Mark Pincus

Who he is

Affiliation (role check run 2026-10-03 — four live hats, three missing from the Sep-5 file): Founder and Chairman of Zynga (FarmVille, Words With Friends, Zynga Poker), a Take-Two Interactive subsidiary since the $12.7B acquisition closed 2022-05-23 (Take-Two 10-K FY2022; title per zynga.com and his LinkedIn). Co-founder and co-CEO, Playful.AI — an "AI-native gaming company" formed with ex-Netflix Games VP Mike Verdu, announced 2025-06-11, bootstrapped so far (PocketGamer.biz; playful.ai). Co-founder, Prentis — an AI lab training computer-use models (AI that drives office software to do back-office work) with Ritankar Das (CEO) and Reid Hoffman; launched April 2026, in talks July 2026 to raise $100M at a $1B valuation with ~$50M of signed contracts (TechCrunch, 2026-07-24). Co-founder and Managing Member, Reinvent Capital (2018, with Hoffman and Michael Thompson; backers of SpaceX, Joby, Oscar, Lyft; three SPACs — blank-cheque shells that took Joby, Hippo and Aurora public in 2021) (Forbes; Axios via Pulse2, 2019-01-24). Author, Life at the Speed of Play (Harper Business, 2026-06-23, foreword by Hoffman) (Amazon). Co-creator of Stanford GSB's core product-management course (ten years running, with Prof. Amir Goldberg; secondary: GamesBeat). Judge, Build with Gemini XPRIZE — the $2M prize decided at Moonshots LIVE on 2026-09-25; Polyfork won the $500K grand prize (XPRIZE). Earlier: Freeloader (1995, sold in seven months for $38M), Support.com (1997, IPO 2000), Tribe.net (2003); founding-round angel in Facebook, Twitter and Napster; Presidio Trust director 2017–2020 (Obama appointee) (Wikipedia; LinkedIn). Forbes net worth $1.4B (2026-10-03).
One-line position: Most founders confuse new with better. Copy what is proven on your exact platform, make it better in a way ten out of ten users would say yes to, then isolate the one new bet — because "all new fails." Distribution decides everything; retention at day 365 (do they still come back a year later?) is the only consumer metric he trusts. He is, in his own words, "an AI maximalist" — and in his own behavior, the co-founder of two AI companies.

Discipline & technical bet

A consumer-product operator whose real discipline is distribution and behavioral loops — Zynga's achievement was not the games but the viral mechanics, retention loops and monetization that ran underneath them, built on Facebook's social graph at exactly the right moment (84% of Q3-2012 revenue still came through Facebook, down from 93% a year earlier — Zynga 10-Q, Q3 2012). His stated method is deliberately unromantic — Proven, Better, New: study the category winner "down to the pixel," make an improvement 10/10 existing users would accept, then confine the novel idea to a small slice (~20%) so the product survives when the new part fails (Lenny's Podcast digest, 2026-06-23; Summation, 2025-01-28). The entry point is the two lists — passions × mature, cash-rich markets VCs have abandoned (Inc., from My First Million, 2026-07-10).

The technical bets, as of Oct 2026 (new this pass): (1) Playful.AI — AI-first game development with free-to-play veterans; its first public move was a game jam (Aug 2025) requiring "at least one AI-enhanced mechanic" in combination-style games (PocketGamer.biz); the site still says "we're looking forward to sharing our work soon." (2) Prentis — a model-layer bet: its Hive-32B computer-use model claims to beat GPT-5.4 and Claude Opus 4.6 on computer-use benchmarks at ~10× lower cost per task, sold into insurance claims and customs refunds (TechCrunch, 2026-07-24). (3) Stem Studio — a browser-based Three.js game engine he spent four years on for his long-running Dot Earth / Erth.ai metaverse idea; he pulled the commercial effort and open-sourced it (MIT) in May 2026 (X, 2026-05-28 — date from post id; GamesBeat). Read together: he is not avoiding AI; he is placing one bet at the model layer (Prentis), one at the app layer (Playful.AI), and giving away the tooling that did not work.

Key claims (Says)

• The two lists → Proven, Better, New. Passions × proven, mature businesses; copy the winner, improve one thing everyone agrees on, isolate the new. Held — restated on every 2026 outlet (Lenny's; HBR IdeaCast, June 2026; HBS Foundry, 2026-09-23). The Sep-5 file had the entry step (two lists) and missed the framework's actual name.
• "Better" is often just "new." "Just making it different for the sake of different isn't going to win hearts and minds" (HBR). "All new fails" (Summation; Altucher 2026-06-25). Held.
• Instincts vs ideas. "Your instincts are right 95% of the time, but your ideas are right 25% of the time, at best" (Deconstructor of Fun, 2024-09-28; repeated 2026). His own example: right that people would hail cars by phone, wrong with smstaxi.com. New this pass — it is the book's spine.
• Test, don't reason. "Test more ideas in a week than your industry tests in a year"; build "failure machines" (cheap tests of hundreds of concepts) — the Mafia Wars poker link "took 5 minutes to build" (Fortune, 2026-08-20; LinkedIn). He rejects the MVP in favour of "maximum potential products" and says "every time we launched a hit game, we already knew it was a hit" (HBR). Held and sharpened.
• On AI founders — what he actually says. The Sep-5 line "do not found a company around AI" is Wrong as stated. His argument: AI "made building easier and winning harder"; founders see "a false finish line" and ship untested MVPs — "the MVP trap, accelerated" (Forbes Israel / LinkedIn, mid-2026); "AI gets us to a B-plus in seconds. Not only does it never get us to an A, it distracts us" (Fortune, 2026-08-20); test by hand first, "only after users prefer it do you add AI agents" — "that's the lightning in the bottle, not the AI" (My First Million, 2026-07-10, via BigGo digest). And in the same breath: "The new thing today is AI and agents. Go for it."
• "Consumer AI is uninvestable" — because of distribution, not AI. 40,000 games launched last year, none sustained in the top 25; his old "90-10 rule" (a platform is investable when an app can reach 10M daily users in 90 days) has no platform that passes today; ChatGPT lacks the multiplayer layer MySpace and Facebook gave Zynga (Lenny's; TBPN, 2026-07-01). He names Suno and Midjourney as the sensible consumer bets and Cursor and Harvey as proof that "wrappers" (apps built on someone else's model) win by going deeper, not shallower (Podcast Alpha digest). Then the flip: "Even though consumer is arguably not investable right now, the opportunity has never been greater" — build for a world where "AI compute is essentially free" (Wantrepreneur, 2026-06-30).
• "I'm an AI maximalist." On the AI trade: "It's not dark fiber. These GPUs are burning hot, and people are getting real value out of it"; long NVIDIA and the memory names, "a PEG ratio below one" (price-to-earnings divided by growth — under 1 means cheap for its growth) (CNBC Squawk Box, 2026-06-25; Podcast Alpha). New this pass.
• No jobs doomsday. He does not believe "the doom and gloom that all the jobs are going to go away"; AI builders "are so close to it … they lose perspective"; AI will let solo specialists serve far more clients (Axios, 2026-06-30). Earlier he predicted "massive painful dislocations" for outsourcing economies like India and the Philippines (Summation, 2025-01-28). New this pass.
• Implicit, now explicit: distribution and timing beat product quality. "Distribution decides everything … Enterprise AI has it through B2B sales and proven retention. Consumer AI does not" (Podcast Alpha). Held — he now frames it himself.
• Founder control. "Your number one job as a founder is to be right," not to be liked; "I'll keep control, to protect you and the other investors from yourselves"; "know your goal, or suffer death by 1,000 compromises" (Tim Ferriss, 2026-06-18; HBS Foundry; How Leaders Lead, 2026-10-01). Going public is "almost never good" for a company (TBPN). New this pass.

Notable predictions — with falsifiable checks

• (2007–12) Social platforms would become the dominant games distribution channel. Held, spectacularly and briefly. Zynga rode it to a $1B IPO in Dec 2011; 93% of revenue ran through Facebook in Q3 2011, 84% a year later as the terms changed (Zynga 10-Q, Q3 2012), and the stock fell from a $15.91 high (Mar 2012) to ~$2 by 2016 (SF Chronicle, 2016-03). A correct prediction that contained its own reversal — still the most useful thing on this page.
• (2026, live) "Consumer is not investable because of distribution … Go for it." His contrarian call: the 2026 consumer-AI winter is 2007 again, and the breakout consumer companies will be founded now while VCs avoid the sector (Wantrepreneur, 2026-06-30; My First Million, 2026-07-10). Check: by 2028-07, does at least one consumer-AI company founded 2025–26 reach 10M DAU (daily active users) or a $5B+ round? His own stake: does Playful.AI ship a title with public retention numbers by 2027-06?
• (2026, live) Enterprise "wrappers" with distribution beat raw model access; the human first-and-last mile is the moat. Revealed version: Prentis's $75M annualised run-rate forecast for Q3 2026 and a $1B raise (TechCrunch, 2026-07-24). Check: raise closes at ≥$1B by 2027-01; a disclosed run-rate ≥$50M by 2027-03. Miss either → Drifted.
• (2026) AI will not cause mass job loss; the builders are too close to see it. (Axios, 2026-06-30.) Check: U.S. unemployment stays under 5% through 2027-06 (BLS) and the Stanford "Canaries" entry-level gap stops widening. The same observable grades Ng and Brynjolfsson on this bench.
• (2026) Games + AI becomes "a $3 trillion industry" once "AI and games get figured out" (GamesBeat, mid-2026). No date given — unfalsifiable as stated. Working check: global games revenue (~$190B, Newzoo 2025) exceeds $500B by 2032 would be the first milestone worth logging.
• (ongoing) "New" underperforms "proven model, new market." Still a heuristic, but he now puts a number behind it: 8 of his 10 major launches were hits (self-reported, Lenny's, 2026-06-14) and "if you isolate the new to maybe 20% of your product … there's a good chance" (Summation). Check: Playful.AI's own titles are the next sample — grade them against his 80% hit rate.
• Career-as-prediction: the $12.7B exit. Drifted. The Sep-5 file called it "selling at the top." He now calls the sale an underperformance (How Leaders Lead, 2026-10-01); the price owed much to his 2015 move of $800M of Zynga's $1B cash into buybacks at ~$2.20 a share versus the ~$10 Take-Two paid (MobileGamer.biz, 2026-09-29); and he did not cash out — of 55M shares, ~$193M came as cash and ~$350M as Take-Two stock (CNBC, 2022-01-11). Not a timed exit; a long hold that worked.

Revealed behavior (Does)

• (2025-06) Co-founded an AI-native game company, Playful.AI, as co-CEO with Mike Verdu; bootstrapped; CTO Richard Sun ex-Netflix GenAI (PocketGamer.biz, 2025-06-11). Directly contradicts the Sep-5 file's "advises founders not to build AI companies."
• (2026-04 → 07) Co-founded an AI lab, Prentis, and let it go to market for $100M at $1B within three months of launch (TechCrunch, 2026-07-24). Hoffman left Microsoft's board around the same time to go "founder mode"; Pincus runs Reinvent with Hoffman as senior adviser.
• (2026-05) Open-sourced Stem Studio after four years and killed the commercial effort — a public write-off of his metaverse tooling (X, 2026-05-28; GamesBeat). Behavior matches the talk: kill the idea, keep the instinct.
• (2026-06 → 10) Ran a full book tour — CNBC, Bloomberg, Lenny's, HBR, Tim Ferriss, Altucher, Fortune, Axios, Commonwealth Club (2026-08-27), HBS Foundry, David Novak. The Sep-5 file's "little visible evangelism" was written in the middle of it.
• (2024-11 → 2025-06) Switched political sides in public. A self-described lifelong Democrat who gave ~$1M per cycle for four cycles (Obama ×2, Clinton, Biden), he wrote "Biden is even riskier than Trump" in The Free Press (2024-07-18), posted "I am voting for Trump" on 2024-11-04 citing Israel, antisemitism and free speech, and on 2025-06-17 wrote "This is why I voted for Trump" over the Iran strikes. Absent from the Sep-5 file entirely. Relevant to the bench because it is the same move — instinct over tribe — applied to politics, and because it tells you which rooms he is now in.
• Still writes cheques. Most recent disclosed angel: Levellr seed, 2026-01-29 (Tracxn, secondary). Cold-emailed and invested in Raya after noticing it on his own phone (Lenny's). Reinvent Capital: SpaceX, Joby, Oscar, Lyft; three 2021 SPACs (Forbes). Long NVIDIA and memory (CNBC, 2026-06-25). Note the gap: no 13F for Reinvent Capital was located this pass — the public-market book is asserted, not filed-and-checked.
• Copies deliberately and says so. "You don't have the right to change something that's proven. You are not good enough at this" (HBS Foundry, 2026-09-23). Held.
• Wrote the method down — ten years of the Stanford GSB course, then the book (2026-06-23). Held.
• Gave up control once, by choice. Cut his voting power from ~70% to ~10% in 2018 (Variety, 2018) — worth holding next to his 2026 "keep control" sermon. In 2012 he chose not to use that control to force a $400M Supercell purchase past his board (MobileGamer.biz, 2026-09-29); Supercell made ~$500M profit the next year.
• Judges an AI prize while building AI companies. The Sep-5 file read this as a contradiction; it is alignment. Resolved, not "worth noticing."

Feels

A pragmatist with a long memory of a bruising ride — the Zynga arc from IPO darling to "worst CEO" lists to a good-not-great exit — who says being named worst CEO "freed" him (How Leaders Lead, 2026-10-01). He wants the method to be the lesson, not the luck. And he is visibly energised: "working harder than I have in years" on AI (Podcast Alpha); two new companies at 60. The grievance channel is politics, not tech.

Hears

Reid Hoffman above all — co-founder at Reinvent, Prentis and the book's foreword. Then founders pitching him, Stanford MBAs, and the product-podcast circuit (Lenny, Ferriss, HBR) that now frames him as a method teacher rather than the FarmVille guy. Politically he reads and reposts the pro-Israel, anti-"woke" tech-right feed. Probably hears little from the AI-safety or attention-economy critics.

Sees

Distribution channels and retention loops where others see products. His home-screen test — which apps earn a slot shows where the early majority will be in 18 months — and friction posts on social media as live product signal (Lenny's). The privileged vantage in 2026: he sits on both sides of the consumer-AI argument at once — a model lab with $50M of enterprise contracts and a bootstrapped consumer studio — so his "uninvestable" is a seller's read, not a bystander's.

Incentive map

The Sep-5 file said "light-to-moderate … no fund raising, no token, no cohort." Wrong. He has: a book to sell (and a speaking agency, Harry Walker); a $100M raise at $1B to close at Prentis, where "enterprise AI has distribution" is also the pitch; a bootstrapped consumer AI studio whose eventual round benefits if "consumer is uninvestable" becomes consensus right up until he raises; a venture firm with Hoffman; and a public-market book long NVIDIA while he tells CNBC "GPUs are burning hot."

Cannot easily say: that Zynga's success was substantially platform accident; that Prentis's benchmark claims are self-reported; that the Facebook-era engagement playbook is what critics now call attention capture. What he can say that others on the bill cannot: that most new ideas fail and that the sale was an underperformance — he is unusually candid about his own misses (Supercell, Tribe.net, smstaxi, Stem Studio).

Theories aligned with

• Distribution over product — the operative belief, now stated outright
• Proven, Better, New — arbitrage rather than invention; "all new fails"
• Instinct over idea — the human read is right, the implementation is wrong; test the implementation
• Augmentation, not replacement — no jobs doomsday; solo experts serve more clients (shares this with Ng and Andreessen, Axios 2026-06-30)
• Human first-and-last mile — Prentis's thesis that judgment stays human and the model learns from it
• Founder control / "founder mode" — protect investors from themselves
• Behavioral design / engagement loops — the Zynga inheritance, ethically contested

What he's reacting against

• Founders in love with originality — the central target
• The MVP trap, accelerated by AI — shipping untested because building is cheap
• AI-jobs doom from the labs — "they lose perspective" (Axios, 2026-06-30)
• VC herding into enterprise AI while consumer sits "uninvestable"
• Boards and bankers — the Supercell veto, "going public is almost never good"
• The Democratic Party as he knew it — Israel, antisemitism, speech (2024–25)

Where he overlaps / splits (with Rich)

• He has a direct quarrel with the Ideas pipeline, and it is sharper than the Sep-5 file said. Your Ideas folder prizes originality, cross-domain mashups and "problems nobody is serving well." Pincus: originality is where founders die; copy the proven, improve one thing, confine the new to 20%. That is a different scoring function from your graduation thresholds and it is not obviously worse. Cheap test: run three promoted Ideas through Proven/Better/New and see which column each lives in.
• The Agency Lab quarrel is now different — and better. The Sep-5 file had him saying "don't found a company around AI," which would have made him a sceptic of Ascend and the CoS Kit. He isn't. His actual rule is §18 No Job, No Build in his own words: find the proven market and the human job, prove people prefer your version by hand, then add the agents. And his Axios line — AI lets a solo specialist serve many more clients — is a description of Agency Lab's premise, not an objection to it.
• Overlaps on §02 (Start with the Human Job-to-Be-Done) — "passions × proven models in dead markets" is a job-first heuristic wearing different clothes.
• Overlaps on §16 (Would I Do This For Free?) — his first list, "things you're genuinely passionate about," is the same gate, and the Ideas-pipeline Authenticity Filter is the same gate again. He arrived there from a different direction, which is mild evidence it is real. (The Travel Test is an Ideas-pipeline gate, not a §; his "forever franchises" idea — pick the few things that run for 19 years — rhymes with it.)
• Overlaps on §09 (Judgment Over Execution) — "AI gets you to a B-plus; it never gets you to an A, and it distracts you." Human judgment is the scarce thing; AI is the fast executor. He and Rich agree on where the moat is.
• Splits on the layer. Rich's Barbell says invest in infra, build at presentation. Pincus just did both at once — a model lab and a consumer studio. Worth asking whether that is a hedge or a thesis.
• Splits hard on ethics. Zynga-era engagement mechanics are the ancestor of what the AI & Society material objects to about attention capture. His only 2026 engagement is oblique — "most games are empty calories" (How Leaders Lead, 2026-10-01) — about value, not capture.
• Splits on politics, which the Sep-5 file omitted. Post Modern Progressive versus a 2024 Trump convert. Note it, don't let it grade his product claims.
• For the bench: still the best adversarial read on the Ideas scoring model — now with a worked 2026 example (the curated-Yelp sketch from My First Million) and two live companies to grade him against.

Track record

Verified this pass: Freeloader (1995, $38M in seven months), Support.com IPO (2000), Tribe.net (2003; tech to Cisco 2007), Zynga founded July 2007, CEO 2007–13 and 2015–16, IPO Dec 2011, Take-Two acquisition closed 2022-05-23 at $12.7B; founding-round angel in Facebook, Twitter, Napster; Reinvent Capital 2018 and three 2021 SPACs; Stanford GSB course; Life at the Speed of Play (2026-06-23); Playful.AI (2025-06); Prentis (2026-04); Stem Studio open-sourced (2026-05).

The honest read, revised: one very large success built on a distribution window that closed, then a seven-year grind (two CEO stints, an $800M buyback, "worst CEO" lists) to a sale he himself calls an underperformance. The pre-Zynga companies were modest; Tribe.net lost to Facebook. The 2021 SPACs (Joby, Hippo, Aurora) are mixed-to-poor public-market outcomes to date — not re-priced this pass; flag for the next one. The method is derived from a sample of one company and eight hit launches inside it, which does not make it wrong, and does mean it is a heuristic, not a law. 2026 is the year he put new money behind the heuristic.

Empirical vs normative

Empirical and checkable: consumer-AI distribution is broken (app-store hit rates; his 90-10 rule); enterprise wrappers with sales channels win (Cursor, Harvey; Prentis's contracts); no mass job loss (BLS, Canaries). Empirical but self-reported: 8-of-10 hit rate; 95%/25% instinct-vs-idea; Hive-32B benchmarks. Normative: start from passion and proven models; keep founder control; don't go public. He does not disguise heuristics as findings — but in 2026 he has an incentive to, because two companies now depend on them.

Weak spots / open questions

• n=1, now with skin in the game. If Playful.AI ships nothing memorable by 2027, the "proven, better, new" record gets its first out-of-sample miss in public.
• "Uninvestable" from a man raising. The consumer-AI winter thesis flatters a bootstrapped studio that will one day want a round, and the enterprise-distribution thesis is Prentis's deck. Grade the lean each pass.
• Benchmark claims are the company's. Hive-32B vs GPT-5.4 / Opus 4.6 has no third-party eval located this pass.
• Timing versus method. Unchanged: the method explains neither the 2007 entry nor the 2022 price (which was buyback math plus a buyer's strategy).
• The ethics of the engagement playbook remain unaddressed in attention-capture terms. Held as a weak spot; he has not engaged the critique.
• The SPAC trio. Three 2021 SPACs under his name; nobody on this bench has graded them. Next pass.
• Questions worth asking: What would Playful.AI have to show for him to call it a hit — day-365 retention on what base? Is Prentis the Barbell's infra leg or an app with a model attached? And: he gave up voting control in 2018 and preaches keeping it in 2026 — which Pincus should a founder listen to?

Rich's take

• (your synthesis here)

Delta log

2026-10-03 — validation + meta repair (parallel batch)

• Grades: 9 Held / 3 Drifted / 4 Wrong, plus 1 unfalsifiable item given a working check, 3 new live predictions with dated checks, and 1 misattributed prediction retired and rewritten. The Sep-5 file was 28 days old and already wrong on its headline claim.
• Wrong: "do not found a company around AI" — he co-founded Playful.AI (PocketGamer.biz, 2025-06-11) and Prentis (TechCrunch, 2026-07-24) and calls himself "an AI maximalist" (CNBC, 2026-06-25). · "AI-first startups will underperform AI-as-cost-lever companies" — a bench inference, never his words; his actual call is the opposite for consumer AI ("Go for it", My First Million, 2026-07-10); retired and rewritten. · "Judges an AI prize while advising founders not to build AI companies" — the contradiction does not exist. · Incentive map "no fund raising" — Prentis is raising $100M at $1B (TechCrunch).
• Drifted: "sold at the top" — he calls the sale an underperformance (How Leaders Lead, 2026-10-01), the price was partly buyback math (MobileGamer.biz, 2026-09-29) and he took mostly stock (CNBC, 2022-01-11) — both the prediction row and the behavior row. · "Little visible evangelism" — a four-month book tour (CNBC, Bloomberg, Lenny's, HBR, Ferriss, Fortune, Axios, Commonwealth Club 2026-08-27, HBS Foundry 2026-09-23, Novak 2026-10-01).
• Held: two lists / Proven-Better-New (Lenny's; HBS Foundry); "better is often just new" (HBR, June 2026); testing over reasoning (Fortune, 2026-08-20); distribution beats product (Podcast Alpha); social-platform distribution 2007–12 (Zynga 10-Q Q3 2012); copies deliberately (HBS Foundry); angel-invests widely (Tracxn, Levellr 2026-01-29); wrote the method down (book, 2026-06-23); engagement ethics still unaddressed (no 2024–26 statement located).
• Most surprising delta: the file's central claim was falsified before it was written — Playful.AI was three months old and Prentis seven weeks old on 2026-09-05. The Inc. piece the file leaned on was a digest of a podcast; the primary (My First Million) says AI is "the new thing … go for it."
• Role check: four live hats, three missing — Playful.AI co-CEO, Prentis co-founder, Reinvent Capital managing member. Zynga title confirmed Founder and Chairman (zynga.com). XPRIZE described him as "Founder of Zynga and Co-Founder of Reinvent Capital" on 2026-08-19; Polyfork won on 2026-09-25 (XPRIZE).
• Omissions filled: 2024 Trump endorsement and 2025 Iran post (X, 2024-11-04; X, 2025-06-17; Free Press, 2024-07-18); Stem Studio open-sourced (X, 2026-05-28); 2018 voting-control cut (Variety); Supercell veto and $800M buyback (MobileGamer.biz); jobs stance (Axios, 2026-06-30); 84%-via-Facebook figure (10-Q). Items the Sep-5 file carried "from general knowledge" (2007 founding; Freeloader/SupportSoft/Tribe) now verified via Wikipedia and zynga.com.
• Meta repaired: the Sep-5 discipline ("physician-engineer turned prize-designer") and wake_triggers ("Giga-XPRIZE … Fountain Life") belonged to Peter Diamandis — bulk-create leakage, same as Kilpatrick, Luckey and Wood. Replaced. connects-to re-pointed from post-labor-economics to augmentation-thesis and distribution-over-product. § citations checked against the canon: §02, §09, §16, §18 confirmed; Authenticity Filter and Travel Test are Ideas-pipeline gates and now carry no § number.
• Not re-verified this pass (flagged, not asserted): Reinvent Capital 13F holdings; SPAC trio performance; Hive-32B benchmarks (company claim); Playful.AI's $770K revenue figure (GetLatka, low-quality secondary — not used in the body). Blocked fetches: Inc., Forbes Israel, Bloomberg, Lenny's podcast page, X post bodies — all cited via digests or search snippets and labeled.
• Tier: annual → quarterly. Two companies in formation, one mid-raise, a daily X feed and a political posture that moved in 2024–25 — behavior will outrun talk for at least a year. Confidence stays medium: the record is dense but several primaries were behind paywalls and the AI-company facts come from trade press, not filings. Rich's take untouched (placeholder preserved).

2026-09-05 — created

• New persona at v2 depth for Moonshots LIVE 2026-09-25 (Build with Gemini XPRIZE judge). Tier: annual.
• Verified this pass: the two-lists formula, the "better is often just new" framing, the warning against founding companies around AI and the "AI as cost reducer" recommendation, the $12.7B Take-Two figure, and Life at the Speed of Play (Inc.). Acquisition timing corroborated by CNBC, 2022-01-11.
• Carried from general knowledge, not re-verified: Zynga founding year 2007, and the Freeloader / SupportSoft / Tribe.net history.
• Framed the finding: he is a direct, credentialed challenge to two of Rich's own structures — the originality-weighted Ideas scoring model, and the AI-founded premise of Agency Lab. Filed as a stress test, not a north star.
• Noted that his AI scepticism runs against his own interests at this event, which raises its credibility. (2026-10-03: this reading was wrong — see above.)
• Confidence medium.

Sources

• Roles and companies: zynga.com — Founder and Chairman · LinkedIn · PocketGamer.biz — Playful.AI formed (2025-06-11) · playful.ai · PocketGamer.biz — game jam (Aug 2025) · TechCrunch — Prentis (2026-07-24) · Forbes profile ($1.4B, 2026-10-03; Reinvent portfolio) · Axios via Pulse2 — Reinvent Capital $700M (2019-01-24) · Wikipedia — career timeline, SPACs · Variety — voting control (2018) · XPRIZE — judges (2026-08-19) · XPRIZE — Polyfork wins (2026-09-25)
• Filings: Take-Two 10-K FY2022 — Zynga closed 2022-05-23 · Zynga 10-Q Q3 2012 — 84% of revenue via Facebook · CNBC (2022-01-11) — 55M shares, ~$193M cash / ~$350M stock
• 2026 book tour (talk): Life at the Speed of Play (2026-06-23) · X — book announcement (2026-04-10) · Tim Ferriss blog (2026-06-18) · Sourcery — Lenny's Podcast digest (2026-06-23) · Lenny's Newsletter · Lenny on X (2026-06-14) · CNBC Squawk Box (2026-06-25) · Bloomberg Talks (2026-06-25) · James Altucher Show (2026-06-25) · HBR IdeaCast (June 2026) · Axios (2026-06-30) · Wantrepreneur (2026-06-30) · TBPN digest (2026-07-01) · Next Big Idea Club (2026-07-07) · My First Million (2026-07-10) and BigGo digest · Inc. — two lists (digest of MFM) · Forbes Israel — AI founders / MVP trap · Fortune (2026-08-20) · Commonwealth Club (2026-08-27) · HBS Foundry (2026-09-23) · MobileGamer.biz — Supercell, buyback (2026-09-29) · How Leaders Lead digest (2026-10-01) · Podcast Alpha — consumer AI uninvestable, long NVIDIA · GamesBeat — Dot Earth, Stem Studio, $3T · X — Stem Studio open-sourced (2026-05-28)
• Pre-2026: Summation with Auren Hoffman (2025-01-28) · Deconstructor of Fun (2024-09-28) · PocketGamer.biz — "f*** scale" (2017-12) · SF Chronicle — Gibeau replaces Pincus (2016-03)
• Politics: The Free Press op-ed (2024-07-18) · X — "I am voting for Trump" (2024-11-04) · X — Iran / Israel (2025-06-17)
• Secondary, used with care: Tracxn (Levellr seed, 2026-01-29) · Harry Walker Agency (speaking) · Stanford GSB course details via GamesBeat and eciks.org. Not consulted: the book itself; any Reinvent Capital 13F. Next pass: read the book's AI chapter and pull the 13F.

validated 2026-10-03 (created at v2 2026-09-05; meta block repaired, tier annual → quarterly) · backup: _archives/mark-pincus.html.bak-20261003 · converts-from: none · Singularity 🧠 / People