Peter Thiel: Google is undervalued, and CEOs are overpaid

Peter Thiel, the co-founder of PayPal and who rarely gives public interviews, spoke with Michael Arrington at the TechCrunch50 conference today.

Topics ranged from politics to finance and technology. On the item of the stock market, Thiel is concerned that our economy keeps producing bubbles--technology, housing, foreign investing, etc.--yet he does not feel the current technology market is a bubble. Referring to the valuation of Google, Yahoo, Amazon, and other tech stocks, he said, "on a relative basis, they're among the cheapest companies in terms of stock."

But the 2000 bubble, he says, "set technology back 10 years," by discouraging people from getting into technology. Thiel raised the political issue here: Why doesn't our system see and proactively address these bubbles. It's not like they don't have precursor signals.

Arrington engaged Thiel in discussing YouNoodle, a project he's funding that applies quantitative analysis to vetting start-ups. YouNoodle examines a ton of variables, including things like how long the founders have known each other, and then predicts not just success for companies but their three-year valuation. Traditionally an exercise for investors, Thiel says that YouNoodle applies, "a division of labor." Computers, he says, do some things better than humans, and "there are certain quantitative metrics that work very well in startups."

For instance? CEO salary. The lower it is, the more like the company is to succeed, since it indicates alignment with growing equity, and also because it keeps the salary cap for all employees reasonable. "It's been powerfully predictive," Thiel says. And the magic number: $100,000 to $120,000. Above $150,000, "you start to have issues."

Thiel, a ranked chess player in his younger years, has interesting thoughts about artificial intelligence and its benefits and dangers. While people used to think that Chess would never be mastered by a computer since it was a reflection of innate human intelligence, clearly people were mistaken about that. And it's instructive, he says. "How many domains are like Chess, where they can be quantified?" Thiel asked. "My thinking is that there are probably more than we think."

Responding to a question from the audience about computers taking over the world: "My own sense, it's not going to happen. But I was wrong about chess."