Law III, The Surface Captures Attention; the Chain Captures Power

The third law is the most counterintuitive of the three, because it appears to contradict everything modern product culture has taught for the last decade. A beautiful UI may get users. It will not, on its own, keep them. The surface captures attention, fast, measurable, satisfying to ship. The chain captures power, slow, invisible to most users, and the only thing that actually compounds over the next decade.

Consider two products in the same category. Gamma generates presentations from a prompt. So does Replit's agent, in a different domain. Both look superficially similar: prompt in, structured artifact out, polished UI in the middle. Gamma owns L7, the surface, and a thin L5 templating layer. The rest is rented. Replit owns L4 (the hosting and distribution), L5 (the agent that writes the code), L6 (the orchestration between editor, runtime, and deployment), and L8 (the memory of every project the user has ever built). Same prompt-to-output pattern. Different fates. The surface is identical. The chain is not.

This is why feature parity is a lie. Two products can look identical on the surface and have radically different futures, because what determines the future is not what the user sees. It is what sits underneath the user's view: which layers the product actually owns, which it rents, and how much friction would be required to leave. The user does not perceive the chain. The user perceives the surface. But the user's decision to stay, year over year, is governed by the chain.

There is a reason surface-only products dominate early product metrics. Surface is cheap to build, fast to iterate, easy to demo, and immediately legible to investors. Chain is expensive to build, slow to compound, invisible in screenshots, and hard to pitch. A founder choosing where to invest a Series A is constantly tempted to invest in the surface because the surface produces visible momentum. The chain produces durability that does not show up for two years.

The market eventually reprices this. Jasper's $1.5B mark in 2022 was a market paying for the surface. Jasper's $300M reality in 2024 was the market correcting once it could see what the surface was actually sitting on (nothing). Chegg's $12B peak was the market paying for a surface that, in retrospect, owned no defensible chain. Stack Overflow's traffic collapse was a community-owned surface that had never converted its corpus into an owned chain. Every one of these stories is the market discovering, with a delay, that surface and chain had been priced as if they were the same thing. They are not.

Notice the symmetric error on the other side. Plenty of companies own real chain depth and never invest in surface, the data is great, the integrations are deep, the workflow is embedded, but the product is ugly, slow, and confusing. These companies do not collapse. They stagnate. They become the kind of vendor every customer renews and nobody loves. That is also a real failure mode, just a slower one. Law III is not 'ignore the surface.' Law III is 'do not confuse the surface for the company.'

The practical test: for every product decision, ask which layer you are reinforcing. A new animation reinforces L7. A new template reinforces a thin L5. A new integration that captures usage data into a structured form your competitors cannot access reinforces L1c and L8. A new compliance certification reinforces L3. A new system-of-record write-back reinforces L4. These are not equivalent investments. They look equivalent on a roadmap. They are not equivalent in the market.

The companies that will be cited as winners ten years from now are not the ones with the most polished prompt UI in 2026. They are the ones who used the prompt UI to acquire attention and then quietly built the chain underneath it, proprietary data the model layer cannot reach, workflows the user cannot leave, memory that compounds with every session, gates the platform cannot legally cross. Surface to acquire. Chain to retain. Both, or neither.

Law III, stated as a prediction: the surface captures attention this quarter; the chain captures power this decade. Anything you ship today that does not deepen at least one layer below the surface is, by definition, a quarter of attention you cannot compound. That is fine, in moderation. As a strategy, it is a graveyard.