I just read Coding Horror: Actual Performance, Perceived Performance (hat tip to Sterling). The explanation seems plausible and makes a good case for Vista (pre-SP1) on the subject of file copy performance, within the context of the information provided. Of course, that’s not the whole story.
First, consider the fact that the problem isn’t just that file copying seems slower to the user, judging by reports like the Register article Vista sets 2007 land-speed record for copying and deleting. As you’ll see if you read the article, the problem actually involves file copying that never finishes, as well as problems with file deletion — which does not have the indicated problem with write-behind caching indicated by a progress bar.
Second, take the BSD Unix FFS/UFS as an example. Somehow, FFS/UFS is able to manage delayed writes in a manner that doesn’t take three years, and also doesn’t give a mistaken impression that data has already been written to disk when it hasn’t, all at the same time.
The conclusion to which I am inexorably drawn is that Vista’s file copying has problems not addressed by the perceptual performance upgrade, aside from those problems is probably actually better than XP’s on a technical level, and for all these years XP has been lying to you about when file copies finish to keep you happy with MS Windows by making you think it’s faster than it really is.
I guess there isn’t a whole lot of new information in this revelation that Microsoft gives an impression of greater performance by lying to the users of its operating systems. This is just one more piece of evidence of the sort of conflict of interest that occurs when you want the best possible software, but you’re getting it from an organization supported by copyright law in its attempts to “get away with” something — making people think they’re getting their money’s worth, rather than actually giving them their money’s worth.