From The American:
Rich Karlgaard, the technology entrepreneur who is publisher of Forbes, tells the story of a trip he took with Microsoft’s Bill Gates in the early 1990s. On the flight, he asked Gates, “Who is your chief competitor?”
“Goldman Sachs” was Gates’s surprising reply.
Gates went on to explain that he was in the “IQ business.” Microsoft needed the best brains available to make top-shelf software. His primary rivals for the smartest kids in America were elite investment banks such as Goldman or Morgan Stanley.
“Microsoft must win the IQ war,” Gates said, “or we won’t have a future.”
Contrast this with open-source-leader Eric Raymond’s beliefs (expressed in this talk) about software development. He repeats the idea that “with enough eyeballs all bugs are shallow” — implicitly meaning enough diverse eyeballs. That I am writing this with Firefox gives some sense of who (Gates or Raymond) was more realistic.
Part 1. Charles Murray vs Charles Murray. How important is IQ?
Podcast: Play in new window | Download
Keep in mind that Firefox is based on the old Mozilla code, which was an open-sourced version of Netscape Navigator, which was developed by one of Microsoft’s fiercest competitors in the early browser wars. Maybe not the best example.
The open source method has produced a small number of very good pieces of software, and a large number of very bad ones. There just aren’t enough eyeballs to go around.
Interesting. I also use R all the time — also open-source.
pdf23ds,
if Linux and Apache are examples of very good pieces of OSS, then what are examples of very bad ones? Thanks in advance.
Just because software is Open Source doesn’t mean that lots of people have worked on it. (And just because lots of people have worked on a piece of software doesn’t mean it is any good.) There are more than 150,000 Open Source projects on SourceForge alone. If you pick one at random, you are much more likely to find a bad one than a good one.
Talk to anyone in the software business. They will tell you that one good programmer is better than 5, or even 10, mediocre ones. Look at Google, and how successful they are. They’ve taken the Gates approach. They’ve hired the smartest of the smartest, paid them much more than their competitors, treated them like kings. And look at the results.
Open source software isn’t a product of the masses working together, it isn’t some emergent phenomenon. And fixing bugs is a very different thing from planning and constructing a piece of first-class software.
In some fields of endeavor, IQ matters a lot. Writing computer software is one of these.
Hmm. Lots of companies try to hire smart people. Those with more money hire more of them. Rich companies like Google hire lots of smart people, yes. But I imagine Google looks for other qualities, too, such as imagination and perseverance.
I suspect that to produce really good software requires several or even many different talents and that only a few of them fall under the term “high IQ”. In my experience, the quality of Microsoft products is remarkably low given the resources at their disposal; a narrow emphasis on IQ may one reason why — but that’s just a guess since I am so far away.
Google doesn’t hire smart people because they are rich, they are rich because they hired the smartest people. Perseverance may be a separate quality from IQ, but imagination, at least in the realm of software development, is not. You need the IQ to be imaginative in that realm. Really in any realm. Just compare the imaginative life of a high IQ child from an average child. It’s worlds apart.
I actually think being in Washington keeps Microsoft from attracting the smartest people. However, Microsoft’s goal has never been to make the best products on earth, it has been to make the most money. And they’ve succeeded at that.
Programming ability is fairly easy to measure directly, so companies will generally do the obvious and correct thing and measure it directly rather than through the proxy of IQ. There doesn’t seem to be much data on how well IQ predicts programming ability, but who needs to predict something trivially measured?