Why don't all companies buy developers the best hardware?

Many companies are certifiably insane around this.

Seriously. If you asked 10,000 tech mangers, "Let's say you paid Danica Patrick $100,000,000. Do you think she could win the Indianapolis 500 by riding a bicycle?", I'm sure not one of them would say, "Yes."

And yet a good percentage of these same managers seem to think that highly-paid software developers ought to be just as productive with crappy tools and working conditions as they are with good ones - because, of course, those lazy, feckless programmers are getting paid lots of money and ought to be able to pedal that bicycle faster.

Now, what exactly good tools and working conditions consist of depends on the job to be done. People who code the Linux kernel need different kinds of hardware than web site designers. But if the company can afford it, it's crazy not to get people what they need to be as productive as possible.

One company I worked for had a 9 GB source code base, primarily in C, and the thing we most needed were fast builds. Unfortunately, we were mostly working with hardware that had been mediocre five years before, so people were understandably reluctant to build much other than what they were working on at the moment, and that took its toll via low productivity, quality problems, and broken builds. The company had money to upgrade the hardware, but was strangely stingy about it. They went out of business last summer after blowing through over $100 million because their two biggest clients dropped them after repeatedly missed deadlines. We were asked one time to suggest ways to improve productivity; I presented the same kind of cost-benefit analysis the OP did. It was rejected because management said, "This must be wrong - we can't possibly be that stupid", but the numbers didn't lie.

Another company I worked for had fine computers for the programmers, but insisted everybody work at little tiny desks in a big crowded bullpen with no partitions. That was a problem because a lot of us were working with delicate prototype hardware. There was little room to put it on our desks, and people would walk by, brush it, and knock it on the floor. They also blew through $47 million in VC money and had nothing to show for it.

I'm not saying bad tools and working conditions alone killed those companies. But I am saying paying somebody a lot of money and then expecting them to be productive with bad tools and working conditions is a "canary in the coal mine" for a basically irrational approach to business that's likely to end in tears.

In my experience, the single biggest productivity killer for programmers is getting distracted. For people like me who work mainly with compiled languages, a huge temptation for that is slow builds.

When I hit the "build and run" button, if I know I'll be testing in five seconds, I can zone out. If I know it will be five minutes, I can set myself a timer and do something else, and when the timer goes off I can start testing.

But somewhere in the middle is the evil ditch of boredom-leading-to-time-wasting-activities, like reading blogs and P.SE. At the rates I charge as a consultant, it's worth it for me to throw money at hardware with prodigious specs to keep me out of that ditch. And I daresay it would be worth it for a lot of companies, too. It's just human nature, and I find it much more useful to accept and adapt to normal weaknesses common to all primates than to expect superhuman self-control.