design, governance and vesting, Apple/Jobs, luck, and education.
• ★★ | Stanford Silicon Genesis interview with Arthur Rock | http://silicongenesis.stanford.edu/transcripts/rock.htm | 2002-11-12 | full Stanford Rob Walker–Arthur Rock transcript recovered from Common Crawl; reliability A for attributed speech and archive fixity | San Francisco interview with speaker labels RW and AR; original is dead, but full 24,433-byte HTML is preserved in CC-MAIN-2014-10 at timestamp 20140312135648, digest sha1:N7RKO3U75OLYMTWKHZTQKXQAXG6QCAPC, WARC crawl-data/CC-MAIN-2014-10/segments/1394021856395/warc/CC-MAIN-20140305121736-00064-ip-10-183-142-35.ec2.internal.warc.gz, offset 162328150, length 9739; WARC-Date 2014-03-12 and page Last-Modified 2003-06-23.
• ★★ | Scientific Data Systems Alumni Oral History Panel | https://archive.computerhistory.org/resources/access/text/2012/04/102658201-05-01-acc.pdf | 2007-11-29 | official CHM 42-page multi-participant transcript, ref. X4207.2008; reliability A for attributed speech | Company-specific direct Rock and participant-witness control on SDS financing, board role, capitalization, competition, and growth; Rock appears at printed pp. 8–9, 12–14, 23–24, 30, and 35–36.
• ★★ | Triumph of the Nerds: The Rise of Accidental Empires | https://www.pbs.org/nerds/part1.html | premiered 1996-06 | official PBS full program transcript, produced by John Gau Productions and Oregon Public Broadcasting for Channel 4 and PBS; reliability B for edited direct testimony | Earliest located public Rock moving-image testimony; Part I preserves his short first-person description of meeting Jobs, but the underlying full interview was not located, so use only the aired clip.
• ★★ | Pioneers of Venture Capital | https://computerhistory.org/events/pioneers-venture-capital/ | 2002-09-30 | official CHM event record plus digitized catalog item 102706131/X4859.2009 with full hosted media asset 140793 and host-generated “Speech to text” layer; reliability A for custody, participants, date, and media, B for context-controlled paraphrase after speaker turns are bounded by named prompts and distinctive facts; exact wording and quotations require official-video validation | Gordon K. Davidson moderates William Draper, Pitch Johnson, Burton J. McMurtry, Thomas Perkins, Arthur Rock, and Donald T. Valentine. The catalog payload contains 1,478 unlabeled timed-text segments whose start times span 00:00:00–01:54:01; its archival extent is 01:41:06. Preserve the discrepancy, do not treat the final segment start as verified runtime, and keep the event page, catalog, and official YouTube upload as one appearance family.
• ★★ | Up on the Ridge: The Story of a Workstation Pioneer | https://www.bitsavers.org/pdf/chac/CHAC_Analytical_Engine/3.4_Fall_1996.pdf | Fall 1996 | Computer History Association of California interview with Ridge cofounder John Sell in The Analytical Engine, vol. 3.4; reliability A for Sell’s participant account and B for archive custody | Printed pp. 13 and 15 fix May 1980 founding, founder self-funding, the pre-Christmas 1981 first VC close, Rock’s technical review/investment, his H&Q introduction, and split investment; printed p. 20 says Ridge did not go bankrupt, but founders and VC owners disagreed over sale terms before new management replaced them.
• ★★ | Oral History of A. Thampy Thomas | https://archive.computerhistory.org/resources/access/text/2019/01/102740189-05-01-acc.pdf | recorded 2017-08-03 | official CHM founder oral-history transcript, ref. X8292.2018; reliability A for Thomas’s account | Printed pp. 18–19 and 23 identify Tata’s recalled investment, Rock as an investor, Hambrecht as a director, the 1984 demand/financing shock, $25m sales against $40–50m plan, and the Trilogy merger/runway failure; amounts are participant recollections, not Rock cash flows.
• ★★ | Swiss Engineers in Early Silicon Valley I Knew | https://archive.computerhistory.org/resources/access/text/2018/06/102738807-05-01-acc.pdf | 2018-03-04 | first-person CHM memorandum by former Intersil/Eurosil executive Luc-Olivier Bauer, ref. X8682.2018; reliability A for Bauer’s experience and B for reconstructed financing | Printed p. 4 says Hoerni put in $300k and describes $4.5m from Rock, Fred Adler, and corporate investors; it must not be restated as a $4.5m Rock check. Printed pp. 7–10 supply operating/board and Eurosil distress context.
• ★ | Arthur Rock — Importance of Luck | https://computerhistory.org/blog/is-it-better-to-be-lucky-than-good-in-silicon-valley/ | remarks 2016-06-02; recap 2016-07-13 | official CHM recap and embedded official video clip by Arthur Rock; reliability A for the attributed excerpt and B for editorial framing | Short gala remarks on early luck and the Fairchild counterfactual; CHM's embedded official video is https://www.youtube.com/watch?v=79M9FqWF_fM. Treat as a partial direct-voice lead, not a complete appearance.
• ★ | Believe in the Value of Investors | https://stvp.stanford.edu/av/believe-value-investors | 2009-01-13 | official Stanford eCorner 1:07 Hugh Martin clip with transcript; reliability A for Martin's participant account and C for company inference | Martin, not Rock, says Art Rock and Bill Hambrecht joined the founders' “gold-plated” board and were initially treated as obstacles rather than resources. The clip does not name the company; chronology and separate Ridge evidence suggest Ridge Computers, but do not infer an investment. Witness/control only, not a Rock TALK appearance.

Authored, coauthored, edited-first-person, and collective writings

• ★★★ | Strategy vs. Tactics from a Venture Capitalist | https://hbr.org/1987/11/strategy-vs-tactics-from-a-venture-capitalist | 1987-11/12 | Arthur Rock-authored HBR article, printed pp. 63–67; reliability A for Rock's stated method | Clearest dated method statement and denominator lead: about 300 plausible plans per year, one or two investments, and roughly half failing expectations; also names Diasonics and Daisy, admits luck, and discusses accounting and management controls.
• ★★★ | Arthur Rock chapter in Done Deals | https://openlibrary.org/books/OL8089991M/Done_Deals | 2000 | edited first-person book chapter, printed pp. 139–148; reliability B pending licensed body retrieval | Longer Rock-selected narrative about firm identity, investing method, and portfolio; controlled borrowing/catalog route identified, but no unauthorized mirror may stand in for the chapter.
• ★★ | Health Reinsurance Bill Supported by New York Young Republican Club | https://archive.org/details/sim_congressional-record-proceedings-and-debates_january-6-1955-march-25-1955_101_appendix | letter dated 1955-03-04; entered 1955-03-21 | Congressional Record reproduction of a sole-signed Arthur Rock letter, Appendix A1929; reliability A for the text and recipient's identification | Earliest located Rock writing: as club secretary he transmitted a resolution supporting private-carrier health reinsurance with a public start-up fund; useful chronology and political-economic evidence, not a venture rule.
• ★★ | Let's address UC's problems, but keep them in perspective | https://www.sfgate.com/opinion/openforum/article/Let-s-address-UC-s-problems-but-keep-them-in-2497325.php | 2006-05-15 | full SFGate op-ed by Gordon Moore, Arthur Rock, and William K. Coblentz; reliability B for co-attributed prose | Defends competitive UC leadership compensation while acknowledging transparency and accountability failures; named coauthorship does not reveal sentence-level drafting.
• ★★ | Brighten the light on UC? Don't stifle the university's ability to recruit top talent | https://www.sfgate.com/opinion/openforum/article/Brighten-the-light-on-UC-Don-t-stifle-the-2513738.php | 2006-08-16 | full SFGate op-ed by Warren Hellman, Arthur Rock, and Gordon Moore; reliability B for co-attributed prose | Opposes public-session mandates for senior UC hiring and compensation deliberations on privacy and talent-competition grounds; additional signers are separately listed.
• ★ | Letter supporting McCain-Feingold campaign-finance reform | https://www.govinfo.gov/content/pkg/CREC-1999-10-27/pdf/CREC-1999-10-27-pt1-PgS13222.pdf | letter dated 1999-10-18; entered 1999-10-27 | Congressional Record reproduction of a collective letter, printed S13233–S13234; reliability A for the text and signature list | Rock appears as one of many signatories supporting a soft-money ban and disclosure; no evidence identifies him as the drafter, so this is civic-position evidence rather than Rock-authored prose.
• ★ | Mr. President, Urge Republicans to End Davis Recall Bid | https://www.latimes.com/archives/la-xpm-2003-jun-29-oe-doerr29-story.html | 2003-06-29 | full Los Angeles Times op-ed co-signed by Eli Broad, Arthur Rock, John Doerr, F. Warren Hellman, and Art Pulaski; reliability B | Opposes the California recall on cost, governance, bond-rating, and recovery grounds; co-attribution does not establish which signer drafted any sentence.
• ★ | SIJ EAD Letter to Biden administration | https://www.uscis.gov/sites/default/files/document/foia/Work_permits_for_Special_Immigrant_Juvenile_petitioners-Schey.pdf | 2022-03-08 | USCIS FOIA-hosted 14-page collective letter and signature list; reliability A for official custody and C for any claim of Rock authorship | Official indexing places Rock at printed p. 14 as Arthur Rock & Co. partner; the PDF returned HTTP 403 here, so this is a partial signatory route, not body evidence attributable to Rock.

Transaction, company, filing, and legal records

• ★★★ | Intel Corp. $2,500,000 Convertible Debentures memorandum | https://law.stanford.edu/wp-content/uploads/2023/01/Intel-Convertible-Note-220920.pdf | 1968 | contemporaneous four-page financing memorandum preserved by Stanford Law; reliability A for the document and B for later participant authorship testimony | Best ex-ante Intel receipt for founders, business, terms, proceeds, and conflicts; Rock and Moore later attributed Intel's plan to Rock, but the surviving scan is unsigned, refers to Rock in the third person, and differs from their three-page recollection.
• ★★★ | Apple Computer registration statement and prospectus | https://www.sec.gov/files/18-02062-FOIA.pdf | 1980-12 | SEC FOIA-hosted statutory filing scan; reliability A | FOIA PDF p. 33 / prospectus printed p. 30 records Rock's October 1980 appointment; PDF pp. 36–38 / printed pp. 33–35 record 640,000 shares, a rounded $0.09 table price, and 1.3% ownership; Item 25 at PDF p. 57 gives the placement-group price as $0.09375. A reputable mirror is not counted separately.
• ★★★ | Intel 35th-anniversary corporate history | https://www.chiphistory.org/chc_upload/content/pdf/7/1525800843/1525800843.pdf | 2003 | Intel corporate retrospective booklet; reliability B | PDF p. 4 reports Noyce $245,000, Moore $245,000, Rock $10,000, and a further $2.5 million raised by Rock through convertible debentures; precise but not independent of Intel's institutional narrative.
• ★★★ | In re Apple Computer Securities Litigation | https://law.justia.com/cases/federal/district-courts/FSupp/690/872/2359401/ | 1987-12-11 | federal district-court summary-judgment opinion, 690 F. Supp. 872; reliability A | Adjudicated control on the distinct 1984-filed Apple case naming Rock: the court granted summary judgment for defendants; this outcome must never be applied to the separate complaints reported in 1993.
• ★★★ | In re Diasonics Securities Litigation | https://law.justia.com/cases/federal/district-courts/FSupp/599/447/1898592/ | 1984-09-21 | federal district-court procedural opinion, 599 F. Supp. 447; reliability A | Records the 1983 IPO, later loss disclosures, price collapse, parties, and plaintiffs' allegations; class-certification and partial-dismissal rulings are not a merits finding against Rock.
• ★★ | Apple Computer confidential offering memorandum | https://archive.computerhistory.org/resources/text/2009/102712693.05.01acc.pdf | 1978 | primary private-placement memorandum in CHM custody; reliability A | Ex-ante Apple product, management, capital, risk, and financial record near Rock's investment; use exact sections and avoid double-counting a later CHM compilation.
• ★★ | Intel historical annual reports and virtual vault | https://www.intel.com/content/www/us/en/history/virtual-vault/articles/historical-annual-report-list.html | reports from 1968 onward | official corporate document index; reliability A for hosted historical filings | Route to financing, governance, ownership, operating results, and outcome bases; current URL paths do not change historical document dates.
• ★★ | Intel 2003 definitive proxy | https://www.sec.gov/Archives/edgar/data/50863/000104746903011595/a2105765zdef14a.htm | 2003-04-04 | SEC-filed DEF 14A; reliability A | “Directors Emeriti” records Rock's 1968–1999 director service, 1999–2003 emeritus status, and chair/lead/audit roles; a board title is not proof of each investment or operating decision.
• ★★★ | Arthur Rock & Co. principal and board biography | https://www.sec.gov/Archives/edgar/data/1107112/000091205701006073/a2023305zs-1.htm | 2001-02-15 | issuer-filed SEC registration statement; reliability A | Director biography says Rock had been principal of Arthur Rock & Co. since 1969 and an Intel director from 1968 to 1999; it does not reconcile the separate Arthur Rock & Associates label or establish fund economics.
• ★★★ | Echelon Corporation initial public offering prospectus | https://www.sec.gov/Archives/edgar/data/31347/000101287098001903/0001012870-98-001903.txt | 1998-07-30 | SEC-filed 424B1; reliability A | Printed pp. 42, 49, and 51–52 show Rock director since Dec. 1988; May 1997 purchase of 121,029 Series E shares at $5 plus a 29,187-share $5 warrant; and 717,716 beneficial shares including the warrant, 2.6% pre/2.2% post, with separate trust disclaimers and 7,500 board-service repurchase shares.
• ★★★ | Nasdaq 2001 Form 10-K/A | https://www.sec.gov/Archives/edgar/data/1120193/000091205702017541/a2078439z10-ka.htm | 2002-04-30 | issuer-filed SEC annual-report amendment; reliability A | Lines 282, 1168–1172, 1351–1352, 1430, and 1442 show July 2000 board election, 200,000 shares held by the Rock 2000 Trust with Rock sole trustee/beneficiary, indirect H&F LP status, and an aggregate Nov. 2001 private sale to five directors and H&F; no filing allocation supports a Rock-specific check.
• ★★★ | AirTouch Communications 1995 definitive proxy | https://www.sec.gov/Archives/edgar/data/904255/000095014995000156/0000950149-95-000156.txt | 1995-03-29 | issuer-filed SEC DEF 14A; reliability A | Printed pp. 3 and 5–6 show Rock director since Jan. 1994 and, as of Feb. 1, 1995, 152,572 beneficial shares including 2,572 held by spouse plus 10,000 exercisable options; the filing does not disclose original purchase, cost, or whether the position arose through the Pacific Telesis separation.
• ★★★ | IBM current report on completion of the Rational Software acquisition | https://www.sec.gov/Archives/edgar/data/51143/000104746903006346/a2104164z8-k.htm | 2003-02-21 | issuer-filed SEC Form 8-K; reliability A | Item 5 and Exhibit 99 report completion of IBM's acquisition of Rational Software; this is company-level exit evidence and does not prove that Rock held Rational securities or received proceeds.
• ★★★ | Echelon current report on the Adesto merger agreement | https://www.sec.gov/Archives/edgar/data/31347/000119312518208447/d809508d8k.htm | 2018-06-29 | issuer-filed SEC Form 8-K; reliability A | Item 1.01 and Exhibit 2.1 document the Adesto merger agreement; this company-level transaction does not establish that Rock retained his 1998 Echelon position or realized proceeds.
• ★★★ | Cell Genesys Calydon resale-prospectus supplement | https://www.sec.gov/Archives/edgar/data/865231/000086523101500053/body424.htm | 2001-11-29 | issuer-filed SEC Form 424B3 supplement; reliability A | Lines 3–18 and 53 update Rock's shares eligible for sale to 13,220; this later date must remain separate from the Oct. 15 registration statement and does not prove a sale.
• ★★ | Cell Genesys registration statement for Calydon acquisition shares | https://www.sec.gov/Archives/edgar/data/865231/000086523101500044/s3body.htm | 2001-10-15 | issuer-filed SEC Form S-3; reliability A | Lines 87–90, 356–365, 410–416, and 962–963 show Cell Genesys shares issued for Calydon stock, Rock with 14,689 shares and 11,751 then eligible for sale, and Sept. 4, 2001 acquisition effective time. This proves pre-acquisition ownership, not original check/security/date or realized proceeds.
• ★★ | Computer Systems Engineering, Inc. v. Qantel Corporation | https://law.justia.com/cases/federal/appellate-courts/F2/740/59/233752/ | 1984-09-11 | published First Circuit opinion, 740 F.2d 59; reliability A | Lines 57–76 record Qantel’s 1976–78 distributorship conduct, affirmed breach/fraud/ch. 93A findings, and $4,997,544.50 judgment. The corporate-conduct control overlaps Rock’s reported backing but does not find Rock personally responsible.
• ★★ | In re Daisy Systems Corporation and Daisy/Cadnetix, Inc. | https://law.justia.com/cases/federal/appellate-courts/F3/97/1171/554835/ | 1996-09-24 | published Ninth Circuit opinion, 97 F.3d 1171; reliability A | Lines 62–90 record the failed Cadnetix acquisition/financing, customer deferrals, refinancing failure, and involuntary Chapter 11; Rock is not named, so use only as an adverse company control for his 1987 Daisy example, not investment proof.

Contemporaneous and later press

• ★★★ | Rock resigns from Apple Computer board | https://www.upi.com/Archives/1993/08/04/Rock-resigns-from-Apple-Computer-board/8072744436800/ | 1993-08-04 | contemporaneous wire report; reliability B | Reports resignation to avoid perceived conflict and identifies complaints alleging insider trading and misleading statements; this is not a pleading, finding, or outcome, and its 1981 board-start date conflicts with Apple's filing.
• ★★★ | Arthur Rock: The Best Long-Ball Hitter Around | https://content.time.com/time/subscriber/article/0,33009,949965,00.html | 1984-01-23 | Michael Moritz contemporary profile in TIME; reliability B | Period account of amounts, portfolio cadence, operating behavior, and Diasonics before later legend hardened; figures and wealth estimates are reported, not audited.
• ★★ | Arthur Rock | https://www.forbes.com/asap/1998/0601/106.html | 1998-06-01 | retrospective business profile; reliability B | Preserves a “thirty or so” Fairchild-rejection version and a dated late-1990s public narrative; it cannot resolve the 20–30/35/45 conflict.
• ★★ | The Folks Who Brought You Apple | https://fortune.com/article/folks-who-brought-you-apple/ | 1981 | contemporaneous business press; reliability B | Reports Apple investor-network and valuation claims, including a $57,600 Rock investment; the prospectus remains the primary date/share/price control and headline value is not realized return.
• ★★ | Et tu Brute? | https://www.forbes.com/1997/08/04/column_side1.html | 1997-08-04 | retrospective magazine account; reliability B | Governance counterstory about the Jobs/Sculley board conflict; compare with Rock, Markkula, company, and board records and avoid the shorthand that Rock alone “fired Jobs.”
• ★★ | New Lease on Life | https://www.bitsavers.org/magazines/Datamation/19700915.pdf | 1970-09-15 | contemporaneous Datamation news item, printed p. 87; reliability B | Reports a $1.5m Xynetics refinancing from multiple sources including Transamerica and new board chair Rock; the aggregate round is not a Rock check. Fialkov’s existing participant oral history separately calls Rock an investor and board member.
• ★ | How Apple became Apple: The definitive oral history of the company's earliest days | https://www.fastcompany.com/91514404/apple-founding-50th-anniversary-apple-1-apple-ii-jobs-wozniak | 2026-03-26 | full Fast Company edited oral history by Harry McCracken; reliability B for edited attribution | Section 5 attributes one five-word Apple-investment assessment to Rock; provenance as a new interview versus archival reuse is unstated, so this is a late quote lead, not an independent full Rock appearance.

Books, scholarship, base rates, and source trails

• ★★ | The Power Law | https://books.google.com/books/about/The_Power_Law.html?id=FqVPEAAAQBAJ | 2022 | Sebastian Mallaby researched narrative history, Rock chapter printed pp. 17–39; reliability B | Reports D&R $3.4 million to nearly $77 million and reported: 22.6x (The Power Law, Rock chapter, printed pp. 17–39, 2022), with concentration in SDS; use a licensed copy and follow the endnotes because this is not a fund audit.
• ★★ | VC: An American History | https://books.google.com/books/about/VC.html?id=uSmWDwAAQBAJ | 2019 | Tom Nicholas scholarly history; reliability B | Historical framing for D&R, SDS, Intel, Doriot, family capital, SBICs, and partnership structures; use underlying records for specific economics and role attribution.
• ★★ | How Venture Capital Became a Component of the U.S. National System of Innovation | https://kenney.faculty.ucdavis.edu/wp-content/uploads/sites/332/2018/03/how-venture-capital-became-a-component-of.pdf | scholarly article accessed 2026-08-19 | peer-reviewed institutional history; reliability B | Counterfactual/base-rate context for SBIC failures, independent partnerships, ERISA, institutions, and scale; useful against attributing an entire system to one investor.
• ★★ | Paths and Regions: The Creation and Growth of Silicon Valley | https://kenney.faculty.ucdavis.edu/wp-content/uploads/sites/332/2018/03/Paths-and-Regions-The-Creation-and-Growth-of-Silicon-Valley.pdf | scholarly paper accessed 2026-08-19 | academic regional history by Martin Kenney and Urs von Burg; reliability B | Reports $3.5 million raised for Davis & Rock, preserving a third capitalization version; trace its reference rather than forcing agreement with HBS or Mallaby.
• ★★ | Private Equity Performance: Returns, Persistence and Capital | https://www.nber.org/papers/w9807 | 2003/2005 | NBER working paper and later Journal of Finance article; reliability A for its defined sample | Quantitative selection, heterogeneity, persistence, and follow-on-fund base-rate control; later vintages and Venture Economics data are not a direct D&R estimate.
• ★★ | Short-Term America Revisited? Boom and Bust in the Venture Capital Industry | https://www.nber.org/system/files/chapters/c10791/c10791.pdf | 2002 | NBER academic chapter; reliability A for the stated data and methods | Fundraising/return cycles and 1979 prudent-man context test Rock's excess-capital critique; historical VC datasets remain incomplete and often self-reported.
• ★★ | The Economics of the Private Equity Market | https://www.federalreserve.gov/pubs/staffstudies/168/ss168.pdf | 1995 | Federal Reserve Staff Study 168; reliability A | Official institutional history of ARD, SBICs, partnership incentives, information problems, fundraising, and returns; a system comparator rather than Rock-specific proof.
• ★ | Robert Noyce and Fairchild Semiconductor, 1957–1968 | https://www.cambridge.org/core/journals/business-history-review/article/abs/robert-noyce-and-fairchild-semiconductor-19571968/D4E663ED75099AA9A3F260BE7D6C13EE | scholarly article accessed 2026-08-19 | Leslie R. Berlin peer-reviewed business history, partly paywalled; reliability B | Archive-rich reconstruction of Fairchild roles and financing; obtain licensed body and follow primary notes before using exact claims.
• ★ | The evolution of new technology ventures over 20 years | https://www.sciencedirect.com/science/article/abs/pii/088390269290003A | 1992 | peer-reviewed longitudinal study of 250 Northern California technology companies, abstract-only open route; reliability B | Period-matched survival/failure denominator for market, design, distribution, capital, VC relations, and management; not a Rock portfolio study.
• ★ | A Dozen Things I've Learned From Arthur Rock | https://25iq.com/2015/02/13/a-dozen-things-ive-learned-from-arthur-rock-about-business-venture-capital/ | 2015-02-13 | Tren Griffin secondary quotation trail; reliability C | Mandatory discovery route to HBR/HBS/CHM/Berkeley passages; repeats disputed shorthand and must never replace the underlying source.

Archive and catalog routes

• ★ | Arthur Levitt papers finding aid | https://findingaids.library.columbia.edu/pdf/cul-4079912.pdf | holdings include 1996 and 1998 | Columbia RBML finding aid; reliability A for metadata only | Printed p. 167 lists incoming and outgoing Rock/Intel/Arthur Rock & Co. correspondence; contents were not inspected and are an archive-request lead, not evidence of subject matter.

Additional primary receipts and historical context

• ★★ | Semiconductor Sector corporate histories report | https://archive.computerhistory.org/resources/access/text/2012/10/102746515-05-01-acc.pdf | 2012 | CHM commissioned historical reconstruction; reliability B | PDF p. 9, “Financing agreement details (September 23, 1957),” gives the 1,325-share allocation, $1.38 million loan, options, and voting trust; it cites the underlying contract but is not itself contemporaneous.
• ★★★ | Scientific Data Systems registration and offering notice | https://www.sec.gov/news/digest/1964/dig051264.pdf | 1964-05-12 | SEC News Digest, printed p. 1; reliability A | Contemporaneous control for SDS's public/secondary offering, Hayden Stone underwriting group, and Rock's board-chair title; it does not establish personal cash flow or sole deal credit.
• ★★ | Intel 1977 annual report | https://www.intel.com/content/dam/www/central-libraries/us/en/documents/2025-05/history-1977-annual-report.pdf | 1977 | official historical annual-report scan; reliability A | Board page identifies Rock as vice chair and general partner of Arthur Rock & Associates; use the historical document date, not its modern hosting path.
• ★★★ | Letter to SEC chairman Arthur Levitt from Arthur Rock | https://sechistorical.org/digital-archive/letter-to-sec-chairman-arthur-levitt-from-arthur-rock-in-support-of-separating-auditing-and-consulting-practices/ | 2000-06-28 | SEC Historical Society catalog and letter; reliability A for Rock's statement | Direct governance writing supporting audit/consulting separation and citing Rock's audit-committee experience at Intel, Apple, and Teledyne; company filings remain role controls.
• ★★★ | Arthur Rock letter on leaders | https://letters.stevejobsarchive.com/volume-2/arthur-rock | 2025-07-03 | Steve Jobs Archive publication of Rock's short letter; reliability A for the dated statement | Direct later voice on choosing, listening to, and evaluating leaders; doctrine is retrospective and does not prove historical application.
• ★★ | Small Business Investment Act of 1958 | https://www.govinfo.gov/content/pkg/COMPS-1835/pdf/COMPS-1835.pdf | enacted 1958; current compilation accessed 2026-08-19 | official statutory compilation; reliability A | Establishes the SBIC policy regime around D&R's formation; it is context, not evidence that Rock or a named vehicle used an SBIC.
• ★★ | Employee Retirement Income Security Act prudent-man clarification | https://www.govinfo.gov/content/pkg/FR-1979-06-26/pdf/FR-1979-06-26.pdf | 1979-06-26 | official Federal Register rule and preamble; reliability A | Contemporary regulatory context for later institutional venture fundraising; causal application to Rock requires dated vehicle and LP evidence.
• ★★ | U.S. Semiconductor Manufacturing: Industry Trends, Global Competition, Federal Policy | https://www.congress.gov/crs-product/R44544 | updated congressional report accessed 2026-08-19 | Congressional Research Service historical and policy synthesis; reliability A | Documents government-purchase and industrial-policy context for early semiconductors, preventing person-only commercialization stories.
• ★★ | Innovation in Information Technology | https://nap.nationalacademies.org/read/12174/chapter/5 | 1999 | National Academies historical synthesis; reliability A for cited research | Chapter reports the scale of military procurement and R&D in early semiconductor markets; later retrospective context, not a Fairchild deal record.
• ★★ | Companies — The Silicon Engine | https://www.computerhistory.org/siliconengine/companies/ | CHM exhibit accessed 2026-08-19 | official CHM institutional company history; reliability B | Lines 77–79 say Amelco was founded in 1961 by Hoerni, Kleiner, Last, and Roberts, with Teledyne financing arranged by Rock, and later became Teledyne Semiconductor/Telcom before Microchip acquired it; this establishes intermediation, not Rock economic exposure.
• ★★ | NVCA Model Legal Documents | https://nvca.org/model-legal-documents/ | current library accessed 2026-08-19 | official industry model-document library; reliability A for current forms and B for normative comparison | Current preferred-equity, voting, investor-rights and tranched-financing controls; the forms require tailoring and are diagnostics for missing historical fields, not evidence of Rock-era terms.
• ★★ | ILPA Principles 3.0 | https://ilpa.org/wp-content/uploads/2019/06/ILPA-Principles-3.0_2019.pdf | 2019 | official Institutional Limited Partners Association principles; reliability A for the published framework and B for normative comparison | Printed pp. 10–22 and glossary p. 41 provide current governance, economics, reporting, key-person, conflict and transparency controls; they do not reconstruct D&R or later Rock vehicles.
• ★ | Fairchild Semiconductor employee papers and oral-history collections | https://www.computerhistory.org/collections/search/?s=Fairchild%20Semiconductor | holdings accessed 2026-08-19 | CHM collection-search and item-level retrieval route; reliability A for catalog metadata | Route to founder, employee, financing, and technical records that could test the public origin narrative; each item requires its own locator and access check.
• ★ | Arthur Rock Center history | https://www.hbs.edu/entrepreneurship/about/history | current page accessed 2026-08-20 | HBS institutional history; reliability B | The “In 2003” entry documents the $25 million gift and namesake center; promotional philanthropy evidence does not validate investing performance.
• ★ | Arthur and Toni Rembe Rock Center for Corporate Governance | https://law.stanford.edu/arthur-and-toni-rembe-rock-center-for-corporate-governance/ | current page accessed 2026-08-19 | Stanford institutional page; reliability B | Establishes the namesake governance institution and a later influence route, not a deal, vehicle, or return record.

Coverage assessment

Mechanical coverage

The map contains 85 rows and 85 unique canonical URLs: 31 ★★★ / 41 ★★ / 13 ★. Exclusive section totals are 26 direct-voice or participant records; 8 authored, edited-first-person, coauthored, or collective written records; 18 transaction/company/filing/legal records; 7 press records; 10 scholarship/base-rate/source-trail records; 1 archive/catalog route; and 15 additional primary, historical, or institutional controls. The priority routing denominator is 72 = 21 TALKS / 8 WRITINGS / 43 DEALS-controls. Mirrors and related documents are mentioned inside annotations but are not counted as independent rows. Every checkbox remains empty because no later-phase processing is being claimed.

Identity and name/date sweep

Searches used Arthur Rock with venture capital, Rochester, Syracuse, HBS, Army, Hayden Stone, Davis & Rock, Arthur Rock & Associates, Arthur Rock & Co., Rock-Kramlich, company names, and archive/SEC/catalog variants. Results tied the target to Rochester, Syracuse, HBS, Hayden Stone, Davis & Rock, Intel, and Apple through HBS, Bancroft, CHM, and SEC records. Searches also surfaced unrelated Arthur Rock obituaries and a namesake commercial site; neither is used. Strong inspected archival records support birth year 1926, but this SCOUT publishes no day/month, current-age, or death claim without target-specific authoritative evidence.

Direct voice and primary-archive exhaustion

The public direct-voice spine is Bancroft 2008–09, HBS 2001, Stanford Silicon Genesis 2002, CHM Pioneers 2002, CHM 2007, the two-session Something Ventured Rock family, CHM Moore–Rock 2014, Stanford Law 2023, Rock's 1987 HBR essay, his sole-signed 1955, 2000, and 2025 letters, the two 2006 named-coauthor op-eds, and participant controls from Last, Markkula, Moore, Dennis, Kramlich, Fialkov, PBS, SDS, and the Fairchild panels. The 1999, 2003, and 2022 collective writings establish only that Rock was a signatory or named co-author; they do not establish that he drafted the prose. These are overlapping retrospective or attribution families, not independent confirmation merely because several institutions host them. Of 21 priority TALKS families, 19 have full public bodies and 2 remain controlled barriers: the Science History Institute Rock oral history requires a name/email request because of interviewee-stipulated privacy, and Stanford Fairchild V0311 is on-site and reproduction-controlled. Eighteen public families expose complete human-edited text layers; Pioneers of Venture Capital supplies the full hosted video and host-generated timed text, which requires speaker and wording verification. The Something Ventured transcripts, Silicon Genesis transcript, and Stanford Law interview are fully retrievable through the disclosed routes above. No local ASR was used; any future transcription of genuinely missing media must use the OpenAI API.

Searches of Stanford Special Collections/OAC, CHM, SHI, Smithsonian, HBS/Baker/HOLLIS, Columbia finding aids, Internet Archive/Open Library, SEC EDGAR, company archives, and exact title/name variants found no open Arthur Rock personal-papers finding aid, D&R investor-report series, capital-call/distribution ledger, complete investment-memo/pass corpus, or audited person-level return record. The Levitt finding aid and Stanford/Smithsonian item records are exact request routes, not inspected body evidence.

Chronology and role conflicts that remain open

• Fairchild prospect count: Jay Last says roughly 20–30, other accounts say about 35, and PBS preserves Rock saying 45. The crossed-out 1957 list is an artifact lead, not yet an enumerated ledger.
• Fairchild capital: the CHM reconstruction reports 1,325 shares, a $1.38 million loan from Fairchild Camera, options, and a voting trust; oral histories often round the advance to $1.5 million or mention a $3 million purchase option. Drawn capital, authorized capacity, and option value must not be averaged together.
• Davis & Rock: public narratives use $3.0 million, $3.4 million, or $3.5 million for capitalization and roughly $77 million or $100 million for outcome. No audited cash-flow basis surfaced, so no new multiple or Rock-personal return is computed.
• Intel: the primary financing memorandum and Intel history support a $2.5 million convertible-debenture syndicate plus Noyce $245,000, Moore $245,000, and Rock $10,000; later $2 million shorthand remains a memory variant, not an equal primary estimate.
• Apple: the prospectus proves Rock bought 640,000 split-adjusted shares in January 1978 and was appointed a director in October 1980. The director table displays $0.09 per share while Item 25 gives the placement-group price as $0.09375; SHI and UPI use 1980/1981 board-start conventions. Board minutes/proxies are needed to resolve appointment, election, and reporting dates.
• Litigation: the 1984-filed Apple matter ended in 1987 summary judgment for defendants. UPI's 1993 report describes different complaints and allegations with no located docket/disposition. The Diasonics opinion is procedural and records allegations, not a merits finding.

People, entities, vehicles, and deal-credit boundary

The map keeps Rock personally separate from Hayden Stone, Bud Coyle, Davis & Rock and Thomas J. Davis, the Rock–Kramlich period, Arthur Rock & Associates, Arthur Rock & Co., personal Apple shares, financing syndicates, and portfolio-company boards. The last two Rock-branded names are not merged without formation/governance records. Fairchild belongs to Kleiner, the eight founders, Coyle/Rock/Hayden Stone, Sherman Fairchild/Fairchild Camera, and employees in distinct roles. Intel belongs operationally to Noyce, Moore, Grove, Vadasz, and employees; Rock's supported roles are capital organization, memo, chair, and governance. Apple requires separate treatment of Jobs, Wozniak, Markkula, Valentine/Sequoia, Venrock/Crisp, Singleton, Rock, and later board actors. “Introduced,” “placed financing,” “invested,” “approved,” “served on the board,” and “later narrated” are not interchangeable.

Candidate portfolio and denominator

The locked public research denominator is 26 company/opportunity families, routed once into five exclusive categories: 16 confirmed Rock/Rock-vehicle capital or ownership cases (Teledyne, Scientific Data Systems, Intel, Apple, Diasonics, Qantel, ELXSi, Intersil, AirTouch, Echelon, Nasdaq, Benrus, Ridge Computers, Xynetics, Astrodata, and Calydon); 4 financing, launch, or intermediation cases without proved Rock economic exposure (General Transistor, Fairchild Semiconductor, Rational Machines, and Amelco); 2 board, acquirer, or spinoff-only controls (Xerox and Argonaut Insurance); 2 explicit passes (ROLM and Compaq); and 2 failure or association controls without proved Rock investment (Electronic Memories and Daisy Systems). [researcher inference · strong] All 26/26 were searched and mapped; 0/26 has a complete source-backed chain of vehicle, decision date, security, check, ownership, follow-ons, exit/write-off, and cash flows, so 26/26 remain unresolved at full reconstruction depth. The 16 capital/ownership cases are not a return denominator: several prove a position or financing relationship without an original check, basis, or realization. AirTouch proxy, printed pp. 3 and 5–6 · Echelon prospectus, printed pp. 42, 49, 51–52 · Nasdaq 10-K/A, cited lines

The Rock–Kramlich recollection of $10 million committed, $6.25 million invested across 13 companies, and a recalled $40 million result remains a separate completeness gap because the transcript does not enumerate the 13. It cannot create 13 inferred company rows or a normalized return. Anadex Instruments and General Capacitor remain provisional searches outside the denominator; Ridge is confirmed and did not go bankrupt. Artisan Partners Asset Management is excluded as later public ownership without a surfaced venture role, Embarcadero Technologies as advisory-only, and MiniScribe as a false-positive search hit. [researcher inference · strong] Kramlich oral history, printed pp. 5–6 · Ridge participant interview, printed pp. 13, 15, 20

Rock's 1987 HBR essay separately reports a funnel of roughly 300 plausible plans per year, one or two investments, and about half of selected investments failing expectations. That is investor-stated context, not an archival portfolio count. A true total-company denominator and any economic contribution/Pareto analysis are not computable in SCOUT.

Historical, regulatory, and skeptical coverage

The 1958 SBIC statute, 1979 prudent-man clarification, Federal Reserve institutional history, CRS semiconductor history, and National Academies chapter cover alternative vehicles, institutional capital, government demand, R&D, and market conditions available around different eras. They prevent military procurement, semiconductor economics, exit markets, and regulatory change from being silently attributed to Rock.

Adversarial searches combined Arthur Rock with criticism, lawsuit, fraud, sanction, SEC, conflict, Apple board resignation, insider trading, Diasonics, failed investment, biggest loser, Daisy, Qantel, ELXSi, ROLM, Compaq, pass, write-off, ordinary investment, and namesake/death variants. Results found Rock's Diasonics admission, the procedural Diasonics case, the adjudicated 1984 Apple case, the unresolved 1993 allegations, governance criticism, and his own criticism of excess capital/corporate VC. No public sanction or merits finding of Rock misconduct surfaced; that is a bounded search result, not proof none exists. Selection, survivor, access, era, founder/team, and one-outlier explanations remain live.

Completed peers for later comparison

• Reid Dennis is the closest same-region, overlapping-era, syndication, and long-governance comparator; Dennis's raw witness transcript also supplies a non-Rock account of Intel board work.
• Laurance Rockefeller is the Apple co-investor/family-capital contrast; Markkula's CHM oral history preserves the distinct Valentine, Venrock, and Rock sequence.
• Georges Doriot is the public-corporation/permanent-capital and defense/research contrast; Federal Reserve Staff Study 168 supplies underlying structural context.

These are completed, evidence-based comparators, not rankings or claims that their vehicles were equivalent.

Highest-value retrievals for later phases

1. Retrieve the SHI transcript/audio through its stipulated institutional request and inspect Stanford V0311 on-site or by authorized reproduction; independently verify CHM Pioneers machine-text wording and speaker turns against the hosted video.
2. Obtain the direct 1957 Fairchild contract/prospect list and Stanford V0311 panel; test the 20–30/35/45 counts, $1.38/$1.5 million language, and Coyle/Rock/Fairchild roles.
3. Locate D&R formation documents, LP letters, capital accounts, complete portfolio, distributions, and Mallaby/Nicholas underlying notes to reconcile $3.0/$3.4/$3.5 million and $77/$100 million.
4. Retrieve Intel's subscription ledger, later financing records, and board minutes; distinguish founder equity, Rock's $10,000, syndicated debentures, and governance.
5. Retrieve Apple's 1978 purchaser agreements, 1980 appointment minutes, 1980–1993 proxies/minutes, and the 1993 complaint dockets/dispositions; preserve the separate 1984 litigation.
6. Complete the locked 26-family ledger: retrieve a full vehicle, decision-date, security, check, ownership, follow-on, exit/write-off, and cash-flow chain for each family, while keeping the 16 capital/ownership cases separate from intermediation, board/acquirer controls, passes, and adverse association controls.
7. Resolve the unnamed Rock–Kramlich 13-company pool and the provisional Anadex Instruments and General Capacitor leads from governing agreements, investment schedules, quarterly reports, and distribution records; do not create company rows from an unenumerated aggregate recollection.
8. Inspect Rock–Levitt correspondence, Stanford/CHM Fairchild collections, HBS/Baker records, licensed Done Deals, and licensed book endnotes.

Source notes