Why the best startup ideas die in coworking spaces

The essay builds on Sam Altman’s explanation for why Y Combinator deliberately avoids putting founders in shared coworking spaces — a model thousands of incubators tried in the 2010s with no notable successes. Altman’s argument: great ideas are fragile and usually sound bad in their earliest form. A room full of peers acts as a band-pass filter, laughing off the worst ideas but also killing the best ones before they can develop, because founders quietly reshape their work toward what sounds plausible rather than what could matter. The real risk isn’t bad feedback; it’s the social pressure that makes people abandon promising, unconventional directions.

The author reframes the artists’ cliché about solitude — Picasso, Baldwin, Dylan — not as physically sitting alone but as a mental state in which others’ opinions stop interfering and one becomes sensitive to half-formed ‘larval’ ideas. To ground this, the piece turns to the private working notes of mathematician Alexander Grothendieck and filmmaker Ingmar Bergman, who documented their thinking as it happened, mistakes and dead ends included.

Grothendieck’s case is the centerpiece. He spent 29 months writing Récoltes et Semailles, a sprawling, emotionally raw preface probing the psychology of discovery itself, protesting mathematics’ bias toward polished finished theorems over the messy, generative process that produces them. Tellingly, he credits his groundbreaking work not to raw talent — he judged his elite Paris peers more brilliant and better trained — but to a deep contact with himself formed in his late teens, the capacity to think outside the context he was handed. His peers did fine work, but within boundaries already drawn for them.