I’ve said it before but the recent googlex buzz and subsquent dock fandom brings another round of a situation akin to tree-trimbing back button (an inferior solution cemented by user learning and pervasiveness).
While the dock does offer a fun factor, it’s tantalizing close to a true performance optimization but fails to meet the bar. Significant gains are to be had by expanding icon based targets based upon mouse trajectory. Waiting until the mouse is actually over the icon looks pretty, but offers a tiny fraction of the potential performance advantage — only separated by few dozens of lines of code.
I’m serious about the dozens of lines of code. One of the most significant patches submitted to the Optimoz gestures project after it’s initial slashdotting was an optimization of the direction monitoring code.
The little bit of performance gain associated with the dock decreases with expertise and is associated with the verification phase of the move-click process in which the eyes are used to verify that a target has been successfully reached.