Eric Ries on why good companies rot: it's not corruption, it's 'financial gravity'

The thesis: corruption is the symptom, gravity is the force

This was an Ask HN AMA with Eric Ries — author of The Lean Startup and the new Incorruptible — and the thread is the source, so it's worth reading his actual answers rather than the book blurb. His central metaphor reframes why mission-driven companies decay. Asked how corruption differs from his idea of "financial gravity," he reached for a bridge:

"Corruption = the symptom. Gravity = the force that causes it. … If you ask an engineer 'why did it collapse' you'll be annoyed if they say 'gravity' … if we … notice that the metal bolts have been corroded beyond recognition, we can start to think through what went wrong." — Eric Ries

His diagnosis of why leaders fail to fix it is specific about what they overlook:

"they tend to focus on the typical management/leadership stuff: business model, org chart, strategy, vision. … But there is a deeper layer that tends to get overlooked … structure, governance, boards, the relationship with investors." — Eric Ries

On missions that outlive the founder

A reader pressed him with counter-examples — the Ford and Hewlett foundations drifting from their founders' visions despite being non-profits. Ries didn't claim missions are easy to preserve; he pointed at a specific structure that the data says survives:

"the two-entity structure, where a foundation oversees or governs a for-profit company, is uniquely suited to longevity of a mission … The data shows … such structures are five or six times more likely to live to year 50 compared to a standard for-profit company." — Eric Ries

And on what actually motivated the book, the answer was less management-guru than confession:

"I'm sick and tired of helping people become rich and miserable, creating great companies only for them to be destroyed out of the gates. What is the point of all this carnage? Who is it for?" — Eric Ries

The accidental demo of his own thesis

The thread's best moment was unplanned. Ries linked a progress site he'd had Claude Code generate to "summarize" his hundreds of book-tour interviews. A reader checked it and found the model had quietly optimized for the wrong goal:

"Claude Code has misinterpreted your request to summarize your interviews and events. Instead, it created a marketing and promotional website with not a summary to be found!" — sdellis

It's a small thing, but it rhymes with the whole book: state one goal, build the incentive structure carelessly, get a glossy artifact optimized for something else entirely. On AI more broadly, Ries said The Lean Startup's principles need no revision — "none of them, really" — while conceding "many of the tactics and examples in the book are now a bit dated." The funnel from build-measure-learn to an LLM agent that "learns" from feedback, he argued, is the same loop with faster iteration.