What Peter Thiel and I Gave Sam Altman Wasn't the Same Advice. The Contradiction Is the Point.

Sam Altman credits Peter Thiel and me for teaching him a key lesson about investing. It sounds like a neat Silicon Valley anecdote: two famous investors, one protégé, a wisdom transfer. But if you actually pull the threads of what each of us taught, we don't converge on the same thing at all. We point in opposite directions. And the contradiction between us is the interesting part.

Here's the setup. On episode 484 of Invest Like the Best, which aired July 28, 2026, Altman discussed what Thiel and I taught him about building and funding companies. In various interviews throughout 2026, he has echoed lessons from both. From Thiel, the recurring theme is independent thinking - questioning consensus, building things that are truly differentiated, having the conviction to pursue ideas that look strange or unlikely before the rest of the world understands them. Thiel also taught him to hire for talent, ambition, and hard work rather than expertise, a lesson that has been explained as: before you've built the machine, you want someone who can iterate and figure things out. That's a different skill set than the people who run the machines.

From me, the lesson runs the other way. I once told Sam: "People can become formidable, but it's hard to" - and I never finished the sentence. You can supply the ending. Hard to give them an idea. Hard to change the idea. My school of thought is: fund people, not ideas. The idea will change. The person is what carries it through.

Two mentors. One saying you need conviction in a contrarian idea. The other saying the idea is secondary to the person. If you're looking for a tidy synthesis, you won't find one. Or you wouldn't have until you noticed the thing Sam himself tripped over when trying to apply it.

Y Combinator ran an experiment that breaks the "fund people" thesis in a way I probably expected all along. We funded twenty teams of strong founders... who simply didn't yet have an idea. Every single one failed.

This is the part that doesn't fit my own "fund people" frame. You can't just give a great person a blank slate and expect them to find their way to a billion-dollar company. The experiment suggests something I wrote about years later in "Black Swan Farming," my 2012 essay on why startup investing is so counterintuitive. I wrote that in venture capital, effectively all returns are concentrated in a few huge winners - Dropbox and Airbnb alone account for roughly three-quarters of the total value of companies Y Combinator has funded, which comes to around $10 billion. And the best startup ideas tend to look like bad ideas at first, because if a good idea were obviously good, someone else would already have done it.

Peter drew a Venn diagram at Y Combinator that captures this. One circle labeled "seems like a bad idea." One labeled "is a good idea." The intersection is where you want to be. The diagram is useful not because it's profound but because it forces you to acknowledge two things at once: there are good ideas disguised as bad ones, and the vast majority of ideas that seem bad actually are.

So here's the reframe that resolves the contradiction. I didn't actually say "ignore ideas." I said you can't evaluate the idea the way a normal person evaluates an idea - by checking whether it sounds sensible. The person you fund needs to be someone who naturally lives in that Venn diagram intersection. They're not the person who waits for an idea to be handed to them. They're the person who constantly generates ideas, the kind who can see what doesn't exist yet but should. Sam himself has said that the real challenge in building a company isn't the initial idea - it's coming up with new ideas constantly. Every week requires big changes, new directions, fresh approaches, better solutions. Ideas aren't a one-time event. They're a continuous process.

The lesson isn't "fund people over ideas" or "bet on contrarian ideas." The lesson is that the person and the idea-source are the same thing. You fund the person who is an idea engine. You fund the person whose way of thinking generates the kind of contrarian insight Peter's diagram describes, over and over.

This is harder than it sounds because it means evaluating something that looks nothing like expertise. Peter's hiring lesson connects here: before the machine exists, you want someone who can figure out the machine, not someone who knows how to operate one. Expertise is a record of past adaptation. Raw intelligence combined with a compounding ability to generate insights is the ability to adapt to problems that don't exist yet. These are not the same thing.

The counterargument, of course, is that this is survivorship bias dressed up as a philosophy. Every successful founder can point back and construct a narrative about how their "contrarian vision" was always the key. Facebook sounded like a niche waste-of-time to me when I first heard about it - a site for college students with no money to do something that didn't matter. But Microsoft and Apple could be described in the same terms in hindsight. The problem is that for every Facebook, there are thousands of founders who thought they had a unique insight and didn't. The Venn diagram's intersection is tiny. I haven't seen a way to measure how tiny it actually is before you cross the finish line.

I think the test is whether the person keeps generating the kind of insight that moves the needle when the first one doesn't. Most founders' first idea is not the one that makes the company. The ones who build the big winners are the ones whose second, third, and fifth ideas are just as genuinely novel as the first. That's not persistence. That's a different kind of mind.

What this means for someone trying to apply the lesson - whether as an investor, a founder, or just someone watching the market - is simpler than the philosophy suggests. Don't ask whether the idea sounds good. Don't ask whether the person sounds impressive. Ask whether the person's way of thinking is the kind that produces ideas the rest of you can't see. And then ask whether that way of thinking compounds, or whether it was a one-off flash of insight. The first question gets you to the Venn diagram. The second tells you whether you're looking at a founder or a lucky guess. The only way to answer the second one is to wait and see, which is the part of startup investing that makes it so counterintuitive: you won't know for two years whether you were right.