Your Red Flag Is My Term Sheet

Psychology has a tidy name for a messy cluster of traits: grandiose self-belief, comfort with manipulation, a light relationship with guilt. The Dark Triad. Narcissism, Machiavellianism, and psychopathy, in their subclinical, wears-a-Patagonia-vest doses. Not movie villains. The person in the room everyone ends up agreeing with, and can't reconstruct why on the drive home.

Venture capital has a name for this cluster too. We call it conviction.

A spicy thing to say about my own industry, so let me pay for it with data instead of vibes.

The pipeline is tilted before anyone pitches.

In 2021 the Journal of Business Venturing ran a meta-analysis with a title I love: "Fatal attraction." Thirty-nine samples, nearly twelve thousand people. All three Dark Triad traits predict wanting to start a company better than they predict being any good at it. Narcissism correlates 0.24 with entrepreneurial intention and a limp 0.09 with performance. Machiavellianism: 0.16 on the way in, minus 0.22 on results. Psychopathy: 0.17 in, minus 0.10 out.

Read those pairs again. The traits cluster at the entrance of the funnel and stop paying almost immediately inside it. "Just pick the impressive ones" starts with a thumb already on the scale.

It gets worse above startup-land. Psychologists ran 203 corporate executives through the standard psychopathy checklist, co-authored by the man who built the instrument. About 4% cleared the clinical research threshold; community samples sit near 1%. The prevalence isn't even the scary part. In the same study, psychopathy scores correlated positively with the companies' own ratings of charisma and presentation, and negatively with ratings of actual performance. The scorecards rewarded the trait and penalized the substance. The authors put it plainly: "charm and grandiosity can be mistaken for self-confidence or a charismatic leadership style."

Investors write checks for a demo.

So what happens when that pipeline hits the filter? Two economists ran machine learning over 1,139 real pitch videos from five accelerators (YC and Techstars among them) and published the result in the Journal of Finance. One standard deviation more passion in the pitch: 35% higher odds of getting funded. Fine, you say, passion probably proxies for something real.

It doesn't. Conditional on getting funded, not one positive pitch feature predicted better long-run outcomes, and plenty predicted worse: more failure, less follow-on funding. The passionate pitches raised more and delivered less. Then they decomposed why investors keep paying, with an incentivized experiment, and got the number I can't put down: 82% of the effect was inaccurate beliefs. Only 18% was taste. We aren't buying charisma with our eyes open. We believe the demo, and we are measurably wrong.

Every engineer knows this bug. Confidence was a proxy for quality. The proxy got promoted into the target. And the people who optimize the proxy hardest are exactly the ones you least want to select. Goodhart's law, running on wetware, with a checkbook.

The pattern-matching is overfit too: trained on a dataset containing one Steve Jobs and survivorship bias all the way down. Nobody writes case studies about the hundred distortion fields that distorted nothing but a cap table.

Now the steelman, because there is one and it's good. Narcissistic CEOs, in the best studies we have, don't produce worse average outcomes. They produce extreme ones: big wins or big losses, variance up, mean flat. A power-law fund is structurally long volatility, so buying variance at a flat mean is defensible portfolio math. Except the 82/18 split says we're not doing it on purpose. And the pitch study couldn't test whether the home-run tail pays for the pattern, because home runs are too rare to measure. The one respectable defense is the one thing nobody has the data to prove.

92% of VCs use AI to find companies. 58% of buyers start there, too. If you're not showing up in AI answers, you're invisible before the conversation even starts. Join HubSpot for Startups, Anthropic, and Marketing Against the Grain on July 30 (11 am ET) for a live AEO teardown. Real startup. Real recs. Register and unlock the free Startup Visibility Bundle.

Register Now

The dictionary does the laundering.

This is where a measurement error hardens into a culture. Follow the language for forty years.

1981: an Apple engineer named Bud Tribble coins "reality distortion field" to describe working for Steve Jobs. Read the original definition slowly: the ability "to convince himself, and others around him, to believe almost anything with a mix of charm, charisma, bravado, hyperbole, marketing, appeasement and persistence." That is a warning label. Engineers invented the term to describe being manipulated. The industry reissued it as a compliment.

2000: John Doerr, the most famous VC alive, tells the world at the exact top of the dot-com bubble that great ventures are "led by missionaries, not mercenaries." Founder psychology becomes item one on the checklist.

2010: Paul Graham writes that the best founders have "a piratical gleam in their eye," that they "delight in breaking rules," that they're "not Goody Two-Shoes type good." YC's application literally asks for a time you hacked a system to your advantage, and weights the answer heavily. (In fairness: the same essay screens hard for cofounder friendship. The filter isn't monstrous. It's confused.)

2022: Marc Andreessen announces a $350 million check, his firm's largest ever at the time, to Adam Neumann, fresh off riding WeWork from a $47 billion valuation to $8 billion and exiting with roughly a billion dollars. The announcement calls him "a visionary leader who revolutionized the second largest asset class in the world" and celebrates repeat founders "growing from lessons learned." Asked about it onstage that fall, a16z's Chris Dixon: "we just don't rely on books and movies for our diligence... we came to a different conclusion on a lot of what happened."

And the cleanest specimen in the jar: Sequoia published a glowing profile of Sam Bankman-Fried whose actual title said he had a savior complex, and that maybe you should get one too. The trait was the headline, on the website of the most storied fund on earth. It stayed up until FTX collapsed, then vanished within a day of the stake being marked to zero. On the way up, a savior complex you should emulate. On the way down, a 404.

Careful, though. I can't diagnose Neumann or SBF or anybody from a conference stage, and neither can the researchers (fraud verdicts are the court's department, not the DSM's). Nobody can, at pitch distance. That's the point. So the only thing standing between "conviction" and the checklist is vocabulary, and our industry's dictionary stocks only flattering translations until the down round.

Which leaves me where, the engineer who now sits on the listening side of pitches?

Go back to Tribble's definition one more time. "To convince himself, and others around him." Himself comes first. That's what makes this genuinely hard: the founder with real, earned conviction and the founder running the checklist emit the same signal at minute ten of a pitch. One built the belief from years of evidence. The other manufactures it on demand, for themselves first. Same eye contact. Same certainty. Same feeling in the room that the future already happened and you're the last to hear.

The founders I've gotten most excited about this year gave me that feeling. I went back through my pitch notes this week. The phrase I keep finding, in my own handwriting: "made it feel inevitable."

The definition I opened with was never a diagnosis of them. It's a description of what works on me.

— SWEdonym