Optimizing Proven Systems Through the Proven--Better--New Framework

The most successful products do not reinvent the wheel. Instead, they optimize friction points in systems that are already proven to work. Mark Pincus uses his "Proven--Better--New" framework to show that a founder’s ego is often the greatest risk to a startup. The desire to be entirely new in a market that rewards familiarity can be fatal. By treating product development as a series of failure machines rather than a quest for a single visionary breakthrough, entrepreneurs can avoid the high costs of discovery. This conversation acts as a manual for founders who prefer the durable advantage of deep, pixel-level iteration over the vanity of originality. It is essential reading for those moving from the app-based internet to an agentic future, where distribution is difficult and trust is the only remaining currency.

The Hidden Cost of the "All New" Illusion

In this conversation, Mark Pincus argues that the industry obsession with all-new ideas is a statistical trap. He notes that 100% of new apps launched in the App Store last year failed to reach the top 10, which shows that consumers crave familiarity with an improved experience more than they crave novelty. The "Proven--Better--New" framework is Pincus’s response to this: study a proven success down to the pixel, make an improvement that users recognize as better immediately, and isolate the new innovation to a small, testable variable.

One of my mantras that still repeated in the hallways at Zynga is all new fails. And that might sound like a beatdown if you start with that mantra you won't be disappointed because there's lots and lots of statistical proof that all new fails.

-- Mark Pincus

This approach creates a competitive moat because it is unglamorous. Most teams are too busy chasing big moments to perform the boring work of benchmarking existing competitors. Pincus points out that Slack succeeded not by inventing a new communication paradigm, but by applying a game-like feel to a boring enterprise category that HipChat had already validated.

Why Obvious Fixes Create Lasting Moats

Pincus emphasizes that "obviously better" is not about what the founder thinks is cool. It is about what 10 out of 10 users would recognize as an improvement without explanation. At Zynga, this meant removing clicks. While competitors forced users through download gates, Zynga’s poker games were accessible instantly. This simple reduction in friction, a better improvement on an existing proven game, drove massive adoption.

If you can perfectly copy a proven successful app and no one thinks it's a copy, that is the fucking magic trick.

-- Mark Pincus

The downstream effect of this strategy is a feedback loop of data. By treating the product as a failure machine, Pincus’s teams could test hundreds of variants in a week. When a product is inconsequential, like a game that has fallen below revenue targets, it becomes the perfect laboratory for innovation because the team is no longer paralyzed by the fear of damaging a major brand.

The System Responds: From Virality to Trust

Pincus looks back on his early social network, Tribe, as a perfect case study in what not to do. Despite rapid, viral growth that outpaced LinkedIn, Tribe lacked retention and trust. Pincus admits he ignored his users' desire for community because it did not align with his pre-determined business model.

This reveals a critical systems-level insight: virality is not a substitute for retention or trust. Platforms like Facebook succeeded by building a social membrane through trusted networks, whereas Tribe’s open, rowdy platform attracted users but repelled the mainstream audience. In the current era of AI agents, Pincus suggests that the social membrane will return. It will be negotiated by agents navigating trust and access to provide high-signal lead generation, rather than just raw connectivity.

Key Action Items

• Implement a Failure Machine: Stop building full products. Start testing headlines, offers, and fake door links to measure actual demand before writing code. If 25% of users do not click on the offer, do not build the product.
• Adopt the Proven--Better--New Framework: Identify a large, boring industry. Map the incumbent’s user experience click-by-click. Identify one obviously better improvement and one small, novel new feature to test.
• Kill the Ego, Not the Instinct: If a product is failing, preserve the underlying instinct, but stop funding the specific expression of it. Pincus notes that even after four years of a passion project, it is better to open-source the engine and move on than to keep funding a sinking ship.
• Build for Agentic Networks: As AI agents begin to roam the internet, focus on building tools for agent-to-agent interaction. Look for ways to create marketplaces where agents earn credits by solving challenges for other agents.
• Prioritize Enterprise Over Consumer: Given the current difficulty of app discovery, focus on prosumer or enterprise AI tools. These markets have higher willingness to pay and do not require the same viral discovery mechanics that currently fail in the App Store.