· Mark Pincus, Reid Hoffman

Mark Pincus & Reid Hoffman: How to Build Breakout Products

Great products come from disciplined copying of what is proven, boxing innovation into one place, and launching wrong-and-fast to learn — and the AI era is a consumer revolution most people are missing, just like the 2002-03 internet nuclear winter.

productai-eraproven-better-newconsumervibe-codingmetaverse0% confidence

Why this is in the corpus

Third Pincus interview in the corpus (Masters of Scale), reinforcing Proven/Better/New and instincts-vs-ideas while adding fresh AI-era material: the consumer-revolution parallel to the internet nuclear winter, the is-AI-a-platform-yet debate, vibe coding democratizing product-making, and the metaverse arriving outside of goggles.

Summary for skimmers

Pincus reframes Proven/Better/New for a world with little "proven," argues AI is not yet a consumer platform but a coming consumer revolution, and pushes founders to build wrong-and-fast because "viable is a weak word."

Briefing

What survives the editorial filter

This page should feel like a smart colleague already listened for you and left only the operating logic worth keeping. Not everything said in the episode makes it through.

Trust signal

Direct episode extraction

Best used for

Decision-grade retrieval metadata not yet added for this episode.

Hold lightly

No explicit downgrade reason stored yet for this episode.

Principles

Durable claims that survive beyond the speaker's biography — each with explicit limits, transferability judgment, and evidence.

Principle

"Viable" is a weak word — build on conviction, not the hope of viability

Replace the MVP's "viability" bar with a conviction bar.

Pincus critiques Eric Ries's "viable": the word smuggles in hope, and hope is a weak reason to build. What you want is conviction, which is strong. Viability is a low, hope-laden bar; conviction demands real signal.

Do not build for viable; build for conviction.

“that word viable is actually hurtful because viable also means hope. Yes. Of viability. Yes. And what you want is not viable on conviction. And viable is actually a very weak word to me.”Mark Pincus

Principle

What you believe is "better" is actually "new" — real better is what 10 of 10 users would call better

If it is not obvious better to 10 of 10 current users, it is "new," not "better."

Pincus draws the line between "better" and "new": your belief that something is an improvement is really untested novelty. Real "better" is unanimous among the product's existing users and usually simple — you already know the consumer experience, so a true better is one they immediately recognize.

Relabel your "better" as "new" until users prove otherwise.

“what you believe is better is called new. Right. What's actually better is what 10 out of 10 of the existing users of that product would say.”Mark Pincus

Principle

Your instincts are right; your ideas are usually wrong

Trust the gut signal that something is broken; distrust your specific idea of how to fix it.

Pincus separates the winning instinct (the felt sense that a category is broken or a demand exists) from any particular idea. You "had the instinct that something was broken there," not the idea for the exact product. Treat the idea as disposable and the instinct as the compass.

Anchor on the instinct, prosecute the idea.

“if you accept this premise that you're instincts are right and your ideas are usually wrong, then what do you do with that?”Mark Pincus

Principle

Real "better" is mundane: free, no download, one inch better on something users already care about

True improvements are boring: cheaper, frictionless, slightly more polished.

Pincus insists real "better" is unglamorous — free, no download, one inch better on a dimension users already care about and would notice, like extra polish. Novel dimensions are "new," not "better."

Ship boring betters on axes users already value.

“better is usually very simple and basic things like it's now free, no download. It's something that they really care about is like one inch better.”Mark Pincus

Principle

Your "new" is probably why people try the product — and probably why it fails

The "new" both attracts trial and is the likeliest failure point — plan for it.

Pincus calls "new and improved" the back of the box: the new is the reason people try, and probably the reason they fail. Because you assume the new idea is likely wrong, you can plan alternates in advance.

Assume your new is wrong and pre-plan alternatives.

“new is your novel idea. That will probably be the reason why people will try this product And it will probably also fail.”Mark Pincus

Principle

Box your innovation into one place; be a scientist prosecuting the instinct

Innovate in exactly one declared place and treat it as a hypothesis on trial.

Proven/Better/New is a way to isolate innovation: declare where you are innovating and where you are not, then act like a scientist in a white lab coat "prosecuting this instinct" — unattached to the idea, giving the team permission to change every Monday.

Declare the one place you innovate; hold it on trial.

“It's a way to isolate and box in your innovation in one place and to say, this is where I'm innovating and this isn't.”Mark Pincus

Principle

If you are not embarrassed, you are not launching early enough

Launch to learn at the point of embarrassment, not pride.

Pincus reframes launch itself: it is not a proud unveiling but "turning it on to the level that you'll learn." If you are not embarrassed, you waited too long.

Turn it on the moment it can teach you something.

“if you're not embarrassed, you're not launching early enough. And it's to learn And it, and launch is the wrong word. It's like you should turn it on to the level that you'll learn.”Mark Pincus

Principle

When nothing is proven, hunt for the "fuzzy proven" transplanted from elsewhere

In no-precedent environments, import a proven behavior from an adjacent context.

For AI, where "there's very little proven," Pincus says look for a fuzzy proven — an experience proven somewhere else you can apply here. It is a better vein than inventing something the world has never seen, because behavior persists across eras.

Find the adjacent proven and port it to the new platform.

“there's a fuzzy proven you're looking for. It's, is there something that was proven, there's an experience that was proven somewhere else and now I could try to apply it here.”Mark Pincus

Principle

The best product makers are the best at copying — with the least ego

Mastery of product is mastery of faithful copying without ego.

The paradox: the most senior product makers copy proven elements most faithfully and feel no need to prove originality, while junior makers feel compelled to change everything and are the worst at copying.

Drop the ego; copy the proven exactly.

“the best masters of the craft are the best at copying. Yes. And they have the least ego around it.”Reid Hoffman

Principle

Real ambition is the willingness to be wrong for a long time

Ambition is measured by tolerance for prolonged wrongness, not intensity.

Pincus probes founders who claim ambition of "a 10." Real ambition means you can be wrong for three years and then very right, versus most people who are "a little bit right every day." The willingness to be wrong long is the differentiator.

Ask if you can stomach three years of being wrong.

“if you're really ambitious, you're willing to be wrong for a long time. Yep. You can be wrong for three years and then really Right.”Mark Pincus

Principle

The magic trick of copying is that the end user does not feel it is a copy

Copy the proven mechanics but make it feel original to the user.

Pincus and Hoffman note that Zuckerberg's Threads copied a proven product "artfully" so "people don't realize it's a copy." The art is staying as close as legally possible to what works while ensuring the end user does not experience it as a blatant copy.

Hide the copy in the presentation, not the mechanics.

“The real art of it is that the end user doesn't feel like it's a blatant copy. That's the real magic trick.”Reid Hoffman

Principle

Do not blame the platform — screen time is a modeling and parenting problem

Screen overuse is downstream of what adults model, not just product design.

Pincus, a builder of addictive apps, refuses to blame Facebook or games for screen addiction, comparing it to blaming cigarette companies while smoking in front of your kids. The real question is what behavior adults model.

Fix the modeling before blaming the product.

“I don't blame Facebook for this, you know, I don't think it's like cigarettes and it's a drug or I don't blame games. I think it is a part of our culture and society that we, what are we modeling for them?”Mark Pincus

Frameworks

Reusable systems and operating models — including when they help and when they break.

Framework

Proven / Better / New

Decompose any product into Proven (copy), Better (unanimous improvement), and New (one novel bet).

Born from Tribe's failure and Zynga's discipline, Proven/Better/New is Pincus's scientific method for products: copy the proven faithfully, add only "better" that all current users would recognize, and confine novelty to one declared "new." Zuckerberg's Threads earns "an A minus" for executing it.

Sort every product element into Proven, Better, or New.

“proven better. New is this framework that we got to out of the failure of tribe came the commitment to winning in Zynga. And with that, this framework of proven better new. And the idea was to say, okay, we've gotta be more scientific about this.”Mark Pincus

Framework

Instincts versus Ideas — the chapter before the framework

Before vetting products, classify each impulse as a trustworthy instinct or a disposable idea.

Pincus frames instincts-versus-ideas as the foundational chapter beneath Proven/Better/New: you have winning instincts in your gut ("you just know it"), but your ideas are usually wrong. The discipline is to kill bad ideas and keep good ones while staying loyal to the instinct.

Name the instinct, then treat every idea as expendable.

“In order to explain proven better new, I gotta start a chapter earlier on instincts versus ideas.”Mark Pincus

Framework

The testing-and-failure machine at the top of the funnel

Institutionalize high-volume, top-of-funnel testing to out-learn your industry.

Pincus's Zynga-proven system: build a machine that tests more ideas in a week than the industry does in a year, concentrated at the top of the funnel — testing click-through on an ad or landing page before building anything downstream.

Make top-of-funnel testing a machine, not a one-off.

“test more ideas in a week than your industry tests in a year. Build a testing and failure machine usually at the top of the funnel.”Mark Pincus

Signals

What appears to be shifting, for whom it matters, and what happens if you ignore it.

Signal

AI is not yet a consumer platform — the front door and portal have not formed

AI lacks a consumer distribution platform, unlike early Google.

Pincus argues that despite ChatGPT's size, AI is not yet a consumer-level platform — there is no SEO/SEM-style front door magnifying distribution the way Google was a "platform enabler." Builders "could see it, hope for it," but are "not yet there."

Do not assume AI is a distribution channel yet.

“I would argue it is not yet a platform at a consumer level. There's, there's no, despite The fact that it's huge. Yes. It's not yet there”Mark Pincus

Signal

AI will unleash a massive generation of new products — about to go bonkers large

Product creation is about to explode as AI democratizes building.

Pincus predicts AI will drive a massive generation of products "about to go bonkers large," citing his partner Hillary and daughter Georgia building working apps via Claude Code vibe coding. Everyone becomes a product maker; the stack automates and it comes down to taste.

Prepare for an order-of-magnitude increase in products.

“one of the things I think we're gonna see with AI is a massive generation of products. Yeah. This is just about to go bonkers large.”Mark Pincus

Signal

The metaverse will be as big as Zuck thought — but not inside goggles

The metaverse arrives via ambient devices and agents, not immersive goggles.

Pincus predicts the metaverse will be as big as Zuckerberg believed, but delivered through ambient devices and consumer agentic AI — avatars that roam finding dates, jobs and money — rather than hot, painful immersive goggles. "It just doesn't have to be immersive 3D."

Bet on ambient and agentic delivery, not headsets.

“I think that the metaverse is starting to happen, and I think it's going to happen and it's going to be as big as Zuck thought. And it's just not gonna happen inside of an immersive game goggle that is painful and hot to wear on your head”Mark Pincus

Signal

We are all about to become product makers — building is being democratized

AI turns everyone into a potential product maker, making taste the differentiator.

Watching his non-engineer partner build a location-aware activity app for their kids via Claude Code, Pincus concludes "we're all gonna become full stack, then the stack's automated and it's just profitable and then it gets down to taste."

When the stack automates, compete on taste.

“We're all gonna become product makers because it's democratized, because it's accessible.”Mark Pincus

Opportunities

Only included where there is a buyer, a real wedge, and a plausible revenue path — not vague idea theater.

Opportunity

The whole AI consumer revolution almost nobody is building for

The AI consumer wave is wide open because the herd is chasing enterprise again.

Hoffman and Pincus recall the internet nuclear winter (2002-03) when everyone fled to enterprise or clean tech and they bet on the just-starting consumer internet (the web-two thesis). AI now rhymes: enterprise is where the money seems to be, and "few people are thinking about the consumer thing" — the bigger unrealized wave.

Build for the AI consumer wave the herd is ignoring.

“it's funny how few people are thinking about the consumer thing right now. It's a very good illustration.”Reid Hoffman

Opportunity

Games that improve your life — get you a job by playing

Games that deliver real-life outcomes (skills, jobs) unlock a far larger market.

Pincus sees an unlock in making games a good use of time: if games improved life skills or effectively "did LinkedIn for you" and got you a job, the industry would be far bigger — merging the real world, real humans and real social into play.

Wire real-world outcomes into game mechanics.

“Imagine if games improved your life skills. Imagine if games were doing LinkedIn for you. You know, imagine if you were getting a job by playing a game, like how much bigger would it be?”Mark Pincus

Lessons still worth keeping

Useful takeaways that did not fully clear the bar for durable principle status.

Lesson

In 2004 Shenzhen, LinkedIn had three clones so exact Pincus could use them without reading Chinese

Exact LinkedIn clones in China proved the mechanic itself is the transferable proven asset.

Visiting Shenzhen in 2004, Pincus found three LinkedIn copies so exact he could complete a profile in Chinese without reading it, navigating purely by the copied mechanic and layout.

Copy the mechanic precisely; it carries the value.

“in 2004 when I went to Shenzhen, there were three copies of LinkedIn. There were such exact copies in Chinese, I could fill out a profile and I don't speak any Chinese speak, read, write, et cetera, but I just like click on this button, type in this field.”Mark Pincus

Lesson

Threads earned an A-minus by executing Proven/Better/New as an artful fast follow

Meta's Threads is a worked example of a disciplined, artful fast follow.

Pincus grades Zuckerberg's Threads "an A minus" as a fast follow: it copied a proven product's mechanics, stayed close to what works, and did so artfully enough that users did not perceive a blatant clone.

Study Threads as the template for an artful fast follow.

“if you're Zuck and you're building threads, which is, you know, a copy, a fast follow of another very successful product, he did a a very good, I give him an A minus very good job of proven better new.”Mark Pincus

Lesson

The Risk redesign destroyed the proven — plastic pieces and Roman numerals nobody wanted

The Risk board-game redesign shows how touching the proven and mislabeling new as better destroys a product.

Pincus's favorite bad example: an MBA at the maker of Risk swapped the wooden pieces he loved for plastic, used Roman numerals instead of visible force, and added an unlearnable Castle Risk mode — "proven worse and bad new."

Never redesign the proven elements users are attached to.

“It was like proven worse and bad new. They took the wooden pieces that I loved and made 'em plastic. They made 'em Roman numerals. Like, I wanna show my force. I don't want Roman numerals. That's not better.”Mark Pincus

Lesson

Friendster grew because the tsunami moved the boat, not because it understood virality

Friendster's explosive growth revealed massive latent consumer demand, not product mastery.

Pincus and Hoffman wrote checks into Friendster "not because we thought it had any chance" but because the experiment needed to happen for the consumer internet. Its growth, despite no understanding of virality, showed how strong the underlying consumer demand was — the tsunami moved the boat.

Read runaway growth as a demand signal, then build the real capability.

“Friendster didn't understand virality or growth hacking. No. They just literally, the wave picked up the boat. Yes. Yes. And the tsunami moved it along.”Mark Pincus

The Plays

Try these this week

Verb-first executable actions — each one tied to a stated outcome in the episode.

Pre-plan four other versions of your "new" so you can learn faster

Outcome: Enumerate alternate "new" variants in advance to iterate the moment the first fails.

Context: Anticipating that the new idea is probably wrong, Pincus plans four other versions ahead of time and studies other products' versions of the same behavior, so learning continues without a restart when the first bet fails.

“if we realize our new idea is probably wrong, then we can plan ahead for it and say, well what are four other ones? And and I might looking around at other products, other versions of what I'm doing so I can learn faster.”
— Mark Pincus

planned upfront, tested weekly per

Before you start

· one isolated new element
· competitive teardown of similar products
· testing capacity for rapid variant swaps

Build a culture of building it wrong — do not build it right

Outcome: Make fast, deliberately imperfect builds the cultural default to get signal early.

Context: Pincus advises building a culture of building it wrong rather than right — turn it on rough, get signal, and avoid the over-commitment that comes from polishing before learning.

“I like to say, do not build it. Right? Yeah. Build a culture of building it wrong. Build it. Right. Presume some level of success.”
— Mark Pincus

days to a few weeks per cycle per

Before you start

· leadership permission to ship rough
· no ego attachment to the current idea
· a defined learning question per build

Test at the top of the funnel with just a landing page and an ad link

Outcome: Validate demand with an ad and landing page before building anything downstream.

Context: Pincus's cheapest test: before building the product, put up just the landing page and the ad link and measure whether anyone clicks. Do not be embarrassed and do not build the rest.

“is anyone even going to click on this? Don't be embarrassed. Don't build the rest of this. Just the landing page, just the ad link and nothing else.”
— Mark Pincus

days per test per

Before you start

· ad account and small test budget
· landing-page builder
· a defined click/conversion threshold

Build levels one to ten and see if it works

Outcome: Ship a bounded early slice and watch whether users advance through it.

Context: Hoffman cites a classic Zynga play he adopted: build just levels one through ten and see if it works, using engagement and progression as the signal on the core loop before building the rest.

“It could be things around we will build level one to 10, see if it works. Yes. Right. Which is kind of a, like I learned from Zynga and kind of a classic.”
— Reid Hoffman

weeks per

Before you start

· a definable early slice
· progression/engagement instrumentation
· willingness to stop if the slice fails

Turn it on only to the level that you will learn

Outcome: Scope every release to the minimum exposure that produces a learning signal.

Context: Instead of launching, Pincus turns a product on to exactly the level needed to learn — a reframe that removes launch pressure and keeps the focus on signal, tolerating embarrassment as proof the timing is early enough.

“launch is the wrong word. It's like you should turn it on to the level that you'll learn. Yes.”
— Mark Pincus

as early as embarrassment allows per

Before you start

· a crisp learning question
· instrumentation to read the signal
· tolerance for shipping something embarrassing

Copy the proven exactly — stay as close as legally possible

Outcome: For anything proven, copy it exactly and change nothing.

Context: Pincus's rule for the proven bucket: copy it legally and artfully, do not change a single thing, and stay as close as possible without doing anything illegal. Changing the proven yields false-negative signal and failure for the wrong reasons.

“you should copy those legally if you do it artfully, people don't realize it's a copy. But you should not change a A single thing.”
— Mark Pincus

upfront before build per

Before you start

· clear read of what the market has proven
· legal check on IP boundaries
· discipline to leave working things alone

Decision Moments

Actual decisions, real outcomes

Specific decisions narrated in the episode with their outcomes and transferable lessons.

In 2002-03, during the internet nuclear winter, Silicon Valley concluded the consumer internet was over and split toward enterprise and clean tech. Pincus (starting Tribe) and Hoffman (starting LinkedIn) had to decide where to place their bets.

Did: They rejected the consensus, bet that the consumer internet had "just started," became the earliest web-two builders, and formed an angel-investing group writing checks into consumer experiments like Friendster — not because they thought it would win, but because the experiment needed to happen for the consumer internet.Outcome: LinkedIn and the broader web-two consumer wave vindicated the contrarian bet; Friendster's growth (with no virality expertise) proved the latent consumer demand was real — "the tsunami moved the boat."

At the start of a platform shift, the herd's flight to enterprise underprices the consumer wave. Pincus and Hoffman frame the AI era as the same moment repeating.

Part of an emerging decision pattern across multiple episodes

Raising money for Zynga, a junior partner at a VC that had backed a successful prior outcome told Pincus he worried Pincus was "not coachable" — implied advice to defer to investor guidance.

Did: Pincus rejected the implied advice. He noted he has many coaches he chose himself, but refused to be coachable by a VC who had never held the operator's job or sat in the seat, declining to pretend he would learn from the person with money.Outcome: Pincus counts this as advice he is glad he ignored; Zynga went on to build household-name products (FarmVille, Words With Friends). Hoffman agrees, calling reflexive deference to pattern-recognizing VCs a classic error.

Choose coaches by relevant operating experience, not by who holds capital. Deferring to money as authority can override a founder's better-informed instinct.

Part of an emerging decision pattern across multiple episodes

Deciding when to expose a new product, teams repeatedly defaulted to polishing until "it's ready," while AI now compresses time-to-prototype from three years to three months.

Did: Pincus advocates the opposite discipline: build a culture of building it wrong, turn the product on only to the level needed to learn, and treat embarrassment as proof the timing is early enough — rejecting "viable" as a weak, hope-laden bar in favor of conviction.Outcome: The doctrine keeps teams unattached and learning fast; its failure mode is "going native the wrong native," where sunk-cost attachment blocks the necessary kill.

Launch to learn at the point of embarrassment; build on conviction, not viability. Guard against sunk-cost attachment to a quick build.

Part of an emerging decision pattern across multiple episodes

Tensions surfaced

Contradictions and trade-offs the episode raises — judgment calls a thoughtful operator has to navigate.

Tension

Copy the proven faithfully vs. the user must never feel it is a copy

Preserve proven mechanics exactly while making the surface feel original.

Two true claims collide: stay as close as legally possible to the proven, yet ensure the end user never feels it is a blatant copy. The doctrine is to copy the underlying mechanic faithfully while making the presentation feel non-derivative — the "magic trick."

Copy the engine, restyle the shell.

“Stay as close to it as possible without doing anything that's illegal. Yeah. The real art of it is that the end user doesn't feel like it's a blatant copy. That's the real magic trick.”Mark Pincus

Tension

Enterprise is where the money is vs. the consumer revolution is the bigger wave

Both enterprise and consumer will be big in AI, but consumer is the underpriced side.

In the internet winter the enterprise-first crowd was partly right (cloud/SaaS did win) but wrong to dismiss consumer. Hoffman resolves the tension: enterprise will be big and consumer will be big — but almost nobody is building for the consumer wave now, making it the asymmetric opportunity.

Take the neglected consumer side of the AI bet.

“everyone's talking, well, well, okay, enterprise is where the money is. That was the same thing back in the internet winner. Yep. You know, et cetera. And like, no, no, no. Consumer is gonna be big. Now Enterprise will be big too.”Reid Hoffman

Corpus connection

Where this episode fits for retrieval

What kinds of decisions this briefing is best pulled into.

Primary decisions

• product-strategy
• strategic-bet
• launch