Korva is a scroll of the internet’s best ideas.
Pedigree can't replace a point of view
Pedigree as a Proxy
Research founders are pitching investors who mostly aren't research scientists, so the quality of the story and how legible the company is matters a great deal when it needs heavy capital spending upfront. Plenty of people are trying to invest in technology bets, and how much they understand varies. Many hand their judgment off to signals: who has invested, who gives references, and who the referral sources are.
A Friend's Question
An investor pressed a very good friend to explain how one company would get big. The friend's whole answer was a question: did the investor know the quality of the founder? The investor did, from eight years of knowing the person, and the technical theory and the business still made no sense.
Building Intuition
Not every one of these research bets will work, and the investor asking the questions may be no better at deciding. The aim is still to build an intuition for the bet itself, because having no point of view on the business besides the pedigree of the person is dangerous.
Why do investors lean on founder pedigree?
Early-stage companies have no revenue or history, so investors look for signals that stand in for evidence. Economist Michael Spence shared the 2001 Nobel Prize in part for his work on signaling, which showed how credentials like degrees can convey quality when quality is hard to observe. A famous lab or a strong former employer plays the same role for a founder. Spence's signaling model dates to 1973.
Why do research startups need so much capital?
Training a frontier AI model requires large clusters of specialized chips running for weeks or months. The Stanford AI Index 2024 estimated the training compute cost of OpenAI's GPT-4 at about $78 million and Google's Gemini Ultra at about $191 million. Those costs come before any product can be sold. Researchers also pay for data, engineers, and experiments that fail.
How do investors end up following each other?
Economists Sushil Bikhchandani, David Hirshleifer, and Ivo Welch described information cascades in a 1992 paper. In a cascade, people copy earlier decisions because they assume earlier decision makers knew something, and they set aside their own information. Once a few investors back a company, later ones can pile in without independent analysis. A cascade can start from only a few early choices.
More from Invest Like the Best with Patrick O'Shaughnessy
