CHAPTERS
- 0:00 – 6:09
Copying as “moral arbitrage”: why originality debates miss the point
Mark Pincus argues that copying is inherent to product creation, and the real differentiator is who performs the “magic trick” better. He distinguishes legal copying from moral/taste judgments and frames successful copying as making something feel new to users.
- •Most products are derivative; the debate is often performative
- •Taste/morality lines vary, but execution quality is what users feel
- •Examples: Jobs/Gates/Zuckerberg copying and why some copies feel “blatant”
- •The goal: emulate proven success while still delivering novelty in experience
- 6:09 – 11:00
Trust instinct, distrust ideas: the mindset behind Proven-Better-New
Before defining the framework, Pincus explains the philosophy: trust your instinct, but assume your specific idea is wrong. He uses consumer AI agents as an example of instinct-level shifts versus unclear concrete use cases.
- •Instinct can be ~95% right; specific ideas are often wrong
- •Instinct vs idea: you can sense the future without knowing the winning product
- •AI agents feel like a new “magical leap” (answers → actions)
- •Adoption hinges on trust, security, and consumer readiness
- 11:00 – 16:06
Consumer AI agents in the wild: what Instinct and Muse might be proving
Pincus breaks down what current agent products suggest about channels, friction, and user trust. He explores why SMS/text as a distribution surface feels powerful and where skepticism remains.
- •Agentic UX progression: search links → answers → actions
- •SMS as high-engagement, low-friction distribution
- •Trust and security concerns shape willingness to delegate actions
- •Early products may prove the direction even if specific use cases are fuzzy
- 16:06 – 18:45
Proven-Better-New defined precisely (and why most teams misuse it)
At the whiteboard, Pincus defines each component operationally: Proven is pixel/click-level benchmarking, Better is what 10/10 existing users prefer, and New is your isolated innovation zone. The purpose is to avoid failing for the wrong reasons by keeping novelty constrained and testable.
- •Proven = replicate the full core experience down to click/pixel level
- •Better = unambiguous user-perceived improvement (not founder preference)
- •New = your hypothesis/innovation; isolate it so it’s the only big variable
- •Bad onboarding kills discovery of the ‘new’ (Sid Meier/SimCity examples)
- 18:45 – 22:56
Zynga Poker as the canonical PBN case study
Using Zynga Poker, Pincus shows how aggressive copying of the core experience enabled a clean test of what mattered. Better was removing download friction; New was identity and friends via social graphs.
- •Proven: copied leading poker UI/UX nearly identically on purpose
- •Better: no-download web play reduced massive funnel drop-off
- •New: real identities and friend tables via Facebook/MySpace graph
- •Mantra: “All new fails” → treat new as sacred when it actually works
- 22:56 – 28:28
Choosing the ‘New’: use instinct, borrow from previously proven patterns
Pincus explains that the best ‘new’ often resembles something already proven in a different context. He discusses how he’d approach competing with agent products today when he can’t honestly claim a clear ‘better.’
- •‘New’ comes from instinct—but is strengthened by past analogs
- •Be humble: if you can’t name a real ‘better,’ don’t pretend you have one
- •Look for proven viral patterns (e.g., TBH/Gas/Honesty Box mechanics)
- •Start with ultra-light tests before building full product
- 28:28 – 29:53
Platform-cycle paradox: the earlier the shift, the less ‘new’ you need
Pincus lays out a key paradox: at the start of a platform wave (internet/mobile/AI), the platform itself is the novelty, so teams should pursue less new. Later in the cycle, attention is scarce and products need more novelty to stand out.
- •Early cycle: users are in discovery mode; ‘AI inside X’ can be enough
- •Late cycle: you need more new to find a vein of demand
- •Framework must be used in context of platform maturity
- •Avoid being ‘too clever for the moment’ early in a cycle
- 29:53 – 32:26
Hay Day vs FarmVille: a ‘0% new’ strategy that wins early mobile
Pincus praises Supercell’s Hay Day as an almost pure copy optimized for the new platform. The differentiation was ‘better’ mobile execution (speed, touch UX, pacing), not novel gameplay.
- •Hay Day ≈ FarmVille 2 visually and structurally (admired copying)
- •Better: built for mobile—faster loads, touch-first UX, tuned crop cycles
- •Why it worked then: mobile itself was the novelty
- •Why it wouldn’t work now: mature market demands more true ‘new’
- 32:26 – 35:44
Don’t be too clever right now: examples from HBS, USV, Fomo, Paraform
Pincus argues many founders over-index on novelty when there’s abundant low-hanging fruit. He contrasts ‘clever’ ideas with companies winning through better execution and accessibility, often enabled by AI leverage.
- •Early-cycle discipline: prioritize better over extreme new
- •Perfume-for-sleep example as ‘too clever for the moment’
- •Fomo: social crypto trading succeeds via accessibility + breadth of coins
- •Paraform: seemingly non-new category made better via AI-driven leverage
- 35:44 – 39:04
FarmVille’s copying playbook: what Zynga copied—and what it deliberately cut
Pincus details how FarmVille was built quickly by benchmarking Farm Town and its predecessors. The key was copying proven viral mechanics while cutting features misaligned with the target audience’s social comfort zone.
- •FarmVille built in ~6 weeks by legally copying what worked
- •Copied: gifting and neighbor mechanics that drove virality and reactivation
- •Cut: town square/stranger interaction due to ‘stranger danger’ for audience
- •Default: copy even what seems illogical—then test variations carefully
- 39:04 – 39:41
What not to copy: user specificity and building for audiences you are part of
Pincus emphasizes that taste decisions require a sharply defined target user—and ideally the builder is part of that audience. Otherwise teams operate with weak intuition and slower learning loops.
- •Target user clarity shapes what ‘proven’ and ‘better’ actually mean
- •Building for yourself increases speed/accuracy of product judgment
- •Consumer empathy is harder when you’re not in the audience
- •Changes you think are ‘better’ may actually be risky ‘new’
- 39:41 – 43:51
Test 100 ideas, don’t ship 1 MVP: faster top-of-funnel learning in the AI era
Pincus critiques MVP thinking as too hope-driven and slow-to-falsify. He proposes using AI to rapidly test demand via ads and concepts, gathering signal before committing to a full build.
- •MVP trap: ‘viable’ creates attachment and delays truth
- •Use AI to generate many concepts, names, and ad variants quickly
- •Focus on top-of-funnel metrics to validate demand early
- •Pocket FM example: CTR thresholds as a predictive growth signal
- 43:51 – 46:07
Consumer AI is (now) investable: distribution, timing, and mass-market adoption
Pincus revisits his earlier view that consumer AI wasn’t investable, noting how quickly market conditions shifted. He argues Silicon Valley overestimates mainstream adoption speed and points to the opportunity created by still-limited AI-native replacements for everyday apps.
- •Market moved fast: Instinct’s valuation changes the investability picture
- •Distribution is still unclear, but awareness grows monthly
- •SV bias: most people can still live fine without AI today
- •Opportunity: many everyday apps still lack compelling AI-native versions
- 46:07 – 48:32
Proven-Better-New beyond software: hot sauce as a product strategy exercise
Pincus applies the framework to physical goods, using hot sauce as an example. He shows how ‘proven’ taste can be identified, but replication and distribution proofs are harder than expected.
- •Proven can be sensory: repeated preference signals a winning baseline
- •Copying a formula is difficult in practice (ingredients/process variance)
- •Proven extends to marketing/distribution paths, not just product taste
- •Without proven distribution + clear better/new, it remains a hobby
- 48:32 – 55:26
The 2006 “think weekend”: unconference with Thiel, Zuckerberg, Parker—and why it worked
Pincus tells the story of convening ~35 smart people in Aspen for deep conversations without networking agendas. He explains the format, why it was special, and how others can replicate the spirit today.
- •2006 context: none of the attendees were ‘legends’ yet
- •Unconference structure: presentations as catalysts for debate
- •No agenda/no selling made the gathering unusually high-signal
- •Advice: host your own deep-topic salon; include counterfactual debates
