I've been building things for thirty years. Here is the only thing that has never changed.

I started my first company in 1996. Since then I have co-founded, led, or invested in companies across semiconductors, displays, biotechnology, cybersecurity, and wine. Different industries. Different countries. Different stages. Different teams.

Almost everything about how I work has changed across those thirty years. The tools. The markets. The way capital flows. The speed at which technology moves. The profile of the people building the most interesting things. The frameworks I use to evaluate opportunities have been updated so many times they barely resemble what I started with.

One thing has not changed. One pattern I have seen hold across every context, industry, stage, and geography.

The people who build things that last are genuinely curious about being wrong.

Not tolerant of being wrong. Not resigned to it. Curious about it.

There is a difference. Tolerance is acceptance after the fact - a founder who handles failure with grace. Curiosity is something that operates before the fact - a founder who actively wants to find the errors in their own thinking before the market finds them.

I have met many founders who were tolerant of failure. Most of them failed in the same ways repeatedly because they processed what happened after each failure without interrogating the assumptions that produced it. The postmortems were honest but shallow. They identified what went wrong without asking why they hadn't seen it coming.

The founders who actually compounded, who got better in ways that made each successive company harder to destroy,  were the ones who treated their own mistakes as genuinely interesting problems. They looked at a failed prediction and wanted to understand the model error that produced it, not just the outcome.

Why this is harder than it sounds

Building something requires a kind of conviction that is, by nature, resistant to revision. You have to believe enough to continue when the evidence is ambiguous, when the market is unreceptive, when the people around you are skeptical. That belief is not optional. Without it, most things worth building would stop before they started.

The tension is that the same conviction that keeps you going can also keep you wrong longer than you need to be. The skill is holding conviction about the direction while remaining genuinely open about the path, committed to the destination without being attached to a specific route that the evidence is telling you isn't working.

I have found this easier in the vineyard than in the tech world, for a reason that took me years to understand. The vineyard gives you feedback you can't argue with. A block that isn't yielding good berries doesn't the following year. You can have theories about why, but the grapes don't care about your theories. They respond to conditions, not narratives.

Business gives you more room to construct alternative explanations. The market timing was wrong. The customer education wasn't complete. The macro environment intervened. Some of those explanations are true. Some of them are stories we tell ourselves so we don't have to revise the underlying model. The vineyard stripped most of those stories away, which was uncomfortable and useful in equal measure.

The thing that has never changed

Across every company I have built or funded, across semiconductors, displays, and wine, across California, Armenia, and Germany, the pattern holds.

The people who are genuinely curious about being wrong, who treat their own mistaken predictions as data rather than as failures to be defended against, are the ones I would bet on in any context.

Not because they fail less. Often they fail more, at least initially, because they take more risks with their assumptions. But they fail faster, update more quickly, and compound their learning in ways that make each subsequent attempt considerably more likely to succeed.

The market eventually teaches everyone the same lessons. The question is how many times you have to be taught before you stop being surprised by them.

Curiosity about your own errors is the closest thing to a shortcut that I know of.

Everything else I believed in 1996 has been revised. This one hasn't needed to be.