NotRocketSurgery

Moore's Law in Reverse

ryan / Thu, 31 Aug 2006 21:04:00 GMT

Tags happiness, mooreslaw, processors, productivity | 4 comments

If the performance of hardware doubles every so many months, why doesn’t the amount of time we spend worrying about the performance of our software get cut in half?

Sure, one might riposte by saying that we’re doing more sophisticated things with our software now than we were that many months ago, so we need to use the processor speed to do more fancy stuff. But I don’t buy it.

Unless you’re writing something with massive data-crunching or vast levels of concurrency and you use, say, C++ or Java to do it, then admit it to yourself: your life is being sacrificed at the altar of Premature Optimization. It’s not just bits of your code; it’s your whole paradigm.

Your blog, your desktop word processor, your personals website, your client’s shopping cart: let’s face it; they aren’t the SETI project. Why are so many programmers who write this stuff being a slave to memory management, static typing, ugly syntax, and all that misery?

If getting more “low level” is always better, then we ought to be writing everything out in machine code ourselves, and our punch-card poking predecessors were the most productive programmers in human history.

Let’s be honest. If your app is running slowly, in 80% of cases, you can throw on another server, or pop in some memory. Those things are cheap. Your happiness, sanity, and productivity aren’t.

All that hard work being done by the diligent folks over at Intel and AMD should buy us something, after all. Why not a little happiness and productivity?

Eight months ago I stopped writing software to be read computers and started writing it to be read by people (myself in particular.)

Let’s figure out how we can make this possible in all domains of programming, not just web applications.

What’s the point of all those megaflops if they can’t make our lives better?