Conviction vs hubris

1. Whose money burns, and is the number written down? Conviction prices its downside and carries it personally. Sara Blakely risked $5,000 and kept her day job. Bezos sized the Fire Phone so total failure was a write-down, then promoted the executive who ran it, because the bet was sized to be losable. Hubris runs unpriced or on other people's money: Neumann had extracted roughly $700M personally before the S-1 ever met the public. The worst case written down, in numbers, carried by the person making the bet: that's the first marker.

2. Private evidence, fact or feeling. The only legitimate ground for defying consensus is something specific you know that the other side can't see. Zuckerberg refused Yahoo's $1B holding News Feed: built, unlaunched, with engagement data behind it. In the CRM, as it were. Yang refused Microsoft's $44.6B holding a feeling about a company with his own name on it. Same refusal, opposite evidence class. The marker isn't whether you have a reason; everyone has a reason. It's whether the reason lives in a system somewhere or in your chest, and whether you'll say which out loud.

3. What would change your mind, and has anything ever? Conviction updates on evidence and defends the destination. Ego defends the route, because the route has become the self. Hastings killed Qwikster 23 days after announcing it, publicly, while keeping the streaming thesis: route sacrificed, destination held. Ron Johnson had seventeen consecutive months of collapsing sales at JCPenney and never ran a single test. The marker with teeth is historical: when disconfirming evidence showed up before, did you move? Your track record of updating predicts this certainty better than anything you currently feel about it.

4. Would you still be you if it failed? The sharpest tell in the record is identity fusion. Zuckerberg at 22, asked what he'd do if the company died: build another one, he liked building them. The company was replaceable; the mission wasn't. Against that, Yang defending the company bearing his name. When the decision and the self are the same object, evidence against the decision arrives as an attack on the self, and your immune system handles it instead of your mind. If the honest answer to "what's left of you if this fails completely" is "I don't know," the bet needs restructuring before it needs making.

5. Who's allowed to tell you you're wrong, and have they, recently? Conviction seeks priced, adversarial feedback: paying customers, public markets, the engineer who quits over the roadmap. Hubris curates friendly, unpriced feedback: the one believer with a checkbook, the board that already cashed out, the patron urging more madness. The test is concrete: who has genuinely changed your mind on something that mattered in the last six months? Nobody means you don't have advisors, you have an audience.

6. Can anyone call your loan while you wait to be right? Being early only works if you can stay solvent while the world catches up. Huang could absorb Wall Street's contempt for a decade because gaming cash flow funded CUDA and nobody could call the loan. Bets funded by narrative (the next round, the patron's continued belief) select for story-management over truth-management, because the story is the collateral. A margin call attached to your conviction isn't a reason to fold; it's a date, and it belongs in writing.

Here's the part that should genuinely reassure you: worrying about this is itself a good marker. The founders who got it right often worried their call was ego-driven, took the worry seriously, and committed only after the worry lost the argument on the merits. The dangerous state isn't doubt. It's certainty that never had to survive an interrogation: the certainty that arrived instantly, never wavered, and got offended when questioned.

So the concept converts to a practice in one move: before any contrarian or irreversible call, run the six markers as questions, in writing, and score them honestly. Six passes: commit, and write the memo. Four or five: commit with named repairs (usually: write the downside, collect the dissent, set the wire). Three or fewer: the certainty is structurally unsupported, whatever it feels like. That doesn't mean the call is wrong. It means you haven't yet earned the right to be sure, and the next move is building structure, not gathering more feeling.