there is a reason aphorisms are often uttered with the finality of scripture. their words are designed for utility—short, unambiguous, unburdened by the poetic excesses of philosophy. the most famous maxim, “make stuff people want,” is at once self-evident and insidious. it is crisp, memorable, seemingly unassailable in its logic. and yet, upon closer examination, it reveals itself to be an incomplete truth, one that, if taken at face value, can lead even the most ambitious creators astray.

the problem? people rarely know what they want.

the verbosity of humanity, the tyranny of consensus

humans, as a species, are verbose—not just in speech, but in thought, in want, in expectation. they mistake the familiar for the desirable, the incremental for the necessary. they articulate their needs in sprawling, contradictory declarations, seeking refinement over revolution.

if one were to build purely by listening to what people say they want, one would find themselves in an endless loop of marginal improvements—faster horses, not automobiles; better landlines, not the internet.

take chatgpt, for instance. had you asked the average person five years ago if they wanted an ai capable of composing essays, writing code, or simulating conversations indistinguishable from human thought, the response would have been one of skepticism, even rejection.

“why would we need that?”

“who would trust a machine to generate knowledge?”

“isn’t this just another gimmick?”

and yet, the moment it existed, its necessity became undeniable.

this is the fundamental flaw in yc’s maxim—it assumes that demand is pre-existing, that desire is static, that people know what they want. but the reality is often the opposite: the greatest products create new desires rather than merely serving old ones.

the technology no one knew they wanted

openai, unlike many yc-backed startups, did not simply respond to demand; it created it. prior to chatgpt’s release, ai was seen largely as a theoretical curiosity, confined to research labs and sci-fi fantasies. the public, when consulted, voiced its hesitations: ai was either unnecessary, untrustworthy, or dangerous.

but the moment it arrived, the world reorganized itself around it. students, programmers, writers, businesses—all found themselves unable to return to a pre-chatgpt reality. its utility was not dictated by existing demand but by its own inevitability.

it did not make what people wanted; it made what people, once exposed, could no longer imagine living without.

the fallacy of listening too closely?

paul graham, one of yc’s founding figures, speaks often of default skepticism—the way truly disruptive ideas are met with initial rejection. he writes about how airbnb was dismissed outright, how stripe was deemed unnecessary in a world dominated by paypal.

but the error here is in assuming that these products were always wanted. they weren’t. at least, not in any way that was clearly articulated. they had to prove their own necessity.

this is the lesson buried beneath yc’s dictum: the best founders do not simply make what people saythey want. they decipher latent desires, things that exist just beyond the periphery of current imagination.

this is the true art of innovation.

the real meaning of “make stuff people want”

perhaps the yc maxim should be rewritten—not in the crisp, tweetable format preferred by the startup faithful, but in a way that acknowledges the complexity of human nature:

“make what, once it exists, people cannot imagine having lived without.”

for to follow demand too closely is to be shackled to the present. but to build in anticipation of what will be wanted tomorrow, despite the skepticism of today—that is where true creation lies.

chatgpt, the automobile, the internet, airbnb—none of these were built because the people demanded them. they were built because their very existence reconfigured reality, forcing humanity to catch up.

so build, not just for the wants of now, but for the inevitabilities of the future.

🦧