Youth vs Taste

Jonathan Goldsmith was auditioning for a role in a new ad campaign centered around "the most interesting man in the world." He almost didn't get it. His agent received the call that they loved the performance, but needed someone younger.

His agent responded, "how can the most interesting man in the world be young?"

He ended up getting the job. Once stated, it seems obvious that to know a thing or two, you have to have been around long enough to do a thing or two.

So why do we keep betting on young founders?

The old edge was stamina#

In the pre-AI startup playbook, a lot of early advantage came from sheer execution volume. Ship more. Iterate faster. Try more channels. Write more code than your competitors. If a founder was young, intense, and all-in, they could no-life their way into product velocity.

When early products were mostly constrained by how much work a tiny team could personally force through, raw stamina covered up a surprising amount of imperfect judgment. Connect that effort to a couple customers who supplied use cases, and sometimes that was enough to stumble into product-market fit.

Investors pattern-matched that. If the game rewards people who can sprint for years, age gets added to the weights.

AI makes execution cheaper#

Now everyone is talking about how AI changes the velocity game, and this part is not hypothetical anymore. A founder can do in a weekend what used to require a small team and a month of coordination. Garry Tan recreated his 2008 startup in 90 hours.

What's more is that agentic coding is not even always deep work. You can dip in and out and still keep a project moving.

That creates a contradiction in the usual founder mythology. We cannot simultaneously believe that youth is a durable velocity advantage and that AI radically democratizes velocity. One of those stories has to be false.

If everyone can ship faster, then "shipping faster" stops being enough of an explanation for why you win. Output still matters, but it compounds less. The bottleneck moves upstream.

The question becomes less "can you build it?" and more "should you build it at all?"

Taste matters more when users want faster horses#

This is where taste starts to matter a lot more, and why Paul Graham talks about it, even if I do not think most people have fully internalized the implications yet.

The tricky part is that AI products are still early enough that customers often do not know what they want yet. You talk to them, and many of them will ask for a faster horse: painkillers for a pre-AI workflow they already understand.

If you listen too literally, they will steer you toward incrementalism. Customers are good at telling you where it hurts. They are much worse at telling you what the new default should feel like.

You cannot outsource taste to your customers. Taste is having a product philosophy and enough conviction to teach that philosophy to the market. It is knowing which parts of feedback to honor, which parts to translate, and which parts to politely ignore.

But taste is expensive#

You do not get it by slamming Celsius and working until 4 a.m. You get it from experience in what works and what does not. From shipping things no one cared about. From underbuilding and from overbuilding. From going down enough dead ends that you develop an instinct for how to use your time.

Because when building is no longer the hard part, the real cost becomes opportunity cost.

That accumulated judgment used to be partially offset by raw output. Now it is harder to compensate for not having it.

That is the young founder paradox in the AI era: the people with the most execution energy often have the least accumulated judgment, right when judgment is becoming the larger source of advantage.

What I would ask founders now#

If I were evaluating founders right now, I would care a lot about a few questions:

• Why this problem? Did you pick it because you felt the pain yourself, or because you saw someone on X claim there was a market there?
• Are you building to a spec, or are you imagining how the world could look different if everyone used your product?
• Are you demoing the first thing you tried, or the best thing you found after killing a bunch of bad ideas?

The old joke in every new tech wave is that recruiters immediately look for people with 5+ years of experience in it.

At the beginning of the mobile wave, no one wrote Objective-C unless you were already a diehard Apple fan, which was not actually a large crowd pre-iPhone. Back then, being a seasoned engineer did not guarantee an advantage over the kind of raw dedication young engineers could bring to a green field.

This wave is the inversion. AI is making execution abundant, which means founder quality gets measured less by output and more by taste. Older founders are suddenly on more equal footing because they are more likely to have already paid for judgment.

For years, the bet was on whoever could run the hardest. Now everyone gets the fast horse. The only question is who knows where to point it.

Maybe this sounds self-serving coming from a 44-year-old founder. That doesn't make it wrong.

If you're still not convinced, ask yourself this: how can the founder with the best taste in the room be the one who’s done the fewest things?