Sympathetic Stupid

Tuesday, August 30, 2005

Reducing Inequality and Graham's Startup Obsession

Paul Graham is an interesting guy. He's regarded as a guru in the computer science field but is smart enough to realise that compter programming can't exist in a bubble (or a black box). As an enabling tech, it needs to consider the social consequences of the work it enables. However, I think he can be quite wrongheaded.

This one's on inequality.

The basic principle: reducing inequality means taking money from the rich. You can't just make the poor richer (especially by increasing productivity) because that just means bigger markets for the rich to employ and sell to. No net increase in equality. It's kinda hard to decrease productivity of rich people (which would have the right net effect) so you need to take away a portion of their surplus resources and distribute it equally.

Fair enough.

Here's the leap of logic. Risk-reward tradeoff. If the rewards of being rich are less, the risks taken to get there will be proportionately smaller. So logically, there's a whole class of risks which are no longer worth taking; especially starting new companies.

Here's the IT connection. IT startups looked dead after the tech-boom-crash but are still around, and Graham had a successful one. His argument is that reducing inequality means no more venture capital and therefore no more startups. The government can't do it because it's too risky and unjustifiable after the fact. And not only that, that it's no longer worth it for startup founders, because they can't get much out of the hard work and stress of the startup.

So therefore, too much equality means there'll be no new companies, therefore no growth or innovation, because established companies are less likely to innovate than startups (cf Microsoft). Finally, he redefines the problem. Corruption, not wealth, is the problem; the connection between wealth and power.

I can't argue that corruption isn't a problem, and fixing that, while not trivial, is probably doable. But I'm not convinced about the rest.

A single example of innovation from a big company would be Xerox Labs - laser printers, Smalltalk, client/server, networks, GUIs and mice, photocopiers, worms. Innovation from the public sector? DARPA invented the Internet. Here in Australia, the CSIRO just participated in a nanotech breakthrough. I think you can innovate in an institution, if the institutional climate is correct. R&D tax breaks are a good start, incentive for the company to 'throw money away'. Get enough smart people together in one place and magic happens, look at Xerox, or MIT, or Stanford.

And my other issue is the implied claim that the only incentive for smart people is money. Certainly this is true for some, but throughout history, the smartest people have been those motivated primarily by curiosity or elegance - Einstein, Da Vinci, Torvalds. So removing the money as a primary incentive should remove those motivated by money, like the guy who wants to make a slightly better system for measuring click-throughs. It will leave the most important group; those geniuses who will innovate no matter where they are and what they're doing.

The whole article seems to be too strongly coloured by Graham's own experience. He's extremely intelligent, the type of person the human race needs to progress, and he enjoyed his time in a startup and came out of it well. But for each who succeeds there are hundreds as intelligent who fail, often not because of the technology but because of unrelated financial matters related to the economic imperative of quick success. How much of that cash was spent on letterheads, business cards, marketing and organising 'face time' with important people so as to get their money? Startups are a preposterously inefficient road to innovation because they waste so much talent.

On inequality, I think Graham is dead wrong.