Paul Wythes

Snapshot and testable edge thesis

The thesis is about a joint operating system, not audited personal alpha. The public record supports institution building more strongly than it supports a Wythes-personal return claim.

Core philosophy

Wythes's strongest direct themes are practical and bounded: make founder contact before the market becomes crowded; let reciprocal interest precede intensive work; respect the investor's expertise limit while checking the people who possess the technical knowledge; recruit or replace management when the gap is remediable; syndicate when capital or expertise is constrained; and build companies over a long clock. [investor-stated · moderate] These are paraphrases of one late oral-history work plus one readable PAW direct-voice source, not a contemporaneous authored doctrine. spoken synthesis · written synthesis

The strict attributable-writing denominator is N=3 and 0/3 full bodies: Done Deals remains controlled unread, while the Mercury and NYSE artifacts are unlocated. PAW is readable journalistic direct voice, not a Wythes-authored work. [researcher inference · strong] writings coverage

The reconstructed operating system is: source locally → verify people/customers → distinguish fatal from repairable gaps → assign capital, ownership, and decision rights → recruit/govern with conflict controls → reserve against milestones → keep person, team, vehicle, and outcome credit separate. [researcher inference · moderate] It is not a claim that every Sutter Hill decision followed this sequence; no original memo or complete ordinary-case ledger demonstrates that. model inventory

Three boundaries are essential. First, a good company outcome does not reveal original investment economics. Second, a Sutter Hill association does not establish Wythes sourcing, approval, board work, or personal capital. Third, patience is capacity, not quality: without milestones and stop rules it can become escalation. [researcher inference · strong] evidence boundary

Sourcing

• Cheap founder contact before expensive diligence — [investor-stated · moderate]. Wythes recalled physically canvassing the Valley and asking companies for meetings. The procedure exploits low contact cost and local density; no funnel proves efficacy. Wythes oral history, printed pp. 39–48 / PDF pp. 43–52
• Reciprocity before depth — [investor-stated · moderate]. His reconstructed early pitch offered a long partnership rather than a financing transaction and waited for mutual interest. That can reveal willingness to work together, but it may also select for network comfort rather than quality. oral-history note
• Relationship reuse and contrast — [retrospective witness · moderate]. Kasper is the documented Wythes canvassing application; Measurex is a founder-originated/referral contrast whose founder independently credits later syndication help; Diablo→Qume shows reused founder/executive paths. None supplies a source-to-return conversion rate. Measurex note · Comstock note
• Negative attribution controls — [retrospective witness · strong]. Apollo is routed to Draper; Linear's founders identify the initial syndicate and do not name Wythes. These prevent the firm label from becoming personal sourcing credit. Apollo note · Linear note

Record introducer, first contact, incremental information, independent channel, conflict, sponsor, and decision owner. Missing are the complete funnel, rejection ledger, referral concentration, sourcing cost, geographic radius, and source-quality/outcome comparison. [researcher inference · strong]

Picking

A no-hindsight decision record must state the invest/pass answer, evidence that could change it, security/ownership, vehicle, decision rights, milestones, reserves, dissent, and kill conditions. The public corpus usually lacks those fields. [researcher inference · strong]

Deal and portfolio mechanics

The final public union contains 35 associated companies, five passes, and one anonymous loss. The associated label spans inherited positions, sourcing, syndication, board roles, firm/group holdings, later-firm holdings, and narration; it does not mean funded by Wythes. Every associated case is materially incomplete, so 0/35 supports a complete vehicle-to-cash-flow chain. [researcher inference · strong] ledger

Measurex supplies the cleanest financing arithmetic: total startup financing about $1.25 million, comprising $250,000 from Sutter Hill/SBIC plus two outside blocks of about $500,000 each. The larger number is the round, not Sutter Hill's check. [retrospective witness · moderate] Measurex lab

The only portfolio-composition estimate is inseparable: Wythes attributed roughly $8 million of a reported ~$10 million earlier Sutter Hill/Genstar transaction value jointly to Kasper and Measurex. It is not two deal returns, cost, proceeds, MOIC, IRR, or a Wythes distribution. Removing it leaves no comparable deal-level measure. [investor-stated · provisional] Pareto analysis

No universal target ownership, security preference, reserve ratio, diversification rule, loss budget, follow-on formula, concentration threshold, distribution rule, or exit doctrine surfaced. Xidex's bridge and long arc show case behavior; Duplicon and Priam/Cohesive are escalation boundaries. IPOs and acquisitions are company outcomes until dated ownership and cash flows connect them to a vehicle. [researcher inference · strong] portfolio forensics

Wythes reported a practice of distributing unrestricted public stock and leaving sale timing to recipients; one retrospective practice without agreement, tax/restriction, distribution, and outcome records is a deal-specific ledger item, not a universal exit model. [investor-stated · provisional] deal-mechanics limits

Vehicle and incentive fit

Vehicle fit includes stage, geography, regulation, capital source, liquidity, duration, reserves, decision rights, incentives, and succession. The evidence supports duration as plausible capacity, not causal alpha. The transcript's later characterization of the 36% figure as carried interest conflicts with the 80/20 and 75/25 economics and cannot be normalized. [researcher inference · strong] return controls

Team attribution and work with founders

Wythes described partnership complementarity and shared credit. Multiple participant families support team framing, while the missing vote/carry ledger prevents a claim that complementarity improved returns. The operating model is still usable: name the capability gap, assign a responsible partner, record the decision right, and preserve founder/operator credit. [investor-stated · moderate] talks frameworks · team model

Views on the craft

• Company building over transaction completion — [investor-stated · moderate]. The oral/written derivative record favors a multi-year operating horizon. It does not establish universal hold duration or return. Darden analysis
• Access is not selection — [researcher inference · strong]. Dense networks and firm reputation can improve opportunity flow; matched decisions are required to show judgment. profile audit
• Governance is a mechanism, not a title — [researcher inference · strong]. Recruiting, information, capital, conflict, and stop rights matter; tenure alone proves none. bounded governance summaries
• Power-law humility — [researcher inference · strong]. Kasper+Measurex may dominate the only portfolio-composition estimate, but missing cash flows prevent an 80/20 result or repeatability claim. Pareto
• Returns belong to defined objects — [researcher inference · strong]. Company outcome, firm IRR, personal return, carry, and filing ownership are different. return controls
• Failure rhetoric — [investor-stated · provisional]. Acknowledging misses is useful against survivor bias, but no complete loss denominator or dated postmortem proves learning. failure ledger
• Mission — [investor-stated · provisional]. Useful products, jobs, and taxes are stated benefits of company building; no evidence shows mission changed selection, terms, or measured outcomes. Darden analysis
• Greatness — [researcher inference · moderate]. The strongest supported definition is building a sourcing/governance institution that survives succession while keeping attribution honest; superior personal alpha remains unproved.

Signature bets and decision process

Use the four bounded no-hindsight labs. Each preserves what was knowable, the role split, a provisional invest/pass, kill criteria, and the missing evidence:

1. Measurex: customer/operator quality versus an SBIC check constraint and local financing refusals; decide whether syndication solves capital without outsourcing judgment.
2. Diablo/Qume: repeat-founder access versus duties and confidential information; specify consent, recusal, board allocation, and walk-away conditions.
3. Apple/Cisco: price or management weakness versus staging, repricing, and recruiting; later success cannot rewrite the original evidence.
4. Duplicon: product promise versus validation, capital, and weak outcome; this is a failure/weak-outcome exercise, not proof of a Wythes hardware rule.

Mental models and maxims

The reconciled inventory contains eight procedures in decision sequence, not rank order: four moderate and four provisional. Generic failure acceptance, public-good rhetoric, a Duplicon hardware slogan, and the researcher-created price/fatal/remediable taxonomy are excluded.

1. Reciprocal local sourcing before expensive diligence — provisional
2. Expertise-boundary people diligence — provisional
3. Make management gaps governable, then recruit or replace — moderate
4. Syndicate when capital, expertise, or governance exceeds one vehicle — moderate
5. Use consent, recusal, and information separation for competitor conflicts — provisional
6. Build companies on a multi-year clock, with explicit stopping tests — moderate
7. Match vehicle duration and incentives to the work — provisional
8. Build complementary investor teams and keep role credit explicit — moderate

Evolution and contradictions

No public sequence proves that Wythes changed his process after MMI, Diablo/Qume, Duplicon, the mining loss, or the missed companies. The contrast between Diablo/Qume and Priam/Quantum suggests a possible conflict-control evolution, but chronology, minutes, and explicit causal statement are insufficient. [researcher inference · provisional] talks evolution

Linear is a negative attribution control rather than evidence of a Wythes selection evolution. The inaccessible Done Deals, roundtable, Mercury, and NYSE bodies cannot fill a chronology by title or later publication date. [researcher inference · strong] Linear note · writings coverage

Return claims remain separate: reported >35% average annual Sutter Hill return over a prior thirty-year period (Wythes/PAW); reported ~36% net IRR for 1970–2005 (Wythes oral history); reported ~40% before 1981 (Draper); and reported 12–13% net for a first India fund (different vehicle). The transcript's later characterization of the 36% figure as carried interest conflicts with the earlier 80/20 and 75/25 splits. None is audited, personal, mutually reconcilable, averageable, or compoundable. [investor-stated · provisional] return ledger

Failures and limits

• MMI — [investor-stated · moderate]: minority control and founder incentives allegedly weakened the investment; no original terms or proceeds. case
• Duplicon — [investor-stated · provisional]: a weak outcome/sale; no proven write-off or documented reliability rule. lab
• Unnamed mining position — [investor-stated · provisional]: write-off without identity, vehicle, date, amount, rationale, or learning record.
• Five passes — [investor-stated · provisional]: Apple, Cisco, Tandem, Siebel, and Webvan. Webvan was examined and passed as too expensive; it was not funded. failure ledger
• Governance/conflict controls — [contemporaneous record · strong]: Teltone, Activision, and T. Rowe matters require careful legal boundaries and do not establish personal wrongdoing. filing controls
• Priam and Cohesive — [investor-stated · provisional]: prolonged adversity/repair constrain the claim that patient intervention reliably worked; exact costs, stop decisions, and outcomes remain incomplete. failure ledger
• Missing ordinary cases — [researcher inference · strong]: selected winners, public companies, and adverse records conceal the full portfolio, recaps, write-offs, dilution, and quiet mediocrity.

No documented process change follows the failures. No complete portfolio, board, or pass denominator exists; LSI remains orphaned; three Biovest names remain outside 41; the authored corpus is 0/3 full; and no associated case has complete economics. These limits prevent a personal hit rate, alpha claim, or causal attribution. residual gaps

Historical operating environment

• 1958–1965 — [contemporaneous record · strong]. The federal SBIC framework expanded organized risk capital; Sutter Hill Capital's 1962 license and Wythes's reported $250,000 check constraint shaped feasible early financing. Small Business Investment Act · license record
• 1960s local cluster — [investor-stated · moderate]. Few local venture sources and dense technical-company formation made direct canvassing and East Coast syndication more valuable. No decision-date comparison set or acceptance rate survives. historical context
• 1970 vehicle change — [retrospective witness · moderate]. Genstar-backed continuing capital plausibly changed duration and fundraising burden. The 1975 SBIC surrender is a separate legal endpoint. vehicle chronology
• 1979 onward — [contemporaneous record · strong]. The prudent-person regulatory change belongs to later institutional-capital expansion, not the cause of the 1970 partnership. Wythes's NVCA participation is not sole authorship. Federal Register, 1979
• 1980s–1990s — [investor-stated · moderate]. More venture competition, changing storage/semiconductor cycles, institutional LPs, and public exit windows altered sourcing and realizations. The record provides no vintage/stage/sector base-rate series. portfolio limits

The favorable thesis is era-bounded: access and patient capital may have mattered more when organized venture supply and experienced technical managers were scarce. Any modern transfer needs current, independently researched base rates; T011 contains none.

Modern VC translation (as of 2026-08-21)

No current company, market, fund-term, or regulatory case was independently researched in T011. Every application below is therefore a [researcher inference · provisional], not a claim about a present firm or market.

The price/fatal/remediable classification remains only a [researcher inference · insufficient public record] and an analyst-created IC exercise. Apple, Webvan, and Cisco come from one retrospective work; later issuer outcomes do not independently prove the rule or an application. Use the Apple/Cisco lab to test the exercise, not as evidence of a ninth Wythes model. candidate-model boundary

Comparison with completed peers

The comparison is dimensional, not a return or greatness ranking. All four records lack complete personal cash-flow denominators. Every dimensional entry below is [researcher inference · moderate], grounded in the linked completed peer syntheses and the underlying evidence in the first row; none is a return claim.

Synthesis — [researcher inference · moderate]. Wythes resembles Brooke and Crisp most in institution/vehicle continuity and Rock in people/governance practice, but his supported edge is more locally reciprocal and explicitly team-based. Brooke's cross-border institution design, Rock's person-centered concentrated vehicles, and Crisp's Rockefeller platform are not templates for Sutter Hill. Shared Apple, Intel, and firm-brand histories require role separation, not cross-validation.

The strongest case against greatness

[researcher inference · strong]. The skeptical thesis is that Wythes was a capable steward and narrator inside an advantaged pre-existing platform, not a demonstrably superior personal investor. Peterson and Lodato supplied the company/SBIC; four D&J positions and Draper supplied assets and networks; Genstar and later LPs supplied capital; founders, syndicates, boards, and successor partners supplied company-building work. Bossen on Measurex and Comstock on Diablo are stronger person-specific controls than most famous associations. Apollo is Draper-led, Linear does not name Wythes, Applied Materials routes through Morgan, and Shutterfly postdates active retirement. The public set is selected, the claimed 27-plus-board roster is unrecovered, 0/35 is full-chain, group holdings are not personal ownership, and the visible economics may be dominated by the joint Kasper+Measurex sentinel. Reported returns are retrospective firm/vehicle estimates with unresolved objects and bases. Local scarcity, technical-cluster growth, SBIC constraints, evergreen capital, founders, successor teams, survivor memory, and exit cycles can explain outcomes without exceptional selection. profile skeptic

The favorable thesis wins only if original sourcing logs, memos/votes, references, cap tables, minutes, recruiting records, partnership terms/capital accounts, and matched ordinary cash flows show repeated Wythes-specific mechanisms and superior ex-outlier outcomes. The skeptic wins if other actors own the decisions, governance is nominal, the vehicle does not change behavior, or returns disappear after controlling access, team, vehicle, vintage, founders, and luck. With 0/35 complete chains and the decisive records absent, the present public evidence cannot discriminate between these explanations. [researcher inference · strong] retrieval ledger

Analyst study guide, glossary, and go deeper

Recommended path

1. Read the profile for chronology, people, and the competing causal theses.
2. Audit sources, then writings and talks, before treating any late recollection or unavailable body as doctrine.
3. Work through the 41-family ledger, vehicle map, ownership controls, and return limits.
4. Replay the four labs below without looking ahead; then compare your answer with the sourced outcome.
5. Use mental models as procedures with exception ledgers, not maxims to memorize.

Investor-specific glossary

Reusable checklists

• Sourcing: introducer; first contact; contact cost; incremental information; reciprocity; independent channel; conflict; sponsor; would the case survive without platform prestige?
• Diligence: technical claim/falsifier; independent expert; customer evidence; management gaps; market/price; legal/IP; alternatives; disconfirming reference; dated evidence.
• IC: invest/pass; fatal versus remediable issue; security/ownership; vehicle; syndicate roles; decision owner/dissent; reserves; milestones; kill/reversal criteria; conflict controls.
• Governance: information/vote rights; recruiting owner; board cadence; capital plan; competitor consent/recusal; escalation; stop rule; measured intervention result.
• Attribution/economics: source, sponsor, diligence, approval, negotiation, board, follow-on, exit; founder/operator; vehicle; cost/ownership/dilution; distributions; fees/carry; benchmark.

Four no-hindsight exercises

1. Measurex — capital constraint and syndication
2. Diablo/Qume — repeat network and competitor conflict
3. Apple/Cisco — price and remediable management
4. Duplicon — validation and weak outcome

Ranked sources and retrieval priorities

1. Wythes CHM/NVCA oral history: broadest direct spine for life, vehicle, cases, philosophy, passes, and failures; late, paraphrase-only, and one witness/work. source note
2. Contemporaneous filings and government/court records: best for legal entities, dates, holdings, boards, conflicts, and adverse boundaries; incomplete for rationale and cash flows. filing controls
3. Bossen and Comstock participant histories: strongest person-specific independent controls for Measurex and Diablo/Qume; retrospective and narrow. Bossen · Comstock
4. PAW biotech feature: best readable published Wythes direct-voice route for expertise humility and the >35% estimate; journalistic, not an authored method. source note
5. Draper/Johnson and founder panels: formation and negative-attribution controls; witness families must not be multiplied. talks guide
6. Darden briefing: compact derivative route to company-building duration; it reuses the 2006 work and adds no independent Wythes thought. source note

Highest-value retrievals are original sourcing logs, investment memos/votes, SBIC files, partnership agreements/amendments/capital accounts, cap tables, board/recruiting/conflict records, full investment/pass/write-off schedules, and dated distributions. Lawful access to Done Deals, the Mercury item, NYSE remarks, and 1995 roundtable could change doctrine coverage; until read, they remain access leads only.

Pareto 80/20 — what drove the record

[researcher inference · strong] A literal 80/20 result, career hit rate, or Wythes-personal return cannot be calculated. The final discovered union is 41 opportunity families, but it is a selected public-name set rather than a complete portfolio, and 0/35 associated companies has the full legal-vehicle → decision-right → security/cost → ownership/dilution → follow-on → board-action → exit/write-off → cash-flow chain. Company scale, acquisition value, filing-group ownership, and reported fund IRR are not interchangeable. denominator reconciliation · source-map return boundary

Economic contribution — directional, provisional, and not a personal-return ranking

Dominant-sentinel test — [researcher inference · strong]. Removing the joint Kasper/Measurex estimate leaves no comparable deal-level economic measure. The remaining order is therefore not computable; the evidence cannot support repeatability, a dominant outlier, or a claim that 20% of investments produced 80% of returns.

Historical importance — ranked independently of economics

Decision-learning value — ranked for replayability

Evidence boundary and searched denominator

Scope and inclusion rules

• One legal lineage, one row — [researcher inference · strong]. Micro Magnetics/MMI, Micrex/Xidex, Manufacturing Management/Pierce Pacific, and Amerigroup/AMERICAID are each deduplicated. Tellabs is not Teltone; Oacis is not Exogen; Core Therapeutics is not Corcept; subject Wythes is not Paul Wythes Jr. identity boundary
• Association is deliberately broad — [researcher inference · strong]. It may mean an acquired portfolio position, referral, syndication, board role, firm/group ownership, remembered evaluation, later-firm holding, or official-biography attribution. It does not mean funded, Wythes-led, or profitable. source-map candidate boundary
• Roles remain separate — [researcher inference · strong]. Sourcing, sponsorship, diligence, approval, negotiation, board work, recruiting, follow-ons, exit influence and later narration are not interchangeable. Unknown remains unknown. Wythes CHM printed pp. 149–151 / PDF pp. 153–155
• Vehicles and holders are not companies — [researcher inference · strong]. Genstar, Biovest, India/China funds, Sutter Hill entities, TOW Partners and personal/trust holdings remain outside the company count. A Schedule 13G is a dated control snapshot, not original-cost or realized-return evidence.
• Outcome is not economics — [researcher inference · strong]. IPO, acquisition, bankruptcy, reverse split or public-company scale proves a company event; without dated ownership and cash flows it does not prove Sutter Hill or Wythes proceeds.

Reconciliation and coverage

The transcript separately reports 46 deal events in 1964–1970—4, 11, 3, 4, 14, 9 and 1 by year—including four acquired Draper & Johnson positions in 1965; the 1966–1970 subtotal is 31. Most events are unnamed, and the event terminology may include follow-ons or other non-unique transactions. Those figures cannot be mapped by assumption to 35 named associated lineages. CHM printed pp. 63–65 / PDF pp. 67–69

The public 27-plus-board claim is another unresolved denominator: board service does not prove an investment, and a funded company does not prove Wythes held its board seat. Princeton memorial

Three Biovest-context names in the PAW feature—Neurex, BioSurface Technology, and Ontogeny—remain outside the accepted 41 pending legal-identity, vehicle, attribution, and deduplication work; Biovest itself is a vehicle. [contemporaneous record · provisional] PAW feature

Search boundary and selection bias

The bounded record covers the 1961 precursor through post-1999 filings and outcomes, using the Wythes/participant oral histories, official filings, the recovered firm biography, company aliases, successor/acquirer names, and explicit pass/failure terms. It searched Sutter Hill Capital, Sutter Hill Ventures, TOW, Genstar, Wythes/trust spellings, board/financing/IPO/merger/write-off/litigation combinations, and later-firm holdings. Commercial portfolio aggregators were discovery-only. The set overweights memorable wins, public issuers, adverse filings and companies Wythes chose to narrate; ordinary private outcomes and unnamed early deals are underobserved. source-map coverage assessment

Historical market, regulatory, and base-rate context

• 1958–1965 — scarce organized venture supply and an SBIC wrapper. [contemporaneous record · strong] The Small Business Investment Act created the federal framework; the official notice fixes Sutter Hill Capital's license to 20 September 1962. Wythes remembered a $250,000 SBIC investment limit, but no license file, leverage draw or examination record was found. Small Business Investment Act · Federal Register, 1975, printed p. 42412
• 1960s — local technical density, few local syndicate partners. [investor-stated · moderate] Wythes described door-to-door Peninsula sourcing and traveling to New York after local rejections. Measurex shows the benefit and cost of a thin financing network; no contemporaneous acceptance/rejection base rate survives. CHM printed pp. 43–46 and 58–64 / PDF pp. 47–50 and 62–68
• 1970s — partnership transition and institutional-capital opening. [contemporaneous record · strong] The 1970 partnership and 1975 SBIC-license surrender are distinct. The 1979 prudence rule belongs to the later ERISA-era institutional context; Wythes's association role does not prove personal authorship or causal responsibility. Federal Register, 1979-06-26
• 1980s–1990s — more capital, more competition, and public-market exit sensitivity. [investor-stated · moderate] Wythes remembered weaker deal sharing in the 1980s and distribution of unrestricted public stock to partners, leaving sale timing to holders. The 1995 Federal Reserve study supplies institutional background, not a Sutter Hill base rate. CHM printed pp. 103–108 / PDF pp. 107–112 · Federal Reserve Staff Study 168
• 1999–2001 — dot-com selection and hindsight. [investor-stated · provisional] Wythes contrasted Sutter Hill's discipline with Webvan-style capital intensity, but the complete dot-com exposure, contemporaneous memos and loss schedule are absent. A selected survivor narrative cannot establish a superior hit rate. CHM printed pp. 124–131 / PDF pp. 128–135

No numerical venture base rate is applied to an individual case: the sources do not make stage, geography, sector, vintage, accounting and exit definitions comparable.

Entity, vehicle, and incentive chronology

Incentive conclusion — [researcher inference · moderate]. The evergreen form plausibly reduced fundraising-timing pressure and supported reserves, while a single continuing valuation/accounting system, personal co-investment, LP withdrawal rules and succession could create other alignment questions. Without the partnership agreements, the structure cannot be declared superior to a dimension-matched finite fund. Kauffman alignment critique

People, capital, and influence map

Eight-role responsibility grid for material cases

Comprehensive 41-family ledger

Each row reports the strongest surviving association and its missing chain. None is a complete Wythes-personal investment reconstruction.

Excluded audit row — Balance Bar. [contemporaneous record · strong] The routed filing belongs to T. Rowe Price Balanced Fund, Inc., not Balance Bar Foods; Wythes appears as a fund director. It is excluded from the company union and retained only as a source-map correction. SEC registration, signature page

Bounded evidence summaries — not ex-ante reconstructions

Micro Magnetics-MMI — control before pedigree

[investor-stated · moderate]. Wythes remembered an attractive automated money-changing product and a founder with HP, consulting and Harvard credentials. Sutter held a minority position without another venture firm, while the founder retained majority control; Wythes later judged that salary, board and liquidity preferences blocked scaling. No memo, security, board minutes or disposition record survives. The narrow documented process claim is that Sutter did not repeat that single-investor/majority-founder structure; whether later deals actually followed the rule requires the portfolio contracts. CHM printed pp. 54–56 / PDF pp. 58–60

Kasper Instruments — local sourcing, co-investment, and founder liquidity

[investor-stated · moderate]. Wythes says he found George Kasper through local canvassing, formed a mask-aligner company in 1967, brought in Page Mill Associates after the MMI lesson, and served with John Young on the board. The founder later sold stock to Sutter/Page Mill for a reported $1 million; Wythes and Young preferred continued growth, and Cutler-Hammer later acquired the company. The source supports sourcing, syndication, board and secondary-liquidity actions, but not initial check, security, ownership, follow-ons, acquisition price, or vehicle cash flows. CHM printed pp. 56–58 and 65 / PDF pp. 60–62 and 69

Micrex-Xidex — legal boundary, management change, and rescue

[investor-stated · moderate]. Five people left Memorex in 1969. Wythes says counsel was used to police trade-secret boundaries; the company changed its name from Micrex after a threatened dispute. In 1972 the board replaced the founding CEO, Wythes recruited Les Colbert, and he later authorized a payroll bridge before a 1973 round. An IPO and later cash sale followed. The case supports persistent governance activity, but the threatened dispute, bridge authority, financing terms and economic outcome lack independent documents. CHM printed pp. 84–89 / PDF pp. 88–93

Priam-Quantum — competitor consent and information separation

[investor-stated · moderate]. Because Sutter already backed Priam, Wythes says he asked Priam CEO Bill Schrader before investing in Quantum. Schrader consented if no Sutter partner joined Quantum's board; Sutter used Qume CEO Bob Schroeder as a proxy and shared the deal with three other firms. This is the strongest narrated conflict-control design, yet no consent letter, recusal policy, minutes, allocation document or proof of information-barrier effectiveness surfaced. CHM printed pp. 102–103 / PDF pp. 106–107

Decision labs

Decision lab 1: Measurex — syndicating beyond the SBIC limit

Decision-date freeze: late 1967 to January 1968. Bossen had operating experience, an MIT engineering background, a process-control concept for paper production, and a business plan; Sutter could not supply the full round and local investors had declined. [retrospective witness · moderate] Bossen printed pp. 18–21 / PDF pp. 24–27

• Known then: Sutter was prepared to support the company; the target round was reported around $1.25–$1.3 million; Wythes remembered a $250,000 Sutter limit/component; an industrial customer reference changed one New York investor's view.
• Unknown/disputed then: Product reliability, sales cycle, customer concentration, exact security, fully diluted ownership, board rights, reserve needs and whether the Wythes versus Bossen chronologies reconcile exactly.
• Believed thesis: A credible operator plus an industrial sensor with customer value could support a new local company. [investor-stated · moderate]
• Alternatives/dissent: Stop after local rejections; reduce the plan; seek strategic capital; syndicate nationally. Davis & Rock and BofA declined in Wythes's account; their original reasons are incomplete.
• Terms/governance/milestones: Wythes says aggregate ownership was negotiated conceptually, but percentages are absent. Before investing, require a customer validation, installation milestone, unit economics, named board lead and pro-rata reserve plan.

Analyst worksheet before reveal: Would you invest then? What fact would kill the deal? What ownership, board and protective terms fit a capital-constrained industrial startup? What customer or prototype evidence unlocks the next check?

Actual decision and outcome — reveal. Sutter invested/sponsored and Wythes helped assemble a wider syndicate; Bossen names Bancroft, Reid and Peggy Dennis and others, while Wythes names additional institutions. Measurex later went public and was sold to Honeywell, but no Sutter or Wythes realization ledger survives. [retrospective witness · moderate] Wythes CHM printed pp. 58–64 / PDF pp. 62–68

Decision lab 2: Diablo-Qume — repeat-founder network with conflict risk

Decision-date freeze: Diablo in 1969, Qume in the early 1970s. Diablo had a team from Frieden and a disk-drive/printer path; David Lee later proposed Qume with manufacturing colleague Lon Israel, neither presented as a ready CEO. [retrospective witness · moderate] CHM printed pp. 90–93 / PDF pp. 94–97

• Known then: Wythes served on Diablo's board; Lee's daisy-wheel work was visible; Qume drew on the prior relationship; Sutter expected to recruit leadership.
• Unknown/disputed then: Patent scope, ownership of inventions, Diablo consent, confidential information, Qume security/ownership, board allocation and whether Wythes personally approved Qume.
• Believed thesis: Proven technical talent could seed a second company if a suitable CEO were recruited. [investor-stated · moderate]
• Alternatives/dissent: Keep Lee at Diablo; license the technology; fund Qume with explicit consent/barriers; refer the deal to an unconflicted firm; pass.
• Terms/governance/milestones: Require written conflict consent, IP opinion, separate directors/data rooms, CEO-search deadline, prototype/customer gates and a recusal protocol.

Analyst worksheet before reveal: Would you back Qume while holding Diablo? What IP or loyalty fact kills it? What information rights and board terms contain conflict? What milestone unlocks follow-on capital?

Actual decision and outcome — reveal. Sutter backed Qume and recruited Bob Schroeder; Diablo was sold to Xerox and Qume to ITT. Comstock later described patent conflict and Sutter concern at firm level; the record does not assign the dispute to Wythes or prove the safeguards used. [retrospective witness · moderate] Comstock pp. 35–36

Decision lab 3: Apple-Cisco — price and remediable management risk

Decision-date freeze: dates not recovered. Wythes remembered Apple at a $65 million valuation after startup stage and Cisco early, with a technically promising company but serious business-management weakness. Wythes and Tench Coxe reviewed Cisco; the original memos and offers are absent. [investor-stated · provisional] CHM printed pp. 153–156 / PDF pp. 157–160

• Known then: Apple price appeared high to Sutter; Cisco founders lacked business-management readiness in Wythes's account.
• Unknown/disputed then: Exact dates, securities, ownership offered, revenue/market evidence, cash needs, alternative valuations, recruiting candidates and internal dissent.
• Believed thesis/risk: Apple failed the remembered price screen; Cisco failed management and market judgment. [investor-stated · moderate]
• Alternatives: Pass; stage capital; negotiate price/ownership; condition financing on a chair/interim executive; recruit before closing; obtain pro-rata rights with smaller exposure.
• Kill criteria/milestones: For Apple, unit economics and price-to-upside under explicit scenarios; for Cisco, founder consent to governance, credible bridge leadership and customer/network adoption evidence.

Analyst worksheet before reveal: Would you invest on the remembered facts alone? What price or founder-governance fact kills each deal? What ownership/terms compensate for uncertainty? What evidence unlocks a first or follow-on check?

Actual decisions and outcomes — reveal. Sutter passed both. Apple and Cisco later became major companies; Wythes credited Sequoia with bridging Cisco's management gap and said Sutter had not attempted the same remediation. That is a claimed lesson, not evidence that a later institutional process changed. [investor-stated · moderate] CHM printed pp. 154–156 / PDF pp. 158–160

Decision lab 4: Duplicon — product promise, reliability, and stop rules

Decision-date freeze: pre-1975, exact date unknown. A former Xerox employee proposed a better copier; Sutter formed/funded Duplicon. The surviving story centers on a demonstration in which the machine failed dramatically, not on a decision memo. [investor-stated · provisional] CHM printed pp. 156–157 / PDF pp. 160–161

• Known then: Relevant founder background and an apparently attractive product concept.
• Unknown/disputed then: Prototype reliability at approval, safety certification, IP freedom, manufacturing cost, service burden, demand, check, ownership, reserves and board rights.
• Believed thesis: A technically improved copier could challenge incumbents. [researcher inference · insufficient public record]
• Alternatives: Fund prototype only; partner/license; require independent reliability testing; stage manufacturing capital; pass.
• Kill criteria/milestones: Repeated duty-cycle test, fire/safety clearance, copy-quality and cost targets, customer pilots, service economics and a fixed stop-loss budget.

Analyst worksheet before reveal: Would you finance before independent reliability evidence? Which failure rate kills the deal? What staged ownership/terms fit hardware risk? Which validation unlocks production capital?

Actual decision and outcome — reveal. Sutter backed the company; Wythes called it unsuccessful and said it was sold to A.B. Dick. The record does not prove a write-off, proceeds or a Duplicon-specific process change. MacNaughton's tolerance of losses is LP-relationship evidence, not proof this deal lost a known amount. [investor-stated · provisional] same CHM locator

Filing-level ownership controls

Portfolio-level forensics and base-rate limits

[researcher inference · strong]. The most plausible observable pattern is not a proven selection edge but a repeatable operating system: local sourcing, syndication when capital was constrained, management recruitment, patient follow-on capital and team deliberation. The strongest rival explanation is access plus survivor memory under a long-lived vehicle. The current set cannot discriminate between them because ordinary outcomes, ownership and cash flows are missing.

Failures, misses, adverse controls, and process change

The legal/adverse record also constrains a celebratory narrative. The Durrum notice is an affiliated-transaction application, not an enforcement finding. Teltone disclosed a reverse-split process without a special committee or independent fairness opinion; those are process facts, not automatic wrongdoing. In Activision, the VC-group Section 11 statutory-underwriter claim was dismissed with prejudice because selling into a firm-commitment underwriting did not itself establish distribution participation; the opinion does not find Wythes committed fraud. Durrum SEC notice · Teltone proxy · Activision, pp. 418–424, 438–439

Reported-return controls — what cannot be claimed

No figure is averaged, compounded, personalized or used to derive wealth. No company multiple is calculated because complete securities, ownership, dilution, follow-ons, dates and distributions are absent.

Uncertainty and retrieval ledger

Until those records are retrieved, the strongest defensible conclusion is narrow: Wythes's record supports historically important evidence about syndication, management intervention, patient vehicle design and team practice, while personal deal economics, career coverage and a literal Pareto result remain unproved.

Coverage and evidence base

The reconciled source map contains 62 rows: 27 ★★★, 31 ★★, and 4 ★. Its 58 priority rows route exactly once as follows; the four ★ discovery rows remain outside the priority denominator.

The reconciled written-corpus inventory therefore contains 7 unique priority artifact families. Three controls are accessible in full, one control is accessible only through derivative fragments, and three strict attributable artifacts have no readable body. The strict attributable-work denominator is N=3: the edited first-person Done Deals chapter, the San Jose Mercury News letter/editorial, and the 1995 NYSE award remarks. 0/3 strict bodies are accessible in full and 0/3 are deeply analyzed: Done Deals has a lawful controlled-borrow route but was unread; the Mercury and NYSE bodies are unlocated. The larger seven-family inventory is 3 full, 1 partial, 3 metadata/body-unavailable, and 3 deeply analyzed. Each family has a note; controls are not silently promoted into Wythes-authored works. Annotated source map, direct voice and writings

The CHM and Bancroft transcripts are two manifestations of one 2006 oral-history work and belong to the spoken corpus, not the written-work denominator. The archived Sutter Hill biography is unsigned corporate copy. The 1987 Florida/Kenney paper and Princeton institutional pages are research controls, not Wythes writings. Source map, access limits

Source-by-source analysis

For lucky alumni, biotech investments pay big dividends (2000)

Context, attribution, and access. Marvin Zim wrote this fully accessible Princeton Alumni Weekly feature; Wythes appears as one quoted investor in a broader survey of Princeton alumni involved in biotechnology. The page is direct voice only where language is explicitly attributed to Wythes, and its descriptions of companies and results remain Zim's editorial reporting. PAW feature, Wythes passage source note

Argument and reasoning. [investor-stated · moderate] Wythes's supported selection claim is that unusually strong scientific people are central to biotechnology investing, while investors should recognize that they are not themselves world-class scientists and compensate through intensive people checking. The causal chain is expertise boundary → deeper assessment of domain experts → reduced risk of backing weak science. The passage does not specify reference questions, independent scientific advisers, disqualifiers, or how scientific quality trades off against market, management, price, and financing risk. PAW feature, Scientists are key

Evidence, examples, and assumptions. The article says Biovest funded six biotechnology ventures and highlights Neurex and BioSurface Technology as successes; it also names Hybritech and Ontogeny among Sutter Hill investments. Those examples are neither a complete company roster nor Wythes-personal decisions. [researcher inference · provisional] The rule assumes scientific excellence can be identified through people diligence and that it predicts venture outcomes, but the article gives no rejected case, failed-science counterexample, capital weights, or complete six-company outcome set. PAW feature, Wythes passage

Decision implications and biography. [investor-stated · moderate] A defensible historical decision rule is: when the investor lacks the core technical competence, make verification of the relevant experts a gating diligence task. The piece cannot establish who sourced, sponsored, approved, governed, or exited any named company. Its 1964 founding shorthand also compresses a pre-existing 1962 SBIC, Wythes's 1964 hire, Draper's 1965 arrival, and the 1970 partnership; it is not a formation authority. PAW feature oral-history formation locators, source map

Economics and contradiction. [investor-stated · provisional] The feature gives reported: >35% average annual Sutter Hill return over the prior thirty years (Wythes's estimate), but supplies no vehicle, cash-flow series, gross/net definition, valuation method, distributions, fees/carry, or audit. [researcher inference · strong] The figure is therefore a firm-level self-estimate, not a personal return or proof that scientist selection caused performance; its two highlighted Biovest successes also create survivor-selection risk. PAW feature, Wythes passage

What Ever Happened to Venture Capital? (2011)

Context, attribution, and access. Sean D. Carr, Malgorzata Glinska, and Andrew King—not Wythes—authored this fully accessible Darden Batten briefing. They argue that larger institutional pools, pressure for earlier returns, reduced monitoring, and weaker IPO conditions moved venture capital away from a small-partnership company-building model. That industry thesis is the authors' synthesis. Wythes appears through three edited excerpts on p. 7. Darden briefing, pp. 1–2, 6–8 source note

Argument and reasoning. The editors combine non-contiguous oral-history remarks: company creation can produce useful products and public benefits; venture work is building businesses rather than executing transactions; and liquidity commonly takes seven to eight years. [investor-stated · moderate] The supported chain is patient capital → continuing company-building work → operating company → eventual liquidity, with useful products, employment, and taxes framed as desired social outputs. The briefing does not supply a Wythes reserves model, board procedure, sale rule, or company-level duration distribution. Darden briefing, p. 7 oral-history origin locators

Evidence, examples, and assumptions. Wythes's excerpts are illustrations in the authors' industry argument, not empirical proof of it. No Wythes investment case, partner vote, intervention, or exit is analyzed. [researcher inference · provisional] The implied operating assumption is that a vehicle and partnership can tolerate a multi-year path and still distinguish productive patience from delayed recognition of failure; neither half is tested here. Darden briefing, pp. 6–8

Decision implications, evolution, and contradiction. [investor-stated · moderate] A defensible historical rule is to underwrite venture as long-duration company creation, not a near-term financing trade. That complements the PAW expertise-boundary screen: select credible technical people, then remain engaged long enough to build. It does not show that every Sutter Hill decision followed either rule. Darden's 2011 page republishes 2006 speech, so it is not evidence that Wythes's thinking evolved after 2006. Its critique of institutional return pressure also sits beside Sutter Hill's own institutional LP base; the unresolved variable may be vehicle design and behavior, not institutional money alone. Darden briefing, p. 7 PAW feature

Wythes Committee plans for coming decade (2000)

Context, attribution, and access. Maria LoBiondo wrote the fully accessible PAW article about a Princeton trustee committee chaired by Wythes. The article describes a 68-page collective report and its university-planning recommendations; it is not the report body. A named chairmanship does not prove that Wythes drafted, signed, or personally endorsed every recommendation. PAW article, opening and report description source note

Argument, evidence, and assumptions. The article's central subject is Princeton's coming decade, not venture selection. Its reported chronology and recommendations belong to LoBiondo's account of a collective governance process. The underlying final report remains unavailable, so its exact authorship statement, evidence base, dissent, and revisions cannot be evaluated. PAW article Princeton finding-aid route

Decision implications and boundary. [researcher inference · insufficient public record] Institutional chairmanship may suggest experience with long-horizon planning, but translating a university committee's work into a Wythes investment rule would be overreach. This family contributes biographical and attribution discipline, not a sourcing, picking, portfolio, board, or exit model. No intellectual-evolution claim is supportable until the report and drafting record are read. source note

Venture Capital Roundtable (1995) — original body unavailable

Renee Deger edited this Venture Capital Journal roundtable; it is not Wythes-authored. The September 1995 issue was not lawfully available. Florida and Kenney's later manuscript and endnotes attest two Wythes fragments on original printed pp. 34 and 36, concerning competition for entrepreneurs and incentives around parallel funds and management fees. The original questions, participant list, complete answers, editing, and surrounding argument are unknown. Florida/Kenney derivative discussion, PDF pp. 147–148 endnote 219, PDF p. 358 endnote 411, PDF p. 369

[investor-stated · provisional] Competition for founder relationships and fund-structure incentives are candidate themes only; no reusable rule is assigned from derivative fragments. Retrieval requires a publisher backfile, licensed periodical database, interlibrary loan, or physical September 1995 issue. source note

Paul Wythes: Sutter Hill Ventures, in Done Deals (2000) — controlled body unread

This is the highest-priority strict artifact because Gupta presents Wythes in edited first person. Lawful public routes establish the book, editor, edition, and chapter start at printed p. 151; authenticated controlled borrowing or a lawful library copy is required for the body, which was not read. No search snippet, unofficial scan, or later summary was substituted. Consequently no argument, reasoning chain, evidence, assumption, rule, decision example, or evolution claim is assigned. Internet Archive controlled-borrow route source note

San Jose Mercury News letter/editorial (circa 2002–2006) — body unlocated

Wythes later recalled writing an item about public disclosure of venture-fund results and alternately called it a letter and an editorial. That one oral-history account supports an authorship and subject lead, not the exact title, format, date, edition, wording, or argument. [investor-stated · provisional] Concern about effects of disclosure on private portfolio companies is a topic signal only; no transparency doctrine or decision rule can be reconstructed. Licensed Mercury/NewsBank or ProQuest archives, San Jose library indexes or microfilm, and UC/CalPERS litigation files remain the lawful retrieval routes. oral-history locator, PDF pp. 44–45 source note

1995 Princeton award remarks at the NYSE — manuscript unlocated

The oral history identifies a 1995 Class of 1955 award dinner and talk in the NYSE Board Room and says a copy later existed. It retrospectively associates the speech with Princeton, venture philosophy, company formation, jobs, and useful products, but the manuscript's sequence, wording, evidence, and recommendations are unavailable. [investor-stated · provisional] These are subject tags, not a written doctrine. The Board Room versus exchange-floor shorthand is also unresolved. CHM/NVCA project files, Princeton Class of 1955 records, and NYSE event archives are the next lawful routes. oral-history event locator, PDF pp. 17–18 copy/topic locator, PDF p. 112 source note

Cross-source synthesis

[researcher inference · strong] The written record is too thin to support a comprehensive Wythes doctrine. Only the PAW biotech feature supplies full, non-derivative published Wythes quotations; Darden supplies a full secondary arrangement of one rights-restricted oral history; the other strict bodies are unread or missing. Repetition across Darden and the NYSE retrieval lead comes from the same 2006 oral-history work and is not independent written confirmation. PAW feature Darden source note NYSE source note

• Picking and people — [investor-stated · moderate]. The clearest written signal is intellectual humility at the expertise boundary: evaluate whether the scientists are genuinely excellent and do not mistake investor confidence for technical competence. The corpus does not reveal an operational scorecard or documented rejection. PAW feature, Scientists are key
• Governance and exits — [investor-stated · moderate]. Darden's traceable excerpts support company building and a long liquidity horizon, but not specific board mechanisms, ownership targets, reserves, sale criteria, or realized duration. Darden briefing, p. 7
• Sourcing — [researcher inference · insufficient public record]. The roundtable derivative suggests competition for entrepreneur relationships, but the original context is missing. The writings do not establish a reproducible sourcing system. roundtable source note
• Pricing and ownership — [researcher inference · insufficient public record]. Across the seven processed families, no accessible writing states a price ceiling, dilution rule, ownership target, or security preference. coverage accounting
• Portfolio construction — [researcher inference · provisional]. Biovest's reported six-company pool could reflect a basket response to scientific uncertainty, but absent the complete roster, weights, and outcomes it cannot support a portfolio rule. PAW feature
• Vehicle incentives — [investor-stated · provisional]. Deger's roundtable may contain a critique of parallel funds and fee incentives, but derivative fragments cannot establish the complete argument or its application to Sutter Hill's structure. roundtable source note
• Mission — [investor-stated · moderate]. The Darden/oral-history material frames useful products, jobs, and taxes as benefits of company building. [researcher inference · insufficient public record] No written corpus evidence measures those outcomes or shows that mission overrode investment economics. Darden source note

The causal synthesis that survives is narrow: domain uncertainty makes expert-person diligence important; venture outcomes then require time and company-building work; vehicle incentives must permit that work. [researcher inference · provisional] Each link is plausible from the sources, but the corpus does not connect all three links to a single documented Wythes decision, compare successes with failures, or demonstrate causation. PAW source note Darden source note roundtable source note

Intellectual chronology, evolution, and contradictions

• 1995: The NYSE award remarks and Deger roundtable are the earliest located attributable/direct-voice leads. One manuscript is unlocated and the other original body unread, so no 1995 baseline doctrine can be reconstructed. NYSE source note roundtable source note
• 2000: Gupta's edited first-person chapter, Zim's biotech feature, and LoBiondo's committee coverage appeared. Only Zim's full article yields an investment rule: deference to and verification of scientific expertise. Gupta's body is unread, while the committee family supplies institutional biography rather than investment philosophy. Done Deals note PAW biotech note committee note
• Circa 2002–2006: The unlocated Mercury item suggests Wythes entered a public debate over LP-result disclosure. Without the body, neither his reasoning nor whether the item changed an earlier view is knowable. Mercury source note
• 2006 and 2011: The oral history articulated company-building, time-horizon, and public-benefit themes; Darden rearranged three excerpts in 2011. The later publication is transmission of 2006 speech, not a new Wythes writing or evidence of evolution. Darden source note

Six tensions remain unresolved rather than averaged away:

1. Reported: >35% average annual Sutter Hill return in PAW, reported: roughly 36% net IRR in Wythes's later discussion, and reported: roughly 40% pre-1981 Sutter Hill IRR in Draper's recollection are separate retrospective claims with different speakers and possible periods. [researcher inference · insufficient public record] The transcript's later description of 36% as carried interest is internally inconsistent with its earlier economics and cannot be normalized into a return or compensation fact without partnership records. PAW feature source-map return barriers
2. Wythes publicly offered an aggregate firm-return estimate while apparently opposing detailed public LP-result disclosure. [researcher inference · provisional] Aggregate performance and company/security-level disclosure can create different risks, so the two acts are not a demonstrated contradiction without the Mercury body. PAW feature Mercury source note
3. Darden frames institutional return pressure as corrosive, while PAW describes institutional Sutter Hill LPs. [researcher inference · provisional] The material could be reconciled by vehicle duration, governance, or behavior, but no partnership agreement tests that explanation. Darden briefing, pp. 6–8 PAW feature
4. PAW's accessible rule emphasizes scientific excellence and investor humility, while the wider oral record emphasizes team selection, shared credit, management, and recruiting. [researcher inference · insufficient public record] These ideas can coexist, and no accessible writing ranks them or shows how Wythes traded technical quality against management quality. PAW feature source-map attribution boundary
5. [researcher inference · insufficient public record] The accessible written corpus states neither a price-discipline rule nor a flexibility rule, so no price-versus-flexibility evolution can be inferred. Importing those themes from unattributed secondary synthesis would erase the corpus and authorship boundary. coverage accounting Darden source note
6. Long patience can enable company building or conceal escalation of commitment. [researcher inference · provisional] The accessible writings contain no stop-loss, milestone, reserve, or write-off rule that resolves the tension. Darden source note

Decision playbook and modern translation

1. Treat expertise boundaries as a diligence trigger

• Historical rule — [investor-stated · moderate]: when the venture's core technical truth lies outside the investor's competence, verify the quality of the domain experts rather than pretending to be one. PAW feature, Scientists are key
• Diagnostic questions: What must be true technically? Who can falsify it? What evidence distinguishes reputation from actual scientific quality? Which references are independent of the founders?
• Boundary and failure modes — [researcher inference · provisional]: expert quality cannot replace market, price, management, regulatory, or financing diligence; prestige, consensus references, and inaccessible technical claims can turn humility into outsourced judgment. The Wythes passage supplies no kill criteria or failed example. PAW source note

2. Underwrite company building and liquidity on a multi-year clock

• Historical rule — [investor-stated · moderate]: treat venture as building an operating company over years, not as completing a financing transaction. Darden briefing, p. 7
• Diagnostic questions: Does the vehicle have time and reserves for repeated operating work? What board contribution is expected? Which milestones justify continued patience? What evidence would make the partnership stop funding or seek liquidity?
• Boundary and failure modes — [researcher inference · provisional]: duration is capacity, not proof of quality. Without explicit milestones, follow-on rules, and write-off discipline, patience can become escalation of commitment. Darden provides no Wythes company-level test. Darden source note

3. Audit vehicle incentives before claiming patience

• Historical candidate — [investor-stated · provisional]: the unavailable 1995 roundtable may connect multiple parallel funds and management fees to organizational incentives. It is not yet a verified rule. roundtable source note
• Diagnostic questions: Which revenue comes from management fees versus realizations? Do new vehicles compete with existing holdings for attention? Who bears extended-hold costs? Can the decision maker wait without disadvantaging LPs or successors?
• Boundary and failure modes — [researcher inference · provisional]: neither an evergreen structure nor a finite term guarantees aligned behavior; claims require the actual agreement, compensation, valuation, and withdrawal mechanics, none of which appears in the written corpus. roundtable source note source-map vehicle barriers

Modern VC translation (as of 2026-08-21)

No current company, market, fund-term, or regulatory case was independently researched for this corpus. The applications below are therefore analyst-created hypotheticals, not claims about present firms.

• Expertise boundary — [researcher inference · provisional]: the principle transfers to any technically opaque company. A modern analyst should identify the claim-specific independent expert, predefine falsifiers, and document disagreement. What does not transfer is treating a famous technical founder or adviser as a substitute for product, market, security, and regulatory work. Applicability signals are high technical opacity, concentrated expert disagreement, and outcomes that can be tested before scale. Likely misuse: prestige diligence. Historical anchor: PAW feature.
• Long-duration company building — [researcher inference · provisional]: the principle transfers only when the actual vehicle, reserves, ownership, and governance permit years of work. A hypothetical capital-intensive technical company may require that patience, but the analyst must map milestones and kill criteria before investment. What does not transfer is Wythes's retrospective seven-to-eight-year estimate as a universal current forecast. Applicability signals are long validation cycles and a credible board work plan; likely misuse is rationalizing a stalled company. Historical anchor: Darden source note.
• Vehicle incentives — [researcher inference · provisional]: test fee dependence, cross-fund allocation, continuation decisions, valuation discretion, and successor incentives in the actual governing documents. Fund, market, and regulatory assumptions differ by vehicle and jurisdiction, so no historical Sutter Hill label can replace a current contract review. Applicability signals are overlapping mandates and economics that reward asset accumulation; likely misuse is assuming evergreen automatically means patient or aligned. Historical anchor: roundtable source note.
• Public-benefit rhetoric — [researcher inference · provisional]: useful products and job creation may inform an impact hypothesis, but a modern IC should define measurable product, stakeholder, and financial outcomes separately. Applicability requires a testable causal metric; likely misuse is treating mission language as evidence of impact or return. Historical anchor: Darden source note.

Best writings and unresolved record

Ranked by present research value rather than presumed literary quality:

1. “For lucky alumni, biotech investments pay big dividends” (2000): [researcher inference · strong] the best currently readable written route because it preserves Wythes's direct comments on scientific expertise and clearly exposes the limits of his reported return estimate. source note
2. “What Ever Happened to Venture Capital?” (2011): [researcher inference · strong] the strongest compact route to company-building and duration themes, but derivative, edited, and traceable to one oral history. source note
3. Paul Wythes: Sutter Hill Ventures, in Done Deals (2000): [researcher inference · moderate] potentially the richest strict artifact because it is edited first person; it ranks below the readable controls until a lawful full body can be processed. source note
4. The San Jose Mercury News letter/editorial: [researcher inference · provisional] potentially decisive for confidentiality, LP transparency, and private-company effects, but title, date, format, and body remain unlocated. source note
5. The 1995 NYSE award remarks: [researcher inference · provisional] potentially the clearest self-selected philosophy statement, but only event and topic leads survive publicly. source note

The Deger roundtable remains an important edited-direct-voice control, but its original issue must be read before its two derivative fragments can influence the synthesis. source note LoBiondo's PAW page is useful for committee/individual authorship discipline, while the 68-page report and drafting record remain archival retrievals rather than investment writings. source note

The conclusions most likely to change after retrieval are the completeness of the people-diligence rule, Wythes's price and ownership views, his actual LP-disclosure reasoning, his treatment of fund incentives, and whether patience was paired with explicit stopping rules. Required lawful retrievals are: authenticated borrowing or a library copy of Done Deals; licensed Mercury/NewsBank/ProQuest or microfilm access; the September 1995 Venture Capital Journal issue; CHM/NVCA project files; Princeton Class of 1955 and Mudd archive holdings; NYSE event files; and the collective Wythes Committee report plus drafting metadata. Until those bodies are read, the corpus supports a narrow, qualified playbook rather than a complete authored philosophy.

Coverage and evidence base

The declared spoken-record plan contains 16 routed record families: 14 spoken works plus 2 archive/access routes. By evidence role, that is 1 direct Wythes work, 13 participant/counterparty control works, and 2 archive routes. All 16 have exact-schema notes and guide entries below. Of the 14 spoken works, 13 have full transcript bodies reviewed and one—the Smithsonian Draper record—is only a five-page edited excerpt. The two archive pages were reviewed as custody/access metadata, not content. Wythes note Smithsonian excerpt note CHM collection note CHM access note

The sole direct Wythes work is one oral history recorded across three sessions on May 8, May 10, and July 10, 2006. The canonical body locator is CHM's 163-page manifestation, where the sessions begin at PDF pp. 5, 56, and 99; physical p. 4 is the unnumbered title page. Bancroft's 177-page donated transcript is a second manifestation, not another interview. The manifestations have different front matter and pagination. No standalone public Wythes audio asset was retrieved or reviewed. CHM transcript, PDF pp. 5, 56, and 99 Bancroft manifestation manifestation/rights note

A percentage of “all Wythes talks” is not computable. No complete universe of private speeches, recorded appearances, or archive audio has been established. The declared plan accounts for 16/16 routed records; it does not mean every historical Wythes appearance is known. The original NYSE award speech remains unlocated, and CHM audio availability remains unresolved without an item-level request. NYSE writing note CHM access note

Rights and independence are equally important constraints. No verbatim Wythes transcript language is reproduced because publication permission was not established. Five Draper works are five interview artifacts but one correlated witness family; two Johnson works are two artifacts but one correlated witness family. Multiple speakers in a panel remain one event family. Repetition demonstrates narrative consistency or memory convergence, not transaction truth or investing efficacy. Draper NVCA note Draper CHM note Johnson Bancroft note Johnson CHM note

Dated source-by-source guide

Recurring stories and frameworks

Wythes's own retrospective account

• Local, reciprocal sourcing — [investor-stated · moderate]. Wythes describes canvassing Peninsula technical companies, making low-cost introductions, and delaying technical work until a founder expressed financing interest. The mechanism was scarce competition plus geographic density; the record supplies no complete contact-to-investment funnel. Wythes note, sourcing CHM printed pp. 43–46 / PDF pp. 47–50
• Long-term equity partnership — [investor-stated · moderate]. His early pitch emphasized equity, mutual reference checking, syndication, and continuing partnership rather than a loan. This is rhetoric from retrospective memory, not evidence that every deal received long-term support. CHM printed pp. 44–45 / PDF pp. 48–49
• People, management, and team credit — [investor-stated · moderate]. Wythes repeatedly frames venture decisions as partnership deliberation, places portfolio-company people ahead of investor celebrity, and treats missing management as a problem investors may need to solve. The Cisco pass is the internal counterexample: Sutter Hill saw the weakness but did not recruit the bridge management that Sequoia later supplied. CHM printed pp. 149–156 / PDF pp. 153–160
• Evergreen alignment — [investor-stated · moderate]. Wythes says the continuing pool supported pacing and pro-rata follow-ons without serial-fund pressure. The causal superiority claim remains unproved because the partnership agreement, capital accounts, valuation policy, withdrawals, and complete outcome ledger are missing. CHM printed pp. 73–76 / PDF pp. 77–80 CHM printed pp. 149–153 / PDF pp. 153–157
• Failure is intrinsic, but the denominator is absent — [investor-stated · moderate]. Wythes discusses Duplicon, Webvan, an unnamed mining write-off, and the Apple/Cisco/Tandem/Siebel passes alongside winners. The transcript still names only a selected subset, so failure rates, ordinary outcomes, and Pareto concentration cannot be computed. CHM printed pp. 50–51 / PDF pp. 54–55 CHM printed pp. 152–159 / PDF pp. 156–163

Participant and counterparty controls

• Formation convergence, not sole-founder proof — [retrospective witness · moderate]. Draper and Johnson consistently place Wythes at the pre-existing Sutter Hill operation before Draper's 1965 arrival. Their accounts support sequence but not Wythes founding the 1961 company or 1962 SBIC. Five Draper records remain one witness family, and two Johnson records remain one witness family. Draper Bancroft note Johnson Bancroft note
• Complementary team claim — [retrospective witness · provisional]. Draper characterizes Wythes as more cautious and technically grounded while casting himself as more outward and risk-taking. That role split belongs to Draper's narration; it is not an independently observed Wythes philosophy. Draper Bancroft, printed pp. 40–41 / PDF pp. 49–50
• Measurex syndication — [retrospective witness · moderate]. Bossen provides founder-side evidence that Wythes helped assemble capital after Sutter Hill decided to support the company. Pete Bancroft, Draper, Reid and Peggy Dennis, and other investors remain part of the financing story. Bossen note Bossen printed pp. 19–20 / PDF pp. 25–26
• Board monitoring and conflict boundary — [retrospective witness · moderate]. Comstock places Wythes on Diablo's board and recalls a narrow expense/profitability challenge. He also describes Sutter Hill financing later competitor Qume during a patent dispute; that tension is firm-level unless decision records assign Wythes a role. Comstock note CHM/PDF pp. 35–41
• Mentorship — [retrospective witness · moderate]. Newhall identifies Wythes as a mentor and NEA management-committee adviser. The collective mentor description does not prove any Sutter Hill deal rationale or return. Newhall note CHM/PDF pp. 39–40
• Negative attribution controls — [contemporaneous record · insufficient public record]. The Linear panel never names Wythes and places Sutter Hill after four initial backers; Poduska assigns Apollo's initiating commitment to Draper; Edwards never mentions Wythes, Sutter Hill, or Measurex. These absences and alternate attributions prevent firm/team stories from becoming personal leads. Linear note Apollo note Edwards note

Chronology and intellectual evolution

The event chronology reconstructed from retrospective speech is specific but not equivalent to an evolution of published thought:

• 1961–1965: Sutter Hill Company, the SBIC, Wythes's hire, the Draper & Johnson portfolio transfer, and Draper's arrival are distinct events. Johnson and Draper support Wythes's prior presence; exact legal roles and transaction terms require contemporaneous records. Wythes formation note Johnson Bancroft, printed pp. 26–28 / PDF pp. 34–36
• 1967–1968: Bossen's Measurex fundraising recollection supplies the clearest independent Wythes-specific sourcing/syndication episode. Bossen printed pp. 18–21 / PDF pp. 24–27
• 1969–1975: Diablo board evidence, the Genstar-backed partnership, and later SBIC surrender belong to different company, vehicle, and legal layers. Participant oral histories support the narrative but not complete legal chronology. Comstock note Draper Bancroft note
• 1999: Wythes dates retirement from active management and describes a later special-LP role; later continuing-firm investments cannot be assigned to him without decision evidence. Wythes note, retirement CHM printed pp. 136–143 / PDF pp. 140–147
• 2002–2019 recorded corpus: Poduska, Comstock, Draper, Wythes, Johnson, Edwards, Bossen, Newhall, and panel participants narrated earlier history decades after the decisions. Recording dates show when memories were captured, not when the underlying beliefs formed. Apollo/Poduska note Comstock note Wythes note Newhall note

[researcher inference · insufficient public record] Intellectual evolution cannot be established confidently from a single located Wythes appearance in 2006. The three sessions show thematic breadth and internal cross-reference, not change over time. Later publications of Wythes excerpts and later participant interviews do not create a dated sequence of Wythes changing his mind. The NYSE manuscript, Mercury item, contemporaneous investment memos, LP letters, and earlier recorded speeches would be required. Wythes note writings evidence boundary

Biography and deal attribution

Contradictions, counterevidence, and access limits

• Formation shorthand: the oral histories converge on Wythes before Draper but do not support assigning Sutter Hill's founding to him in 1964. Public shorthand collapses a 1961 company, 1962 SBIC, 1964 hire, 1965 portfolio/team transition, 1970 partnership, and 1975 SBIC surrender. Wythes note Johnson Bancroft note
• Who initiated the 1965 transfer: Johnson's 2008 account describes a Sutter Hill offer generally; his 2019 account names Wythes as approaching the partners. [retrospective witness · provisional] The versions may differ only in granularity, but a purchase agreement is needed before assigning negotiation or signature credit. Johnson Bancroft note Johnson CHM note
• Returns: PAW reports Wythes's >35% annual Sutter Hill estimate over the prior thirty years; Wythes later discusses reported: roughly 36% net IRR; Draper separately remembers reported: roughly 40% pre-1981 Sutter Hill IRR. The speakers, periods, vehicle definitions, gross/net bases, and denominators differ, and none has a retrieved cash-flow schedule or audit. A later transcript label of 36% as carried interest conflicts with earlier 80/20 and 75/25 economics and is not normalized as fact. Wythes note, economics limits Draper NVCA note PAW writing note
• Evergreen and pioneer claims: Wythes and Draper retrospectively describe alignment benefits and several claimed “firsts.” [investor-stated · provisional] Without contemporaneous partnership agreements and industry controls, the record cannot verify first/only status or connect the structure causally to returns. Wythes note Draper CHM note
• Deal counterevidence: Apollo evidence favors Draper as lead; Linear offers no Wythes attribution; Qume raises a later-competitor conflict; Cisco shows management concerns did not become corrective action. These controls prevent a universal company-building or personal-lead narrative. Apollo note Linear note Comstock note
• Editorial/front-matter divergence: CHM and Bancroft Wythes manifestations are not byte-identical and contain different project/editor material. Interviewer praise, collection descriptions, footnotes, and introductory industry statistics are not Wythes speech. Wythes note, editorial boundary
• Metadata corrections: source bodies date Bossen to 2010/2011, Edwards to 2009 and Berkeley record 218456, Comstock to 2003/2004, Linear to 2014, Apollo to 2002, the Smithsonian excerpt to 2015, Newhall to 2017/2018, and the CHM collection announcement to 2019. These source-level dates supersede stale routed labels and are preserved in the notes. Bossen note Edwards note Linear note Smithsonian note
• Survivorship and memory: oral histories favor memorable winners, misses, and origin stories. Recollection decades later, repeated prompts, interview editing, common NVCA context, and same-witness retellings create correlated error. No complete public portfolio, board, decision, or appearance denominator cures that bias.

Audio, transcript, and archive attempts

• The full CHM Wythes transcript, ref. X8628.2008 / asset 102781073, was processed as the canonical body; the Bancroft 177-page manifestation was checked for custody, rights, and variant-front-matter context. No transcript manifestation was counted twice. CHM body Bancroft manifestation
• CHM's announcement confirms the transferred NVCA collection and names Wythes, but the public route did not expose a standalone audio object. CHM collection announcement route note
• CHM's access page asks researchers to identify catalog numbers and submit requests in advance. No request was sent, denied, or left unanswered during the documented search. Audio access is therefore unresolved, not proven impossible. CHM access instructions access note
• Searches of CHM, YouTube, Stanford/eCorner, C-SPAN, major podcast indexes, and general web/video results found no independent Wythes talk, panel, podcast, or broadcast interview. The unlocated 1995 NYSE remarks remain a written/speech retrieval lead, not a reviewed appearance. source-map access assessment
• The HBS Draper route exposes a 48:04 video and official transcript; the transcript has no timestamps, so none were inferred. The Smithsonian record is only an edited excerpt. No raw transcript, audio, video, or caption file is stored in the investor folder. HBS note Smithsonian note

If Wythes audio is later obtained lawfully, the next researcher should record acquisition route/date, runtime, file size and hash, session boundaries, and its relationship to both transcripts; any ASR should be treated only as navigation until audited against the audio. Permission for research access and permission to publish wording must still be resolved separately.

Five records to study first

1. Wythes's 2006 oral history — direct voice; full transcript; audio unresolved; paraphrase only. It is the indispensable spine for formation, sourcing, vehicle design, company narratives, retirement, and misses, while its retrospective and self-reported limits remain visible. note
2. Draper's 2008 Bancroft oral history — participant control; full transcript; restricted quotation. It is the richest counter-account on formation, Genstar, team roles, and Apollo, but should be read once as part of one correlated Draper witness family. note
3. Pitch Johnson's 2008 Bancroft oral history — participant/counterparty control; full transcript; restricted quotation. It best separates the 1962 Draper & Johnson firm, 1965 sale, Wythes's prior presence, and Johnson's exit. note
4. Bossen's Measurex oral history — founder-side control; full transcript; paraphrase only here. It supplies the strongest independent company-specific evidence of Wythes's syndication work while preserving the full financing group. note
5. Comstock's Diablo/Qume oral history — founder/operator control; full transcript. It adds a specific Wythes board-monitoring episode and an adverse same-firm competitor/conflict boundary absent from the celebratory accounts. note

For a fast attribution stress test after those five, read the Apollo founder control, Linear founders panel, and Edwards negative control. For retrieval rather than content, use the CHM access route.

Reciprocal local sourcing before expensive diligence

• Rule — [investor-stated · provisional] In a dense technical community, create many low-cost founder contacts, offer a credible long working relationship, and reserve expensive diligence for cases where interest is reciprocal. Proximity is a sourcing-effort variable, not a return predictor. Wythes oral history, printed pp. 39–48 / PDF pp. 43–52
• Evidence status — Kasper is the only documented Wythes-applied canvassing case. Measurex is a relationship/referral contrast rather than a second canvassing application; Johnson independently supports the thin local-market practice, not efficacy. No contact funnel survives, so the model remains provisional. Kasper case · Johnson note
• Provenance and origin — Wythes dated the behavior to his 1964 entry into an existing Palo Alto SBIC. His operating jobs and local residence are chronology, not proved causal origins. [researcher inference · insufficient public record] profile, 1964–1970
• Mechanism and reasoning — [researcher inference · provisional] Low contact cost expands trials; repeated local interaction can reveal people and spinouts; reciprocity supplies a weak working-fit signal before expensive diligence. Density can instead produce homophily, repeated-source dependence, and consensus.
• Historical and vehicle context — The mechanism is bounded to a sparse 1960s Peninsula market and a small SBIC. Later competition, firm reputation, and national/sector investing changed contact costs and signal scarcity. historical context
• When to use — Use when a target population is dense, direct contact is cheap, the investor can offer useful context, and incremental information can be logged. Do not use locality or reciprocity as substitutes for independent evidence. [researcher inference · provisional]
• How to apply it — Map the population; log source/contact cost; state the value proposition; record reciprocity and new information; request one falsifiable next datum; advance or stop; compare channel conversion and ordinary outcomes; preserve rejections. [researcher inference · provisional]
• Examples —
  • Kasper — documented application: Wythes remembered finding it through local prospecting, bringing Page Mill, and joining the board; security and proceeds are missing. Kasper case
  • Measurex — documented contrast: Bossen independently credits Wythes's later syndication help, but not initial canvassing. Bossen note
  • Apollo — counterexample/attribution boundary: founder Poduska assigns the initiating Sutter Hill action to Draper, not Wythes. Apollo note
• Evidence — Wythes proves the later-stated procedure and Kasper application; Johnson supports era practice; Bossen and Poduska constrain extension. The record does not prove conversion, exclusivity, cost advantage, or return efficacy. talks synthesis
• Limits, failure modes, and exception ledger — Locality can become insular and reciprocity can select agreeable founders. Route results: Wythes and Johnson transcripts were processed; Kasper is the sole application; Bossen and Apollo provide contrasts; the unavailable roundtable only signals competition; the 41-family ledger has no source funnel; archived/current firm pages, obituaries, video/podcast sweeps, and filings yielded no activity log, contact ledger, rejected sample, or matched nonlocal set. writings route limit
• Evolution and contradictions — Wythes later described more competition and less sharing, but one 2006 appearance cannot establish when the method changed. National and sector cases show literal geography was not universal. [researcher inference · provisional] talks chronology
• Modern VC translation (as of 2026-08-21) — [researcher inference · provisional] Transfer useful low-cost contact, reciprocity, and incremental-information logs—not industrial-park nostalgia. Signals are cheap contact and unique learning; likely misuse is high-volume outreach or calling access skill. A conditional application is to make every channel earn deep work through a documented nonpublic insight; T011 contains no researched current sourcing market.

Expertise-boundary people diligence

• Rule — [investor-stated · provisional] When technical truth lies outside the investor's competence, identify independent experts capable of falsifying it and evaluate the scientific people while retaining responsibility for market, management, price, security, regulation, and the final decision. PAW feature, journalist subheading and Wythes passage
• Evidence status — PAW is the only full rule-bearing source. Biovest is a vehicle/program context rather than a reconstructed Wythes decision; Hybritech and other healthcare names lack source/sponsor records. The model is below the independent-source and application bars. writings synthesis
• Provenance and origin — The earliest accessible articulation is the 2000 PAW biotechnology feature. Engineering education may have enabled informed humility, but no source establishes causation. Unread Done Deals and missing speeches cannot supply origin. profile formation boundary
• Mechanism and reasoning — [researcher inference · provisional] Explicit competence limits reduce false precision; independent experts can test claims; retaining the whole decision prevents prestige diligence. It works only if incentives, independence, disagreement, and milestones are visible.
• Historical and vehicle context — The visible rule is biotechnology-specific. Biovest and later specialist pools are vehicles, not accepted company decisions or personal portfolios; the hardware stories do not prove a universal science protocol. reconciliation
• When to use — Use when a core scientific, technical, clinical, manufacturing, or regulatory claim is material and outside demonstrated team competence. Do not delegate market, governance, valuation, or portfolio responsibility. [researcher inference · provisional]
• How to apply it — Write the claim/falsifier; map experts/conflicts; seek dissent; request replicable evidence; translate it into a dated milestone; make a separate customer/economics/people/security decision; stop if the claim fails. [researcher inference · provisional]
• Examples —
  • Biovest — documented program context, application-poor: PAW records the rule but not the complete roster, decisions, weights, or outcomes. PAW note
  • Hybritech — researcher-applied illustration: a healthcare association exists, but no Wythes expert/reference decision is reconstructed. 41-family ledger
  • Measurex — researcher-applied contrast: customer/reference diligence illustrates independent evidence, but no source says the biotechnology rule was applied. Measurex lab
  • Duplicon — counterexample/boundary: weak product outcome illustrates validation risk, not a documented expert protocol. Duplicon lab
• Evidence — PAW proves the stated competence boundary; the oral record provides company context; Bossen supports a separate reference-oriented case; Duplicon supplies adversity. No source connects the rule to a named biotech approval/rejection, dissent, terms, or return. written playbook
• Limits, failure modes, and exception ledger — Experts can share incentives, consensus, or prestige; technical quality does not resolve market or price. Route results: PAW full body was processed; Wythes oral history adds context; Darden is derivative; firm/Biovest/issuer leads and participant histories were checked; Done Deals remains controlled unread and other strict artifacts unavailable; no six-company roster, advisers, expert memos, rejected scientists, disagreement record, or outcomes surfaced. writings coverage
• Evolution and contradictions — No dated sequence shows how the model moved from hardware to biotech or changed after failure. Scientific excellence and management quality can conflict; no accessible source ranks them. [researcher inference · insufficient public record] writings contradictions
• Modern VC translation (as of 2026-08-21) — [researcher inference · provisional] Transfer claim-specific experts, independence checks, dissent, and falsifiable milestones. Do not transfer deference to reputation. Signals are technical opacity and testable evidence; likely misuse is outsourcing judgment. A conditional application is a memo separating expert evidence from customer, economics, governance, and terms judgment; no current case was researched.

Make management gaps governable, then recruit or replace

• Rule — [researcher inference · moderate] Do not back pedigree or product promise unless ownership, board rights, founder consent, and co-investor structure make a named management gap governable; then recruit, replace, condition, or stop by a defined date. MMI and Xidex bounded summaries
• Evidence status — Wythes supplies MMI, Xidex, Qume, Cohesive, and Cisco; Comstock independently supports a Diablo board-monitoring action and Qume lineage; Bill Younger independently describes business-plan and recruiting work. Applications are retrospective, but breadth clears the moderate model bar. Comstock note · Almanac profile
• Provenance and origin — MMI is the earliest narrated control failure and claimed structural response; Xidex supplies a later recruitment/bridge intervention. No contemporaneous policy proves durable implementation. [investor-stated · moderate] Wythes note
• Mechanism and reasoning — [researcher inference · moderate] Rights determine whether diagnosis can become action; naming the gap makes it testable; recruiting/replacement can repair a viable thesis; conditions buy evidence; stopping limits endless repair. Founder consent, runway, authority, and candidate supply govern feasibility.
• Historical and vehicle context — Small SBIC/minority positions and scarce technical managers made structure and networks material. Later evergreen duration could fund repair but also escalation. Public-board filings do not reveal original venture rights. vehicle context
• When to use — Use when founder control or a specific executive/operating gap is material and the core thesis remains plausible. Stop when the gap is unfillable within runway, the founder will not consent, rights are unusable, or a generic team-risk label masks a broken market. [researcher inference · moderate]
• How to apply it — Map ownership/votes/information; define the missing role and success evidence; obtain consent; assign search owner, interim control, budget, deadline, financing condition, and stop trigger; measure operating change without claiming operator execution. [researcher inference · moderate]
• Examples —
  • MMI — documented adverse origin: founder control allegedly left Sutter Hill unable to correct behavior despite pedigree; original terms/outcome remain absent. MMI case
  • Xidex — documented application: Wythes described recruiting Les Colbert and supporting a bridge amid legal/operating stress; causality and economics are unproved. Xidex case
  • Qume — documented application: Wythes described recruiting Bob Schroeder; Comstock independently supports the founder lineage, not the full rationale. Comstock note
  • Cohesive — documented-but-thin application: Wythes reported board involvement, CEO problems, and merger/recruiting work; entry and outcome economics are incomplete. ledger
  • Cisco — counterexample: Wythes treated management as the reason for a pass and later framed failure to recruit as the miss; feasibility is unknown. Apple/Cisco lab
• Evidence — Wythes supplies the repeated management/governance pattern; Comstock and Younger independently support board/recruiting mechanisms; MMI and Cisco constrain efficacy. No minutes, contracts, or matched outcomes prove the interventions caused returns. talks attribution
• Limits, failure modes, and exception ledger — Investors can overreach, hire too late, alienate founders, or confuse replacement with progress. Route results: oral histories, 2003 colleague reporting, Teltone proxy, firm bio, issuer filings, and litigation routes were processed; no MMI terms, Xidex authority, job specs, search logs, candidates, bridge votes, conditions, post-hire metrics, or outcome counterfactuals were located. Teltone supplies adverse process, not misconduct or model application. filing controls
• Evolution and contradictions — MMI→Xidex/Qume suggests learning; Cisco shows diagnosis without execution; Teltone shows long board involvement is not automatically good governance. No dated checklist establishes the threshold. [researcher inference · provisional]
• Modern VC translation (as of 2026-08-21) — [researcher inference · moderate] Transfer rights-to-intervention mapping, role specificity, consent, deadlines, and stops. Do not assume every gap is repairable. Signals are usable rights and credible candidate supply; likely misuse is perpetual team-upgrade rationalization. A conditional application is a recruit/replace/condition/stop annex to IC; no current recruiting case was researched.

Syndicate when capital, expertise, or governance exceeds one vehicle

• Rule — [retrospective witness · moderate] When one vehicle cannot fund or govern adequately, recruit complementary investors and explicitly allocate capital, diligence, ownership, board responsibility, reserves, and exit authority. Syndication expands capacity; it does not pool credit. Measurex lab
• Evidence status — Bossen independently supports Wythes's Measurex syndication help; Wythes supplies Kasper and Quantum; Comstock supports shared Diablo financing; MMI is the weak-structure counterexample. Case/source breadth clears the moderate bar, while documents and economics remain absent. Bossen note · Comstock note
• Provenance and origin — The practice emerged under the reported $250,000 SBIC check constraint and a thin local capital market. No authored policy or contemporaneous allocation method survives. historical context
• Mechanism and reasoning — [researcher inference · moderate] Distinct investors can add capital, diligence, expertise, reserves, and board coverage; explicit boundaries reduce free riding, duplicate credit, unclear follow-on duty, and conflict.
• Historical and vehicle context — Measurex's approximately $1.25m total startup round comprised $250,000 from Sutter Hill/SBIC plus two outside blocks of about $500,000 each. The round is not Sutter Hill capital. Later evergreen scale changed the constraint, not the attribution rule. vehicle chronology
• When to use — Use when a named capital, reserve, technical, customer, geographic, regulatory, or governance gap exists and a participant closes it. Avoid prestige accumulation and ownerless decisions. [researcher inference · moderate]
• How to apply it — State the constraint; assign lead, check/security/ownership, diligence module, vote/board, information rights, conflicts, pro rata/reserves, and exit authority; stop if complementary value or accountability cannot be named; maintain a source-to-cash-flow ledger. [researcher inference · moderate]
• Examples —
  • Measurex — documented application: Wythes helped assemble capital beyond the Sutter/SBIC block; Bossen independently supports the work. Bossen oral history, printed pp. 18–21/PDF pp. 24–27
  • Kasper — documented application: Wythes remembered bringing Page Mill and John Young into distinct investor/board roles. Kasper case
  • Quantum — documented-but-incomplete application: Wythes described a multi-party allocation and proxy representation; subscriptions are missing. Priam/Quantum
  • Diablo — documented participant control: Comstock places Itel and Sutter Hill actors in financing; exact allocations are absent. Comstock note
  • MMI — counterexample: a lone minority structure allegedly left weak leverage. MMI case
• Evidence — Bossen plus Wythes support Measurex; Comstock independently supports Diablo; Kasper/Quantum extend the pattern; MMI supplies adversity. They prove participation patterns, not approvals, reserve duty, or outcomes. people/capital map
• Limits, failure modes, and exception ledger — Syndicates can fragment diligence, accountability, information, reserves, and exit timing. Route results: Bossen/Wythes/Johnson/Edwards records, issuer/SEC routes, firm and participant archives were processed; the 41 cases retain role boundaries; no subscription agreements, cap tables, side letters, board allocations, pro rata schedules, or distributions were found. uncertainty ledger
• Evolution and contradictions — Early capital constraint is visible; an evergreen pool could reduce dependence, yet later syndication still supplied expertise and governance. No dated selection policy shows the change. [researcher inference · provisional]
• Modern VC translation (as of 2026-08-21) — [researcher inference · moderate] Transfer constraint-first design and role/ownership ledgers. Do not assume a large round or famous party improves quality. Signals are a named gap and complementary resource; likely misuse is social proof/accountability diffusion. A conditional application is a responsibility/reserve matrix attached to the decision; no current syndicate was researched.

Use consent, recusal, and information separation for competitor conflicts

• Rule — [investor-stated · provisional] Before backing an overlapping company, disclose the conflict, obtain informed consent where required, recuse the conflicted partner, assign unconflicted representation, separate information, and decline if the boundary cannot be enforced. Priam/Quantum case
• Evidence status — Priam/Quantum is the sole narrated designed application. Diablo/Qume is an independently supported adverse overlap; Durrum/Teltone are different disclosure/process controls, not additional competitor applications. No underlying policy or effectiveness proof survives. Comstock note
• Provenance and origin — Wythes described Priam/Quantum after the Diablo/Qume lineage. Chronology suggests learning, but no contemporaneous source explicitly connects them as a process change. [researcher inference · provisional] talks chronology
• Mechanism and reasoning — [researcher inference · provisional] Consent surfaces loyalty conflict; recusal/proxy representation separate decisions; barriers reduce leakage; refusal preserves trust when structural conflict is non-remediable. Informal acquiescence or cosmetic proxies can fail.
• Historical and vehicle context — Small partnership teams and repeated storage-company networks made personnel, market, and board overlap likely. This historical procedure is not current legal advice. historical context
• When to use — Use when companies overlap in products, customers, employees, data, IP, strategy, or board duties. Obtain deal-specific legal advice; decline when duty or confidentiality cannot be separated. [researcher inference · provisional]
• How to apply it — Map overlap/duties; request legal and company evidence; obtain written consent; recuse conflicted decision makers; assign independent representation; restrict data/access; audit escalation; stop if residual conflict is not governable. [researcher inference · provisional]
• Examples —
  • Priam/Quantum — documented application: Wythes said Priam's CEO was consulted, Sutter Hill avoided the Quantum board seat, and Schroeder served as proxy; documents are absent. case
  • Diablo/Qume — counterexample/adverse boundary: repeat-founder lineage created overlap with no surviving consent or barrier record. lab
  • Durrum — researcher-applied disclosure illustration: a federal conflict application proves a disclosed affiliated setting, not a competitor protocol or wrongdoing. SEC News Digest, 1974-02-26
• Evidence — Wythes proves a later account of the Priam/Quantum design; Comstock independently supports Diablo→Qume; the Durrum filing supplies a separate disclosure control. The record does not prove consent quality, recusal, containment, or outcome effect. filing controls
• Limits, failure modes, and exception ledger — Consent can be uninformed; proxies can remain controlled; market boundaries move. Route results: Wythes/Comstock transcripts, Durrum/Teltone filings, and Micrex/Memorex plus Diablo/Qume litigation routes were exhausted publicly; no consent letters, recusal records, board minutes, IP opinions, barrier audits, or breach files surfaced. Below three direct applications, the model remains provisional. retrieval ledger
• Evolution and contradictions — Priam/Quantum may respond to Diablo/Qume, but no causal document proves evolution. Durrum is different in kind and cannot inflate repetition. [researcher inference · provisional]
• Modern VC translation (as of 2026-08-21) — [researcher inference · provisional] Transfer overlap mapping, written consent, recusal, barriers, independent representation, and refusal criteria. Do not infer present legal sufficiency. Signals are overlapping customers/data/IP/boards; likely misuse is cosmetic proxying. A conditional application is an independently audited conflict memo; no current legal/fund-policy corpus was researched.

Build companies on a multi-year clock, with explicit stopping tests

• Rule — [researcher inference · moderate] Underwrite venture as building an operating company over years rather than closing a financing; assign board, recruiting, and follow-on work, while defining milestones that distinguish productive patience from delayed failure. Darden analysis
• Evidence status — Wythes is the rule-bearing source and Darden is derivative. Younger independently supports company-building/recruiting work, Comstock board monitoring, and Newhall ongoing difficult-deal involvement. Xidex, Qume, and Cohesive supply breadth; countercases prevent efficacy inference. Newhall note · Almanac profile
• Provenance and origin — The early long-partnership pitch dates behavior to 1964; public articulation survives in one 2006 work and its 2011 derivative arrangement. Publication date is not evolution. Wythes note
• Mechanism and reasoning — [researcher inference · moderate] A multi-year frame makes recruiting, product iteration, financing, and governance part of underwriting. Explicit milestones, reserves, opportunity cost, and stop/merge/sell dates prevent patience from becoming escalation.
• Historical and vehicle context — Early technical companies faced scarce management/capital and long product cycles; evergreen capacity may have helped, but this company-work model is distinct from the vehicle agreement. No duration distribution or reserve policy survives. portfolio limits
• When to use — Use when company milestones genuinely require long work and the vehicle can support downside cases. Do not call delay long term without a falsifier, reserve budget, governance owner, and review date. [researcher inference · moderate]
• How to apply it — Write decision-date milestones, owners, board cadence, reserve/follow-on requests, failure signals, and a stop/merge/sell review; update after evidence; stop when the core thesis fails or repair exceeds time/capital budget. [researcher inference · moderate]
• Examples —
  • Xidex — documented application: management replacement, bridge, later financing, IPO, and sale form a long arc; causal vehicle/economic proof is absent. Xidex case
  • Qume — documented application: executive recruitment and company building appear in Wythes/Comstock accounts; role/economics remain shared. Comstock note
  • Cohesive — documented-but-thin application: CEO problem, recruiting, and merger show prolonged work; no complete cost/outcome chain. ledger
  • Duplicon and Priam — counterexamples: weak outcome and prolonged adversity show why duration requires stops; no documented Wythes stopping rule follows. failure ledger
• Evidence — Wythes supplies the philosophy and cases; Darden adds no independence; Younger, Comstock, and Newhall independently support operating involvement. The record does not prove time horizon, reserve discipline, or performance effect. writings playbook
• Limits, failure modes, and exception ledger — Patience can create escalation, stale values, weak governance, and successor burden. Route results: Wythes/Darden, Younger/Comstock/Newhall, issuer filings, and adverse cases were processed; no board calendars, reserve schedules, milestone votes, write-offs, duration distributions, or stop decisions surfaced. Duplicon does not create a hardware rule. uncertainty ledger
• Evolution and contradictions — Later narration presents stable patience, but no dated process change follows MMI, Priam, Duplicon, Cohesive, or Cisco. One appearance cannot establish intellectual chronology. [researcher inference · strong] talks evolution
• Modern VC translation (as of 2026-08-21) — [researcher inference · moderate] Transfer staged operating milestones, owners, reserves, falsifiers, and stop/merge/sell dates. Do not transfer a universal hold period. Signals are long validation and measurable work; likely misuse is rationalizing delay. A conditional application is a quarterly continue/repair/exit memo; no current company was researched.

Match vehicle duration and incentives to the work

• Rule — [researcher inference · provisional] Use a continuing pool only when the actual agreement aligns duration, commitments/recycling, reserves, fees/carry, valuation, withdrawals, co-investment, LP fairness, succession, and liquidity with the company-building work. vehicle chronology
• Evidence status — Wythes and the Draper witness family separately describe the Genstar-backed evergreen pool; they are participants, not governing documents. Organizational cases exist, but no company-level causal or cash-flow test clears the primary-document bar. Draper evidence guide
• Provenance and origin — The event is the distinct 1970 partnership after the circa-1969 Genstar/company transaction. The reported 80/20 then 75/25 split and later inconsistent characterization of the 36% figure as carried interest remain unreconciled. return controls
• Mechanism and reasoning — [researcher inference · provisional] Compatible duration and reserves can reduce premature fundraising/liquidity; aligned economics and succession can preserve decisions. The same structure can increase valuation opacity, fee duration, withdrawal conflict, and cross-generation allocation problems.
• Historical and vehicle context — The 1962 SBIC, 1965 acquired positions, circa-1969 transaction, 1970 LP, 1975 license surrender, 1986–1991 LP transition, and 1999 retirement are legally/analytically distinct. Personal/trust/TOW and later firm holdings are separate. profile chronology
• When to use — Use when the actual governing documents—not a label—support the work, downside reserves, valuation, LP liquidity, and succession. Do not assume evergreen means patient or aligned. [researcher inference · provisional]
• How to apply it — Obtain the agreement/amendments; map commitments, recycling, reserves, fees/carry, valuation, withdrawals/transfers, cross-fund/co-investment, key-person/succession, governance, and liquidity; stress-test delayed exit and partner retirement; stop if the mandate cannot support the strategy. [researcher inference · provisional]
• Examples —
  • 1970 Genstar pool — documented organizational case: continuing capital/discretion are participant-reported; agreement and accounts absent. formation
  • 1986–1991 LP transition — documented organizational case: additional institutions are manager-reported; commitments and confirmations absent. profile
  • Post-1999 special LP — documented organizational case: retirement/continuity are reported, while exact rights and economics are missing. profile
  • Shutterfly — researcher-applied boundary: firm ownership postdates active retirement; it does not prove vehicle mechanism or Wythes action. ownership controls
• Evidence — Wythes/Draper support structure; the Federal Register fixes the SBIC endpoint; filings bound later holder/retirement cases. No agreement, capital account, LP confirmation, or company cash flow proves alignment or performance. Federal Register, 1975
• Limits, failure modes, and exception ledger — Evergreen can create stale marks, fee duration, withdrawal, cross-generation, and key-person problems. Route results: Wythes/Draper histories, Federal Register, Genstar history, current/archived firm pages, PAW, external context, and roundtable derivative were checked; no partnership agreements/amendments, capital accounts, annual reports, valuations, reserves, withdrawals, distributions, or LP confirmations surfaced. The roundtable body is unavailable and supplies no doctrine. writings incentive limit
• Evolution and contradictions — Vehicle layers changed materially, but decision thresholds did not survive. Long duration can conflict with absent stops and successor fairness; post-retirement holdings cannot be back-attributed. [researcher inference · strong]
• Modern VC translation (as of 2026-08-21) — [researcher inference · provisional] Transfer document-level duration/incentive stress tests. Do not transfer the evergreen label. Signals are long validation, reserve capacity, and enforceable succession/liquidity terms; likely misuse is assuming alignment. A conditional application is a downside schedule joining company milestones to fund cash/LP obligations; no current fund terms were researched.

Build complementary investor teams and keep role credit explicit

• Rule — [researcher inference · moderate] Select partners with complementary judgment and backgrounds, deliberate through explicit decision rights, and ledger source, sponsor, diligence, approval, negotiation, board/recruiting, follow-on, exit, and narration separately. Shared credit is not personal alpha. people map
• Evidence status — Wythes and the Draper witness family support complementarity; Bossen, Poduska, Linear founders, Comstock, and Newhall independently allocate or constrain roles. Case/source breadth clears the moderate bar, while votes/carry and performance effect remain absent. talks attribution
• Provenance and origin — The Wythes/Draper pairing begins in 1965; later Anderson, Baker, Younger, and Coxe additions extend succession. The rule survives retrospectively, not in an authored organization design. Draper note · Newhall note
• Mechanism and reasoning — [researcher inference · moderate] Complementarity broadens networks/judgment; role ledgers reduce gaps and duplicate credit; succession preserves capability. Vague team rhetoric can conceal free riding, dissent, and responsibility.
• Historical and vehicle context — Firm identity spans a pre-existing SBIC, acquired positions, Wythes/Draper, later partners/LPs, 1999 retirement, special-LP status, and a continuing firm. Authority and incentives changed across layers. profile chronology
• When to use — Use whenever multiple partners, syndicates, vehicles, founders, boards, or successors contribute. It is essential for firm histories and group-holder filings. [researcher inference · moderate]
• How to apply it — Assign the eight roles, decision owner/dissent, vehicle, and capital before approval; update after boards/follow-ons/exits; request votes and allocations; keep founder/operator, trust/TOW, and continuing-firm layers separate; stop sole-credit claims when evidence is group-level. [researcher inference · moderate]
• Examples —
  • Wythes/Draper — documented organizational application: participant families describe complementary work; exact votes/economics are absent. talks guide
  • Measurex — documented shared work: Bossen operated; Wythes helped syndicate; three capital blocks remained distinct. Measurex lab
  • Apollo — negative attribution control: founder evidence assigns the initiating Sutter Hill action to Draper. Apollo note
  • Linear — negative attribution control: founders identify initial funders and later Sutter Hill interest; no Wythes role. Linear note
  • Shutterfly — succession boundary: firm ownership postdates active retirement. ledger
• Evidence — Wythes/Draper support team framing; Bossen, Poduska, Linear, Comstock, and Newhall are distinct participant controls; filings bound holders/retirement. They prove attribution must be granular, not that team design caused returns. responsibility grid
• Limits, failure modes, and exception ledger — Shared credit can obscure decision rights, carry, dissent, and poor work; later holdings can be pulled backward. Route results: Wythes, Draper, Johnson, Bossen, Poduska, Linear, Comstock, Newhall, corporate bio, and filings were processed; no IC votes, work logs, compensation/carry allocations, board-assignment policy, performance comparison, or successor authority surfaced. talks source guide
• Evolution and contradictions — The 1981 Draper exit, later hires, LP transition, and 1999 retirement changed authority even if late team rhetoric stayed stable. Team credit cannot erase dates. [researcher inference · strong]
• Modern VC translation (as of 2026-08-21) — [researcher inference · moderate] Transfer capability mapping, decision rights, dissent, role ledgers, and succession. Do not transfer vague collective credit. Signals are multiple contributors/key-person dependence; likely misuse is socializing wins while hiding decisions. A conditional application is an eight-role ledger tied to IC and allocation; no current partnership was researched.

Evidence gaps and candidate models

The inventory contains four moderate models—governable management gaps, syndication, multi-year company building, and complementary investor teams—and four provisional models—reciprocal sourcing, expertise-boundary diligence, competitor conflict protocol, and vehicle fit. They are decision-sequenced, not ranked. Local canvassing/reciprocity, governance/management repair, patience/company building, and team/role credit were merged; company-work duration and vehicle architecture remain separate because their causal decisions differ.

• Price discipline / objection taxonomy candidate — [researcher inference · insufficient public record]. Wythes's Apple and Webvan accounts are two retrospective price anecdotes, while Cisco is a people counterfactual, all within one oral-history work. Issuer records independently bound later outcomes but do not establish a Wythes rule or application. No decision-date memo, offer, terms, dissent, fatal-model case, recruitment file, reversal rule, or follow-up survives. The Wythes oral history, issuer/filing/press routes, firm and source archives, and the Done Deals, Mercury, NYSE, and roundtable writing routes were exhausted without independent rule/application proof. Preserve the three-way price/fatal/remediable classification only as an analyst-created IC exercise linked to the Apple/Cisco lab, not as a Wythes mental model.

• Duplicon hardware reliability/stage gates: one weak-outcome story; missing validation plan, kill criteria, proven write-off, and later process change. Duplicon lab

• Failure as tuition / generic acceptance: no dated postmortem, revised checklist, or repeated procedure. failure ledger

• Public-good company formation: aspiration, not an operational selection rule or measured outcome. Darden note

• Public-stock distribution timing: one reported practice; missing authorization, cases, tax/liquidity constraints, distributions, and outcome comparison.

• Biovest/India/China allocation or basket rule: incomplete manager authority, roster, weights, governance, and economics; Neurex, BioSurface Technology, and Ontogeny remain outside 41. reconciliation

• Parallel-fund/management-fee critique: original 1995 roundtable body unavailable; derivative fragments cannot establish context/mechanism. roundtable note

• LP-result confidentiality: Mercury title, date, format, body, and reasoning unlocated. Mercury note

• Legal/IP diligence for spinouts: one Xidex narrative and missing dispute record; retained only as management/governance context.

• Avoid majority-founder/single-investor structures: MMI-specific claimed change, merged into management and syndication rather than duplicated.

• Invest alongside LPs as universal alignment: reported mechanism without governing documents; retained only as vehicle/team candidate evidence.

• Formal price ceiling, ownership target, security preference, reserve formula, stop-loss, or concentration formula: no accessible memo, agreement, or complete set establishes thresholds. deal mechanics

• Princeton/Wythes Committee planning: collective university governance, not Wythes investment prose or doctrine. committee note

• LSI Logic or celebrated-company model: LSI remains orphaned; Apollo and Linear are negative attribution controls. ledger

Evidence that could change the inventory: original memos/votes and pass logs; portfolio/board/write-off schedules; reference files; terms/cap tables; board/recruiting/conflict records; partnership agreements, reserves, capital accounts, carry allocations, and distributions; lawful bodies for Done Deals, Mercury, NYSE, and the roundtable; and matched ordinary outcomes. Until then, all eight are bounded by 41 = 35 associated + five passes + one unnamed loss and 0/35 complete chains.

What the evidence supports

The defensible favorable thesis is that Wythes helped turn an existing Palo Alto SBIC into a durable, team-based venture institution: cheap local contact generated opportunities; syndication extended a small vehicle; boards and recruiting addressed governable people gaps; and continuing capital supported multi-year company work. [researcher inference · moderate]. It is not a founder-origin story, a proved personal-selection edge, or an audited alpha claim. Peterson and Lodato formed Sutter Hill Company in 1961; its Sutter Hill Capital Corporation subsidiary obtained the SBIC license in 1962. Wythes joined in 1964, Draper arrived in 1965, and the distinct limited partnership began in 1970. formation chronology · Wythes oral history, printed pp. 39–48 / PDF pp. 43–52

The strongest visible company-level evidence is the joint Kasper/Measurex sentinel, which Wythes remembered as roughly $8 million of a reported roughly $10 million Sutter Hill Company transaction valuation. That estimate is inseparable across two companies, retrospective, and unsupported by transaction or cash-flow schedules. Measurex is nevertheless unusually useful because Wythes's exact three-block financing account—$250,000 from Sutter Hill plus two outside blocks of about $500,000—is independently bounded by founder Dave Bossen's account of roughly $1.3 million total and Wythes's syndication help. [investor-stated · provisional] for the transaction-valuation estimate; [retrospective witness · moderate] for the shared financing work. Wythes, printed pp. 62–64 / PDF pp. 66–68 · Bossen, printed pp. 18–21 / PDF pp. 24–27

Exactly four cases qualify as no-hindsight decision labs: Measurex, Diablo/Qume, Apple/Cisco, and Duplicon. MMI, Kasper, Xidex, and Priam/Quantum remain bounded evidence summaries because the record cannot freeze enough decision-date alternatives, terms, milestones, follow-ons, or stops. labs · bounded summaries The eight formal mental models are procedures bounded by source and exception ledgers, not maxims proven to cause returns. model inventory

What could falsify the thesis

The favorable thesis predicts that original sourcing logs, memoranda, votes, reference files, board minutes, recruiting records, partnership documents, and cash flows would repeatedly show Wythes-specific actions: low-cost contacts producing differentiated evidence; co-investors closing named capital or expertise gaps; board rights enabling timely management repair; and company/vehicle duration matching measurable work. It also predicts that ordinary cases—not only Kasper and Measurex—would compare favorably after controlling for vehicle, partner, stage, sector, vintage, syndicate, and founder quality. [researcher inference · moderate]. thesis and predictions

The thesis would fail or materially weaken if those records assign celebrated decisions to Draper, founders, other partners, or syndicates; show locality without information advantage; show syndication without incremental capability; show ceremonial boards or late intervention; reveal that evergreen terms did not alter reserves or holding decisions; or show ex-sentinel cash flows at or below matched alternatives. It would also weaken if the eight procedures appear only in the 2006 retrospective narrative rather than in decision-date records. skeptical case

The strongest alternative currently fits at least as well: Wythes was a capable steward and narrator inside an advantaged pre-existing platform, while local access, founders, Draper and later partners, Genstar/LP capital, syndicates, survivor selection, era effects, and two early outliers explain the visible record. [researcher inference · strong]. The public record cannot discriminate because none of 35 associated companies has a complete vehicle-to-cash-flow chain. denominator

Attribution and denominator audit

The final opportunity union is exactly 41 = 35 associated companies + five remembered passes + one unnamed mining loss. “Associated” can mean inherited position, sourcing, syndication, board work, firm/group ownership, later-firm ownership, or narration; it does not mean Wythes personally sourced, approved, funded, owned, governed, or profited. All 35 associated cases lack at least one decisive field, so the complete-chain fraction is 0/35. [researcher inference · strong]. reconciliation The oral history's 46 events and 31-event subtotal are activity counts, not unique companies or a portfolio denominator. Wythes, printed pp. 62–64 / PDF pp. 66–68

The source map reconciles 62 rows = 27 ★★★ + 31 ★★ + 4 ★. Its 58 priority rows route once into seven writing families, sixteen spoken-record families, seventeen transaction/legal/denominator families, and eighteen biography/context families. Seven writing notes and sixteen talk notes exist. The strict Wythes-attributable writing denominator is three, with zero full bodies accessible and zero deeply analyzed; the one three-session Wythes oral history has two institutional manifestations, not two independent interviews. source accounting · written accounting · spoken accounting

Attribution remains role-specific: source, sponsor, diligence, approval, negotiation, board/recruiting, follow-on, exit, and later narration are separate. Firm, fund, SBIC, partner, founder, trust, TOW, and individual holdings are not interchangeable. Bossen supports Wythes's Measurex syndication help; Poduska assigns Apollo's initiating Sutter Hill action to Draper; Linear's founders do not name Wythes; and post-1999 firm ownership cannot be pulled backward into his active record. responsibility grid

Transfer limits for modern VCs

The eight models transfer as auditable questions: log source cost and new information; use independent experts within explicit competence boundaries; map rights before promising management repair; name each syndicate member's contribution; obtain written conflict controls; join patience to dated stopping tests; inspect actual vehicle terms; and ledger team credit. They do not transfer as evidence that locality, evergreen capital, board tenure, or larger syndicates cause returns. eight-model inventory

Every modern translation is dated 2026-08-21 and is a [researcher inference · provisional] because T011 did not independently research a current company, market, fund agreement, or regulatory application. Changed capital supply, competition, information costs, technical specialization, securities law, governance practice, and fund structures can break the historical mechanism. modern matrix

The former price/fatal/remediable taxonomy is not a ninth Wythes model. Apple and Webvan are two price anecdotes and Cisco is a people counterfactual within one retrospective work; issuer records independently bound later outcomes, not the rule or its application. It survives only as an analyst-created IC exercise linked to the Apple/Cisco lab. [researcher inference · insufficient public record]. candidate-model boundary

Unresolved evidence

The decisive missing record is company- and vehicle-level: original opportunity logs; investment memoranda and votes; rejected-deal files; reference reports; offers and final securities; cap tables; board minutes; recruiting and conflict records; follow-on/reserve decisions; exits, write-offs, distributions, and partner capital accounts. These would resolve repeated process, decision rights, personal economics, and ex-outlier performance. retrieval ledger

The three strict writing bodies remain unavailable: Wythes's edited Done Deals chapter is lawful only through controlled borrowing and was not read; the Mercury letter/editorial and 1995 NYSE remarks remain unlocated. The CHM archive has an item-level route for audio/permissions, but no standalone audio was retrieved; transcript use remains paraphrase-only absent written permission. writing routes · audio/archive attempts

The most material case-specific gaps are: Kasper/Measurex cash flows and allocation; MMI terms and proceeds; Xidex authority, dispute record, and bridge conditions; Priam/Quantum consent and information-barrier documents; Duplicon validation and proceeds; Apple/Cisco offers and pass memos; and Amerigroup's direct SEC exhibit match. The reproduced Amerigroup agreement supports Schedule 8.1.4/Attachment A names and aggregate class totals, but not holder-level share counts or personal return. Amerigroup ledger row

Corrections recorded from independent audits

• Corrected every official 163-page Wythes transcript locator to the uniform substantive-page mapping physical PDF = printed + 4, with the unnumbered title at physical p. 4 and session starts at physical pp. 5, 56, and 99. This includes the central $8m/$10m, ~36% IRR, and carry locators.
• Corrected the carry conflict to printed p. 151 / physical PDF p. 155 and preserved it as unresolved speech/transcription/economic-term ambiguity rather than fact or a confident typo.
• Corrected Measurex's exact three-block arithmetic to printed pp. 62–63 / PDF pp. 66–67 and restricted Bossen's independent support to the approximate total and syndication role.
• Reclassified MMI, Kasper, Xidex, and Priam/Quantum under Bounded evidence summaries — not ex-ante reconstructions; preserved exactly four formal labs, worksheets, and separate reveals.
• Demoted the former price/fatal/remediable Model 8 to an analyst-created candidate. The final formal inventory is eight models × twelve required fields = 96 fields, four moderate and four provisional.
• Corrected the Amerigroup loan/security agreement date to 1999-11-09; localized the reproduced evidence to Schedule 8.1.4/Attachment A, pp. 142–146; graded it moderate pending direct SEC exhibit match; and preserved 1994 incorporation versus 1997 retrospective participation timing.
• Normalized material attribution/support labels to the literal §1 vocabulary and removed stale status language and the duplicated source-map phrase from reader-facing prose.

Reproducible central-claim audit trail

The following 23 claims were fixed before completion-state bookkeeping. “Pass” means the claim is reproducible at its stated boundary, not that the missing economics are supplied.

Independent attribution controls

These are independent only for the named facts. Filings that bound an issuer's later outcome do not independently establish a remembered Wythes decision rule, motive, or application.

Failure-resolution audit

The unnamed mining write-off remains in the 41-family denominator but cannot support an eighth postmortem because identity, date, vehicle, amount, thesis, governance, and learning record are all absent.

Reproducible ordinary-claim audit trail

The fixed sample contains exactly sixteen ordinary claims—two from each of the eight core documents, including this review—and tests reproducibility outside the central-claim set.

Mechanical package reconciliation

• Sources/notes: 62 sources = 27 ★★★ + 31 ★★ + 4 ★; 58 priority routes; seven writing notes and sixteen talk notes.
• Portfolio: 41 rows = 35 associated + five passes + one unnamed loss; 0/35 complete chains; four full labs and four bounded summaries.
• Synthesis: seventeen philosophy H2 sections; eight formal mental models × twelve labeled bullets = 96 fields; four moderate and four provisional; one explicitly excluded price/taxonomy candidate.
• Citations: the nine-file source corpus (excluding this review) contains 120 official Wythes-PDF links across 108 lines—117 with fragments and three intentionally bare. With fourteen claim-local links across thirteen lines in this review, the final package contains 134 links across 121 lines—131 fragmented and three bare. No duplicate fragment suffix, target above p. 163, printed/PDF mismatch, or first-page fragment mismatch remains.
• Review controls: 23 high-risk rows, 16 ordinary-claim rows across all eight core documents, nine independent attribution controls, seven failure-resolution rows, and a final 20/20 acceptance matrix.
• Rights/hygiene: source notes remain paraphrase-led; no full transcript, bulk excerpt, image, placeholder, or temporary artifact was added.

Twenty-point acceptance matrix

All twenty conditions pass after the corrections above. Acceptance is limited to the bounded public record; it does not turn missing archives, cash flows, decision rights, or personal economics into evidence.

Direct voice and writings

• ★★★ | Paul Wythes: Venture Capital Greats oral history | https://archive.computerhistory.org/resources/access/text/2019/03/102781073-05-01-acc.pdf | sessions 2006-05-08, 2006-05-10, and 2006-07-10; public transcript 2019 | full 163-page official CHM/NVCA transcript, ref. X8628.2008; reliability A for custody/attributed speech and B for memory; research access open, quotation/publication permissions restricted | One spoken work with two institutional manifestations, not two interviews: CHM supplies stable locators; Bancroft/UC Berkeley supplies the 177-PDF-page donated-transcript route at https://digicoll.lib.berkeley.edu/record/103633/files/wythes_paul_donated.pdf. Core direct spine: formation pp. 39–48, Kasper/Measurex/Genstar pp. 52–76, Xidex/Diablo pp. 84–93, later companies pp. 104–135, retirement pp. 136–143, economics/team pp. 127–131 and 149–151, and misses/write-off pp. 152–159; printed CHM pages run four behind physical PDF pages. Public-repo use is paraphrase only unless written permission is documented.
• ★★★ | Paul Wythes: Sutter Hill Ventures, in Done Deals | https://archive.org/details/donedealsventure00gupt/page/150/mode/2up | 2000 | Udayan Gupta edited first-person interview chapter beginning printed p. 151; reliability B; controlled-digital-lending/account login may be required for the full body | Mandatory written-corpus route on formation, Genstar, portfolio companies, partnership practice, and stated principles; edited recollection, not contemporaneous or audited evidence.
• ★★★ | For lucky alumni, biotech investments pay big dividends | https://www.princeton.edu/~paw/archive_old/PAW99-00/10-0223/0223feat.html | 2000-02-23 | full Princeton Alumni Weekly feature by Marvin Zim with direct Wythes quotations; reliability B | Quoted-source control, not a Wythes-authored writing: contains reported: >35% average annual Sutter Hill return over the prior thirty years (Wythes's self-estimate) and comments on Biovest/biotech diligence; no cash-flow series or return definition.
• ★★ | What Ever Happened to Venture Capital? | https://www.darden.virginia.edu/sites/default/files/2025-09/VC_2011.pdf | 2011-02 | full Darden Batten Institute PDF by Sean D. Carr, Malgorzata Glinska, and Andrew King; edited research brief using attributed practitioner excerpts; reliability B/C for Wythes application | Derivative control, not Wythes-authored: republishes/adapts oral-history material on company building, long horizons, and useful products; trace each excerpt to the oral history and honor the transcript's paraphrase-only restriction.
• ★★★ | Venture Capital Roundtable | https://www.venturecapitaljournal.com/ | 1995-09 | Renee Deger edited roundtable in Venture Capital Journal, printed pp. 34 and 36; current official publisher route does not expose the original issue body; reliability B for the later academic citation trail | Edited-roundtable direct-voice lead, not an authored work: Florida and Kenney's derivative manuscript and its endnotes attest two reported Wythes fragments only; do not reconstruct the unavailable original from derivative quotations.
• ★★★ | San Jose Mercury News letter/editorial opposing public LP-result disclosure — body unlocated | https://digicoll.lib.berkeley.edu/record/103633/files/wythes_paul_donated.pdf#page=44 | circa 2002–2006; described in the 2006 interview, exact date unknown | attributable-writing lead only at donated-transcript PDF p. 44; title, letter-versus-editorial label, publication date, edition, and body unavailable | Strict attributable artifact: Wythes describes writing during the Mercury News litigation over University of California/CalPERS venture results; no title, prose, or publication details may be inferred from that summary.
• ★★★ | Princeton Class of 1955 award acceptance remarks at the New York Stock Exchange — body unlocated | https://digicoll.lib.berkeley.edu/record/103633/files/wythes_paul_donated.pdf#page=17 | award/dinner in 1995; described in the 2006 interview | attributable speech/manuscript lead; exact event date, title, recording, transcript, and manuscript unavailable | Strict attributable artifact. Donated-transcript locator clusters: award/dinner/talk PDF pp. 17–18; remembered manuscript and tie to Princeton/venture philosophy p. 112; public-good/jobs/products pp. 164–165; copy/CV lead p. 176. These are retrieval leads, not a substitute body or basis for inferred doctrine.

Participant, team, and counterparty oral histories

• ★★★ | William H. Draper III: Early Bay Area Venture Capitalists oral history | https://digicoll.lib.berkeley.edu/record/218454/files/draper_william.pdf | interviews 2008-04-07 through 2008-05-15; published 2009 | full 104-PDF-page Bancroft oral history; reliability A for custody and B for memory | Strong cross-check on D&J, the pre-existing Sutter Hill real-estate/SBIC structure, Wythes's prior hire, the 1965 transition, Genstar, shared team work, and Draper's 1981 departure.
• ★★★ | William H. Draper III: Venture Capital Greats oral history | https://digicoll.lib.berkeley.edu/record/103633/files/draper_william_donated.pdf | interviews 2005-10-24, 2005-11-02, and 2005-12-06; published 2009 | full official NVCA/Bancroft transcript; reliability A for custody and B for memory; quotation/publication restricted | Distinct interview work with the clearest explicit credit to Wythes, Draper, Baker, and Anderson as a four-person team and reported: roughly 40% Sutter Hill IRR by 1981 (Draper's recollection).
• ★★ | Oral History of William H. Draper III | https://archive.computerhistory.org/resources/access/text/2013/04/102743096-05-01-acc.pdf | recorded 2011-04-14 | full 26-page official CHM transcript, ref. X6084.2011; reliability A for custody and B for memory | Compact later cross-check on D&J/Sutter Hill chronology, Genstar, decision authority, and Apollo; overlapping testimony is not independent transaction proof.
• ★★ | William H. Draper III — HBS Entrepreneurs oral history | https://www.library.hbs.edu/special-collections-and-archives/exhibits/entrepreneurs/william-h-draper-iii | interview 2010-12 | official HBS exhibit with video/transcript route; reliability A for event custody and B for retrospective content | Direct counterparty discussion of Sutter Hill's evergreen arrangement, team selection, Genstar, and venture base rates; dynamic transcript assets may expire.
• ★★ | Bill Draper interview excerpt — National Museum of American History | https://americanhistory.si.edu/sites/default/files/SI-NMAH_Draper-Interview_Final.pdf | published 2015-08-19 | five-page edited Smithsonian/NMAH excerpt, not the complete interview; reliability A for custody and B for memory | A separate late Draper account useful for formation and family/firm disambiguation; the complete interview is unavailable, and repeated stories do not become multiple confirmations.
• ★★★ | David Bossen oral history | https://digicoll.lib.berkeley.edu/record/103597/files/bossen_dave.pdf | interview 2010-07-08; transcript 2011 | full Bancroft oral history; reliability A for custody and B for memory; literary rights reserved | Founder-side Measurex control: printed pp. 18–21 names Wythes as helpful in assembling a wider financing syndicate while preserving Bancroft, Draper, Reid and Peggy Dennis, and other contributors.
• ★★★ | William C. Edwards oral history | https://digicoll.lib.berkeley.edu/record/218456/files/edwards_william.pdf | interview 2009-05-12; published 2009 | full Bancroft oral history; reliability A for custody and B for memory; quotation requires written permission | Negative participant-context control: the body has no Wythes, Sutter Hill, or Measurex occurrence; its early-network and SBIC discussion cannot establish a person- or deal-level role.
• ★★★ | Franklin “Pitch” Johnson oral history | https://digicoll.lib.berkeley.edu/record/103604/files/johnson_franklin.pdf | interviews 2008-05-28, 2008-06-25, and 2008-07-07; published 2009 | full Bancroft/NVCA transcript; reliability A for custody and B for memory | Counterparty account of Draper & Johnson, SBIC origins, early syndication, NVCA, and the Sutter Hill transition; prevents a Wythes-only formation narrative.
• ★★ | Oral History of Franklin Pitch Johnson | https://archive.computerhistory.org/resources/access/text/2020/05/102795131-05-01-acc.pdf | recorded 2019-09-12; public transcript 2020 | full official CHM transcript; reliability A for custody and B for memory | Later Johnson cross-check on early firms and institutional history; use only after separating this interview from the Bancroft work.
• ★★★ | Oral History of George Comstock | https://archive.computerhistory.org/resources/access/text/2015/07/102658008-05-01-acc.pdf | recorded 2003-08-13; public transcript 2004 | full official CHM transcript; reliability A for custody and B for memory | Founder/operator control for Diablo and Qume; spoken pp. 35–41 support a Wythes board-monitoring episode and a firm-level competitor conflict, while the p. 48 gallery is editorial rather than Comstock testimony.
• ★★★ | Linear Technology Founders Panel | https://archive.computerhistory.org/resources/access/text/2014/07/102746888-05-01-acc.pdf | recorded 2014-03-31 | full 51-page official CHM panel transcript; reliability A for custody and B for participant memory | Printed pp. 14–15 name the four initial venture backers and place Sutter Hill among firms seeking entry afterward; no person-level Wythes role is identified.
• ★★ | Oral History of John William (Bill) Poduska Sr. | https://archive.computerhistory.org/resources/access/text/2012/06/102658256-05-01-acc.pdf | recorded 2002-10-09 | full official CHM founder oral-history transcript; reliability A for custody and B for memory | Printed p. 37 says Bill Draper committed the Sutter Hill lead and separately notes Wythes; blocks transfer of Apollo lead attribution to Wythes.
• ★★ | Charles W. Newhall III oral history | https://archive.computerhistory.org/resources/access/text/2018/11/102738556-05-01-acc.pdf | recorded 2017-09-26; public transcript 2018 | full official CHM/NVCA transcript; reliability A for custody and B for memory | Later-VC witness describing Wythes as a mentor/adviser to NEA's management committee; useful for influence, not a Sutter Hill return record.

Talks, recordings, and collection routes

• ★★ | NVCA Oral History Collection Unveiled at the Computer History Museum | https://computerhistory.org/press-releases/nvca-collection/ | 2019-03-28 | official CHM collection announcement and named-interview inventory; reliability A for custody/metadata | Confirms transfer of Wythes and peer transcripts/audio to CHM and supplies collection context; no standalone public Wythes audio or video object was located.
• ★★ | Computer History Museum access and research route | https://computerhistory.org/access-and-research/ | current page accessed 2026-08-21 | official reading-room and collection-request instructions; reliability A for access procedure | Request CHM ref. X8628.2008/asset 102781073 audio and permissions metadata; no standalone public Wythes audio object was located and no request was sent during the documented search.

Books, scholarship, and historical analysis

• ★★★ | The Startup Game: Inside the Partnership Between Venture Capitalists and Entrepreneurs | https://us.macmillan.com/books/9780230339941/thestartupgame/ | first published 2011; paperback 2012-06-19 | official publisher record for William H. Draper III's memoir; reliability A for metadata and B for retrospective content; body licensed | Participant account of Sutter Hill, Genstar, Measurex, and team practice; any return figure remains Draper's recollection unless supported by partnership ledgers.
• ★★ | VC: An American History | https://books.google.com/books/about/VC.html?id=uSmWDwAAQBAJ | 2019 | Tom Nicholas scholarly history; limited-preview book route; reliability B | Historical framing for SBICs, early partnerships, Sutter Hill, incentives, and industry evolution; inspect endnotes and original sources before using person-level claims.
• ★★ | Financiers of Innovation: The Development of Venture Capital in Silicon Valley | https://www.creativeclass.com/rfcgdb/articles/Financiers%20of%20Innovation.pdf | manuscript 1997 | full scholarly manuscript by Martin Kenney and Richard Florida; reliability B | Broad institutional/firm history that acknowledges and quotes a Wythes interview on competition; useful context, not an audit or complete portfolio.
• ★★ | Venture Capital-Financed Innovation and Technological Change in the US | https://ipmall.info/sites/default/files/BAYHDOLE/3_DIV_SCAN/3156_001_OCR_NEXT_0.pdf | 1987-08 | full scanned academic paper by Richard L. Florida and Martin Kenney; reliability B | Earlier historical source acknowledging a Wythes interview and mapping the regional venture system; the 1986 interview body remains unavailable, and scan OCR/endnotes must be checked before extracting claims.
• ★★ | Growth of the Multidivisional Corporation: The Genstar Case | https://www.cambridge.org/core/services/aop-cambridge-core/content/view/15D5F574D294DC62B531F814D9B4DEFD/S0007680500051527a.pdf/growth_of_the_multidivisional_corporation_the_genstar_case.pdf | 1985 | peer-reviewed Business History Review article; reliability B | Independent corporate-history context for Genstar and Sutter Hill's electronics-startup strategy; does not reveal partnership economics or Wythes allocation.
• ★★ | The Economics of the Private Equity Market | https://www.federalreserve.gov/pubs/staffstudies/168/ss168.pdf | 1995 | full Federal Reserve Staff Study 168; reliability A for its sourced institutional analysis | Base-rate control for SBICs, limited partnerships, governance, incentives, fundraising, and reported returns; not a Sutter Hill estimate.
• ★★ | 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 2026-08-21 | full peer-reviewed institutional history by Martin Kenney; reliability B | Context for SBIC outcomes, independent partnerships, policy change, and regional scaling; prevents crediting one investor for system-level developments.
• ★★ | We Have Met the Enemy and He Is Us | https://www.kauffman.org/reports/we-have-met-the-enemy-and-he-is-us/ | 2012-05-07 | official Kauffman report page with full PDF; reliability A for the foundation's defined portfolio and B for generalization | Adversarial base-rate control on unaudited VC narratives, confidentiality, cash-flow evidence, fees, persistence, and LP/GP alignment; not Sutter Hill-specific.
• ★★ | Why Critics Are Slamming an Increasingly Popular Private Equity Format | https://www.institutionalinvestor.com/article/2bsvwt4a7hdicrwt1cpa8/portfolio/why-critics-are-slamming-an-increasingly-popular-private-equity-format | 2017-08-14 | reported Institutional Investor analysis; reliability B; access may be metered | Skeptical route on evergreen/permanent-capital valuation and alignment, including Sutter Hill/Yale context; later criticism must not be projected backward without governing documents.
• ★ | A Radical Way to Build New Companies | https://www.forbes.com/sites/brettnelson/2012/08/01/a-radical-way-to-build-new-companies-not-just-web-startups-create-jobs-and-boost-the-economy/ | 2012-08-01 | Forbes explanatory article; reliability C for historical superlatives; access may be metered | Secondary route calling Sutter Hill an early evergreen model; use only to locate stronger partnership or LP records.

Press, biography, institutional legacy, and archives

• ★★ | Former Princeton trustee Paul Wythes dies | https://www.princeton.edu/news/2012/11/09/former-princeton-trustee-paul-wythes-dies | 2012-11-09 | official Princeton obituary; reliability A for university service and B for family/firm biography | Useful for education, death, trusteeship, awards, and the Wythes Report, but its 1962 “founded” date conflicts with direct formation evidence.
• ★★ | Paul M. Wythes ’55 memorial | https://paw.princeton.edu/memorial/paul-m-wythes-55 | 2013-02-06 | Princeton Alumni Weekly class memorial; reliability B | Supplies exact birth date, education, 1964-era entry phrasing, and a claimed 27-board denominator; roster and outcomes remain absent.
• ★★ | Paid obituary: Paul Morrison Wythes | https://www.almanacnews.com/wp-content/uploads/2012/11/2012_11_14.alm_.section1.pdf#page=18 | 2012-11-14 | full family-paid obituary, printed p. 20/PDF p. 18; reliability B for family biography and C for deal attribution | Best public full-name/family chronology, but “leading investments” and “48 years” are family assertions without transaction schedules.
• ★★ | Silicon Valley Nation: Venture capital pioneer Paul Wythes dies | https://www.eetimes.com/silicon-valley-nation-venture-capital-pioneer-paul-wythes-dies/ | 2012-11-07 | full trade-press obituary by Brian Fuller; reliability B | Reuses the door-to-door sourcing story and summarizes firm/company history; derivative of oral history and not independent deal evidence.
• ★★★ | Venture-capitalist Paul Wythes to get lifetime achievement award | https://www.almanacnews.com/morgue/2003/2003_05_28.wythes.html | 2003-05-28 | full local-news profile with named colleague quotation; reliability B | Bill Younger's contemporary description of Wythes's business-plan, recruiting, and IPO work and the 27-board claim provide a witness lead.
• ★★ | Paul Wythes — archived official Sutter Hill Ventures biography | https://index.commoncrawl.org/CC-MAIN-2012-index?url=www.shv.com%2Fteam%2Fwythes.html&output=json | snapshot timestamp 2012-02-04 07:38:12 UTC; origin Last-Modified 2010-09-08 | unsigned corporate firm page recovered through Common Crawl index/ARC; reliability A for the firm's then-published claims and B for attribution; original URL http://www.shv.com/team/wythes.html; ARC byte-range retrieval is not a simple browser route | Names Tellabs, Xidex, Linear Technology, and AmeriGroup as companies Wythes led and claims 40-plus venture years and 27-plus boards; not Wythes-authored, and firm-reported discovery evidence is not a complete denominator or audited performance record.
• ★★ | Princeton's Center on Contemporary China named in honor of Paul and Marcia Wythes | https://alumni.princeton.edu/stories/princetons-center-contemporary-china-named-honor-paul-and-marcia-wythes | 2017-10-27 | official Princeton alumni announcement; reliability A for institutional legacy | Posthumous philanthropy and university-role route; outside the venture-performance denominator.
• ★★ | Navy Supply Corps Distinguished Alumni Directory | https://navysupplycorpsfoundation.org/wp-content/uploads/2024/08/DISTINGUISHED-ALUMNI-DIRECTORY.AUG24.pdf | directory updated 2024-08 | official foundation directory, Wythes entry printed p. 19; reliability A for award/service metadata and B for career summary | Confirms Supply Corps affiliation and award route; company/board claims are compressed biographical statements, not primary deal records.
• ★★ | Princeton University Archives — Shirley M. Tilghman records | https://findingaids.library.upenn.edu/records/PRIN_MUDD_AC379 | collection dated 1988–2013; finding aid accessed 2026-08-21 | official-archive-network finding aid; reliability A for holdings metadata; request/on-site access required | Box-level route to Trustees—Paul Wythes files for 1988–1999 and the strategic committee; bodies were not inspected.
• ★★ | Wythes Committee plans for coming decade | https://www.princeton.edu/~paw/archive_old/PAW99-00/11-0308/0308notx.html | 2000-03-08 | full Princeton Alumni Weekly article by Maria LoBiondo summarizing the committee process and recommendations; reliability B | The article is not the 68-page collective committee report and neither is Wythes-authored. The report body remains unavailable; this contemporary control is not evidence about venture selection or returns.
• ★ | Sutter Hill Ventures current site | https://shv.com/ | current site accessed 2026-08-21 | official firm landing page; reliability A for current address only; no public history, team archive, portfolio, or document library | Negative archive result: the continuing firm's public site does not supply a Wythes-era denominator or confirm historical deal attribution.

Transactions, filings, legal controls, and denominator routes

• ★★★ | Sutter Hill Ventures EDGAR filer page, CIK 0000879051 | https://www.sec.gov/edgar/browse/?CIK=879051 | filings from the early 1990s onward; accessed 2026-08-21 | official SEC filing index; reliability A for filed group ownership and dates | Highest-yield public denominator route for later holdings, but it misses private/pre-EDGAR investments and group filings do not prove Wythes selection or economics.
• ★★★ | Durrum Industries affiliated transaction — SEC News Digest | https://www.sec.gov/news/digest/1974/dig022674.pdf | 1974-02-26 | official SEC News Digest, Durrum/IPC application item; reliability A for application disclosure | Identifies Wythes's Sutter Hill and Durrum roles and the affiliated-transaction route; an application/disclosure is not a violation finding.
• ★★★ | Sutter Hill Capital Corporation — surrender of SBIC license | https://www.govinfo.gov/content/pkg/FR-1975-09-12/pdf/FR-1975-09-12.pdf | 1975-09-12 | official Federal Register notice, printed p. 42412, License No. 09/12-0087; reliability A | Confirms license issuance on 1962-09-20 and SBA acceptance of surrender effective 1975-09-04; does not reveal portfolio, leverage, examination history, or returns.
• ★★★ | Teltone Corporation 2002 proxy statement | https://www.sec.gov/Archives/edgar/data/96890/000009689002000064/defproxy.htm | filed 2002-05-06 | official SEC proxy; reliability A | Establishes board service since 1976, 1999 retirement, reverse-split terms, and lack of an independent fairness assessment. Its 152,105-share voting total includes 75,983 TOW Partners shares under footnote 4 and is not wholly personal ownership; ordinary/adverse control without implying illegality.
• ★★★ | In re Activision Securities Litigation | https://law.justia.com/cases/federal/district-courts/FSupp/621/415/1368334/ | decided 1985-11-04 | full federal district-court opinion, 621 F. Supp. 415, especially pp. 418–424 and 438–439; reliability A for allegations, procedure, and holding | Names Wythes, Sutter Hill, and Genstar among VC selling-shareholder defendants; dismisses the Section 11 statutory-underwriter claim against that group with prejudice because a firm-commitment sale did not itself constitute distribution participation. This is not a merits finding of fraud, and rulings concerning other defendants must not be transferred to Wythes.
• ★★★ | AMERIGROUP 1999 loan and security agreement | https://contracts.justia.com/companies/amerigroup-corp-35296/contract/1086359/ | dated 1999-11-09 | reproduced issuer agreement; reliability B/moderate pending direct SEC exhibit reconciliation | Schedule 8.1.4/Attachment A, printed pp. 142–146, gives aggregate class totals and separately names Sutter Hill, TOW Partners, Wythes's living trust, founders and other holders; it does not give per-holder counts or prove that Wythes founded, led, or personally owned every grouped share.
• ★★★ | Amerigroup merger proxy | https://www.sec.gov/Archives/edgar/data/1064863/000119312512338614/d390909dprem14a.htm | filed 2012-08-02 | official SEC proxy; reliability A | Fixes December 1994 incorporation as AMERICAID, November 2001 IPO, and later corporate chronology; does not enumerate all early decisions or Wythes's role.
• ★★★ | Shutterfly S-1 | https://www.sec.gov/Archives/edgar/data/1125920/000095014906000362/f21300sv1.htm | filed 2006-06-23 | official SEC registration statement; reliability A | Shows Sutter Hill-affiliated 9.1% pre-offering ownership and governance context; Wythes had retired from active management, so person-level lead remains unproved.
• ★★ | Exogen Schedule 13G | https://www.sec.gov/Archives/edgar/data/934094/000091205797006148/0000912057-97-006148-index.html | filed 1997-02-19 | official SEC ownership filing index; reliability A | Separates Wythes's 8,684 shares from 438,671 aggregate group shares, including 379,250 held by Sutter Hill Ventures and 50,737 held by TOW; group ownership is not personal sourcing, capital, or realized-return evidence.
• ★★ | Lexitron SEC News Digest ownership record | https://www.sec.gov/news/digest/1976/dig102776.pdf | 1976-10-27 | official SEC News Digest; reliability A for filed ownership disclosure | Contemporaneous route to a Sutter Hill position absent from the oral-history headline set; candidate for denominator expansion and ordinary-outcome research.
• ★★ | Prime Computer secondary-sale notice | https://fraser.stlouisfed.org/docs/publications/cfc/cfc_19770922.pdf | 1977-09-22 | FRASER scan of Commerce Clearing House Federal Securities Law Reports; reliability B pending issuer/SEC filing retrieval | Routes a Sutter Hill sale of Prime Computer shares and exposes another portfolio candidate; allocation to Wythes personally is unproved.
• ★★★ | T. Rowe Price Associates exemption application | https://www.govinfo.gov/content/pkg/FR-1982-12-17/pdf/FR-1982-12-17.pdf | 1982-12-17 | official Federal Register notice; reliability A | Discloses Wythes's fund-director role and under-one-percent H&Q partnership interest in a conflict application; process record, not wrongdoing or venture performance.
• ★★★ | Migdal v. Rowe Price-Fleming International, Inc. | https://law.justia.com/cases/federal/appellate-courts/F3/248/321/550734/ | decided 2001-04-26 | full Fourth Circuit opinion, 248 F.3d 321; reliability A for procedural history and holding | Adversarial control naming Wythes among directors; claims were dismissed for failure to state a claim, so allegations must not be restated as findings.
• ★★ | Oacis Healthcare Holdings Schedule 13G | https://www.sec.gov/Archives/edgar/data/1011671/000091205797006217/0000912057-97-006217-index.htm | filed 1997-02-19 | official SEC ownership filing index; reliability A | Separates Wythes's 18,762 shares from 893,012 aggregate group shares, including 819,216 held by Sutter Hill Ventures and 55,034 held by TOW; proves a distinct issuer association but not sourcing, cost, economic entitlement, or realized return.
• ★ | Sutter Hill Ventures portfolio companies hub | https://www.crunchbase.com/hub/sutter-hill-ventures-portfolio-companies | current database accessed 2026-08-21 | commercial aggregator; partial/paywalled; reliability C | Discovery-only continuing-firm roster; cannot isolate Wythes's 1964–1999 active era, legal vehicle, decisions, passes, failures, or cash flows.
• ★ | Sutter Hill Ventures investment profile | https://www.cbinsights.com/investor/sutter-hill-ventures | current database accessed 2026-08-21 | commercial aggregator; paywalled/partial; reliability C | Cross-check route for company aliases and later rounds only; no person, vehicle, cost, or complete outcome ledger.

Statutory and regulatory context

• ★★★ | Small Business Investment Act of 1958 | https://www.govinfo.gov/content/pkg/COMPS-1835/pdf/COMPS-1835.pdf | enacted 1958; current compilation accessed 2026-08-21 | official statutory compilation; reliability A | Governing context for the Sutter Hill Capital SBIC; the statute cannot prove its 1962 license, leverage draw, portfolio, compliance, or return.
• ★★ | Small Business Investment Companies | https://www.sba.gov/for-partners/small-business-investment-companies/ | current page accessed 2026-08-21 | official SBA program page; reliability A for program structure | Explains that SBICs are privately managed, SBA-licensed funds with government-guaranteed leverage; modern summary must not be assumed identical to 1962 terms.
• ★★★ | Employee benefit plan investment prudence rule | https://www.govinfo.gov/content/pkg/FR-1979-06-26/pdf/FR-1979-06-26.pdf | 1979-06-26 | official Federal Register final rule and preamble; reliability A | Controls the timing and text of the post-ERISA prudent-person clarification; Wythes's NVCA association role does not establish personal authorship or causal effect.

Coverage assessment

Mechanical coverage

The map contains 62 source rows: 27 ★★★, 31 ★★, and 4 ★, for 58 priority rows. Exclusive section totals are 7 direct voice/writing leads; 13 participant/team/counterparty oral histories; 2 recording/collection routes; 10 books/scholarship/history controls; 11 press/biography/archive routes; 16 transaction/filing/legal/denominator routes; and 3 statutory/regulatory routes. Every checklist row uses exactly six pipe-delimited fields. Mirrors and manifestations are collapsed inside one annotation or identified as the same work; repeated interviews remain separate works but not independent transaction records.

Identity, name, and date-variant sweep

Searches used Paul Morrison Wythes, Paul M. Wythes, Paul Wythes, P. M. Wythes, the Activision misspelling Paul M. Whythes, and company/board/filing/lawsuit combinations. The target is separated from Paul Wythes Jr., the contemporary Oakdell professional, and the continuing firm. Public sources compress 1961, 1962, 1964, 1965, and 1970 into founding shorthand, but the evidence supports distinct company, SBIC-license, hire, portfolio-transition, and partnership events. Conflicting shorthand is preserved, not repeated as fact.

Direct voice, writings, talks, and access limits

The three-session, 163-page CHM/NVCA transcript is the only located comprehensive Wythes interview and the direct evidence spine; UC Berkeley's 177-page donated transcript is another manifestation of that one spoken work. Gupta's Done Deals chapter is edited first-person material, but the lawful route is controlled borrowing and its body was not authenticated or read. The PAW biotech feature quotes Wythes, the Darden brief derives excerpts from the oral history, Deger's roundtable is edited direct voice attested only through later endnotes, and LoBiondo's committee article summarizes a collective report. None of those four controls is a Wythes-authored writing. No personal blog, paper collection, testimony, investment memo, LP letter, annual letter, or complete speech corpus surfaced.

The strict attributable-work denominator is N=3 attributable artifacts: Gupta's edited first-person Done Deals chapter, the San Jose Mercury News letter/editorial, and the 1995 Princeton award acceptance remarks at the NYSE. 0/3 full bodies were accessible and 0/3 could be deeply analyzed: one is a controlled-borrow body not read, and two are unlocated. The Mercury item's title, letter-versus-editorial label, exact date, edition, and prose remain unknown; the NYSE item's exact date, title, recording, transcript, and manuscript remain unknown. Oral-history references are retrieval leads only. Searches of CHM, YouTube, Stanford/eCorner, C-SPAN, major podcast indexes, and general web/video results produced no independent Wythes talk, panel, podcast, or broadcast interview. CHM says the NVCA collection contains audio, but no public standalone Wythes audio object was located. No archive, firm, person, or publisher was contacted.

Formation, vehicle, economics, and return barriers

The public spine distinguishes the real-estate sponsor/company, Sutter Hill Capital SBIC, Genstar transaction, Sutter Hill Ventures partnership, later evergreen pools, management organization, and Wythes's post-1999 special-LP role. No SBIC examination file, acquisition agreement, original LP agreement, amendment, capital call, valuation policy, carried-interest allocation, withdrawal mechanism, annual statement, or distribution ledger surfaced. The transcript later labels a 36% figure as carried interest at printed p. 151/PDF p. 155, internally inconsistent with its earlier 80/20 and 75/25 economics; unresolved speech, transcription, or economic-term ambiguity prevents normalization into fact.

The PAW feature gives reported: >35% average annual Sutter Hill return over the prior thirty years (Wythes's estimate). Draper separately recalled reported: roughly 40% pre-1981 Sutter Hill IRR, while Wythes later discussed reported: roughly 36% net IRR. These are three separate retrospective estimates: none has a retrieved calculation date, complete vehicle boundary, cash-flow series, gross/net reconciliation, valuation policy, fee/carry treatment, or audit, and none is a Wythes-personal return. Further claims require original partnership and LP records before using any performance number.

Candidate denominator and attribution coverage

The frozen discovery baseline routed 40 distinct company/opportunity families once: 35 provisional associated candidates + 4 remembered passes + 1 unresolved unnamed mining write-off, with 0/35 associated candidates having the complete public chain of legal vehicle, Wythes decision right, security/cost, ownership/dilution, follow-ons, board actions, exit/write-off, and cash flows. The baseline included Webvan as associated and a supposed Balance Bar Foods lead. Subsequent body inspection established that the routed “Balance Bar” filing is actually a T. Rowe Price Balanced Fund registration and supplies no Balance Bar Foods association. The recovered official firm biography adds Tellabs; the Oacis Schedule 13G adds a distinct issuer; and Wythes's explicit non-investment account moves Webvan to the pass class. The strict final union is therefore 41 = 35 associated + 5 passes + 1 unnamed loss, still 0/35 complete chains. The 35 associated rows are Pacific Communications and Electronics, ICORE, Electroglas, R. O. Associates, Kasper Instruments, Measurex, Micro Magnetics/MMI, Micrex/Xidex, Diablo Systems, Qume, Teltone, Durrum Industries, Priam, Quantum, Duplicon, Manufacturing Management/Pierce Pacific, LifeScan, Interventional Technologies, Core Therapeutics, Molecular Devices, Cohesive Networks, Linear Technology, Applied Materials, PMC-Sierra, Apollo Computer, Hybritech, Amerigroup/AMERICAID, Shutterfly, Exogen, Activision, LSI Logic, Lexitron, Prime Computer, Tellabs, and Oacis. The five passes are Apple, Cisco, Tandem Computer, Siebel Systems, and Webvan. LSI Logic is an orphan lead retained from the frozen baseline: no claim-local supporting source has been recovered, so its firm association remains provisional rather than upgraded by repetition.

The oral history reports 46 deal events in 1964–1970—4, 11, 3, 4, 14, 9, and 1 by year, with a 31-event subtotal for 1966–1970—including four acquired Draper & Johnson positions in 1965; it separately reports 22 events during 1971–1975. Most are unnamed, and the account does not consistently separate new companies, follow-ons, or acquired positions, so neither 46 nor 31 is a named-company or career denominator. Biographical sources claim at least 27 Wythes boards without a roster. EDGAR often proves group ownership, not personal sourcing or economics. Teltone and weak outcomes are tags within the 35, not extra rows. Genstar, Biovest, India/China funds, firm vehicles, TOW Partners, and personal/trust holdings are vehicle/holder routes outside the company count. One legal company is counted once despite aliases; Core Therapeutics is not silently merged with Corcept Therapeutics.

Person/team/vehicle boundaries are material. Measurex credit is distributed among Bossen, Wythes/Sutter Hill, Bancroft/Bessemer, Reid and Peggy Dennis, and others. Apollo's founder panel identifies Draper as the Sutter Hill lead. Linear's founders name four initial funders before Sutter Hill sought entry. Applied Materials is routed through Jim Morgan. Amerigroup filings show a 1994 company and multiple holders, while Shutterfly's filing follows Wythes's active-management retirement. Sourcing, selection, capital, syndication, board service, executive recruitment, follow-on support, exit, and later narration are separate roles.

Historical, skeptical, legal, and failure coverage

The official SBIC statute/SBA program record, Federal Reserve study, Genstar corporate history, 1979 prudence rule, Kenney/Nicholas histories, and Kauffman base-rate critique frame government leverage, regional networks, LP structures, exit markets, confidentiality, and return-selection problems. These factors prevent industry development or portfolio-company success from being silently assigned to one investor.

Adversarial searches found no verified Wythes-specific SEC or DOJ sanction. They found the Micrex/Memorex threatened dispute in oral testimony; the 1974 Durrum/IPC and 1982 H&Q conflict applications; Teltone's reverse-split process; Migdal, dismissed for failure to state a claim and affirmed; and Activision, where the Section 11 statutory-underwriter claim against the VC group was dismissed with prejudice because selling into a firm-commitment underwriting did not itself constitute distribution participation. The 1982 Federal Register item is an application, not the final order. Allegations, applications, disclosed interests, and procedural rulings are not misconduct findings. Absence of a located sanction is not proof none exists.

The transcript's five passes—Apple, Cisco, Tandem, Siebel, and Webvan—plus the unnamed mining write-off, other weak-outcome discussions, and Teltone control reduce but do not cure survivor bias. A complete failure, recapitalization, dilution, shutdown, and ordinary-outcome schedule remains unavailable; hit rates, Pareto concentration, and realized multiples cannot be computed.

Archive routes and barriers

Highest-value retrieval targets are CHM reference X8628.2008 audio/permissions metadata; Bancroft/NVCA donor files; SBA license/examination records; Sutter Hill formation/partnership documents; Genstar records; Cooley files; Princeton trustee folders; Mercury/UC/CalPERS files; and issuer records. The current Sutter Hill site exposes no history, but a 2012 Common Crawl capture recovered the official Wythes biography through a reproducible index/ARC route; it is not a simple browser page and remains firm-reported evidence. No dedicated public Wythes personal-papers or Sutter Hill partnership-record finding aid surfaced; that is not proof the records do not exist.

Completed peers for later comparison

• Peter Brooke is the institutional venture-formation, attribution-boundary, archive, and incomplete-denominator comparator.
• Arthur Rock is the people-selection, Apple, syndication, finite-partnership/person-dependent vehicle, governance, and adverse-case comparator; a pass, board role, ownership stake, and realized return must remain distinct.
• Peter Crisp is the same-era institutional-capital, Apple-board, long-hold, vehicle, and succession comparator.

Bill Draper's adjacent dossier has not passed the completed-comparator gate; his oral histories are participant evidence only.

Highest-value later retrievals

1. Obtain Sutter Hill's SBIC license/examination file, 1965 acquisition/transition documents, 1969 Genstar transaction records, 1970 LP agreement, amendments, capital accounts, annual reports, investment schedules, valuations, and distributions.
2. Retrieve and process the complete Done Deals chapter, San Jose Mercury letter, 1995 NYSE remarks, CHM audio, and any class/alumni/personal correspondence without inferring missing prose.
3. Reconstruct the 31 remembered 1966–1970 deal events, determine which are distinct companies versus follow-ons or acquired positions, and reconcile the 27 claimed boards, post-1970 investments, passes, write-offs, recaps, and ordinary outcomes into one deduplicated company/vehicle schedule.
4. Build responsibility ledgers for Kasper, Measurex, Xidex, Diablo, Qume, Quantum, Linear, Applied Materials, Amerigroup, and Shutterfly from decision-date documents, separating Wythes, partners, founders, syndicates, boards, vehicles, and acquirers.
5. Reconcile every performance number to exact vehicle, time span, gross/net basis, capital calls, valuations, fees/carry, distributions, and audit status before making any return or edge claim.
6. Retrieve the Micrex/Memorex dispute record, later Teltone transaction materials, complete Migdal docket/outcome history, and other board litigation while preserving allegation, disclosure, procedure, holding, and wrongdoing as separate evidence states.

Source notes