MacBook Air is enough for a work machine

I'm waiting to see what comes out in the next few months, as apparently OCZ and Intel are building drives with the new SandForce chipset. Don't know if the improvements will be significant, though...

But for full-time programming work, I think it is absolutely fucking batshit insane to use anything less than a modern 12-core Mac Pro with 32 GB RAM and an SSD RAID boot volume (at which point multiple 2560 x something monitors are de rigueur).

That's certainly true for [Obj-]C[++] development, where large compiles still take minutes, even on such hardware--and can take an entire lunch break on the Air.

Obj-C coding is my primary background, so buying the fastest possible machine every year has always made sense in that way; perhaps I have a leftover cognitive bias toward buying the fastest possible machine. But, for the past year or so, I've been working mainly with Ruby and Objective-J, which don't have these burdensome compiles. Yet buying the fastest mainstream hardware still makes tons of sense to me.

Running a few hundred unit tests, one second is way better than eight seconds. Running a bunch of convoluted RSpec integration specs, 34 seconds is way, way better than seven minutes. And although I mainly use simple editors like MacVim and BBEdit, I also really like the psychotron-dynamic-heuristic CodeSense-style introspection-based context-aware autocompletion features of RubyMine--a heavyweight Java-based IDE that eats 800MB of RAM and can barely even launch on a MacBook Air, and certainly isn't usable on it. And of course the modern Pro can run Win7, WinXP, and Linux in VMWare in the background without breaking a sweat (that's why 16GB RAM doesn't really suffice).

Don't get me wrong; I think the recent MacBook Air is awesome. I have one, that I use for... well basically for word processing on airplanes (haha). And whatever else on airplanes.

But for working programmers, I think computers are still not nearly 'powerful enough'--there are still huge gains to be had by getting the fastest machine that it is reasonably possible to buy.

For roughly the price of one decently tricked-out Mac Pro, you can buy ten MacBook Airs. And for roughly the price of ten decently tricked-out Mac Pros, you can hire one additional developer.

That notwithstanding, I still think it makes business sense.

Sometimes, in my dreams, an AI built by an alien race visits my house, from the future in another galaxy, and explains to me the precise opportunity cost of a dozen human developers waiting three or four extra seconds, a hundred times a day for a year... and it's more than the ten grand that an appropriately provisioned Mac Pro costs

P.S. A bootstrapped startup where money is really tight is the obvious exception--the exception that proves the rule!

SSDs are interesting in that they get faster as they grow in size. top-tier 64GB SSD drives are quite a bit slower (at least in write speeds) than their 128GB counterparts, which are slower than their 256GB counterparts. Write speeds scale almost linearly with size.

I'd salivate for something a little quicker.