Measuring things that don’t matter

I’ve grown used to absolutely everything being measured in hardware, but
I’m surprised to see the habit spreading to
software. Take Bench JS, which measures
how fast client-side scripts run in the browser.

To be clear, I have nothing against measuring
performance as a fundamental characteristic of something like a processor. But I can’t think of
a single case where timing how fast
client-side scripts run in a browser would come in handy. Execution speed
across the scripting engines in today’s browsers rarely
matters as much as compatibility with ECMAScript and
the DOM.

Measurements that would be much more interesting and yield
useful information aren’t far removed from the original idea. Why
not compare the speed of certain DOM operations with the original DHTML, as Peter Paul Koch did
in DOM vs. innerHTML?

Related articles

• (April 26, 2004, Milan Kryl) Browser Speed Test — Opera Wins
• (April 25, 2004, A Computer Enthusiast’s Blog) Browser Speed Test — Opera First, IE Second