Bruce Tognazzini is a very experienced UI/UX/HCI guy. I don't always agree with his priorities though, which is especially frustrating because he tends to speak a bit ex cathedra. Recently he picked on Safari's "make the tabs as big as the space" approach in an article Providing Predictable Targets. This was my response:
Compared to a computer, a fighter plane is rather much a “single tasking” device! What on earth (or the sky) makes you think the “second tab” will always have the same information? That the use case for wanting to get to it is so similar every time that the muscle memory effect has meaning?
In this case, Safari made the decision that having a bigger space to read the title of what the content actually is, is more important than providing a consistent target. That’s a reasonable tradeoff… in my mind the drawback is that it LOOKS less like a button at that width.
There are two competing, but still both true, thoughts: “if you can’t measure it, you can’t master it” vs “not everything that counts can be counted, and not everything that can be counted counts”. You pay so much more attention to the former at the expense of the latter, it distorts your view. With Fitts’ Law, you have a very powerful tool for describing very simple, static interfaces. But with, say, the Mac vs Windows menu bars? Sure, Mac is faster to zip the mouse to, but with Windows, I never have to think about which window on the screen the menubar refers to– unlike the Mac. If I have a large sized screen and am using both firefox and chrome, I have to burn a lot of time scanning to see if the menu screen is one or the other. That cognitive time and effort is a lot more significant than the physical motion, but because it’s tougher to make a good test for it (since most tests implicitly tell you what to do so that A/B tests make more sense) it gets much less attention from you.Anyway, I really think the "stopwatch" brigade has its limitations, especially in today's world of computers offering a lot more variety in experience and usage modes.