First off, I'd like to say that the dancing bars in the old defrag UI was a
model. By definition, ALL models are inaccurate. The only question is how
much time and effort you're willing to put into the model and how useful it
needs to be. If you could get a 100% accurate model, it wouldn't be a model
any more: it would be reality (and we'd be gods
). Anyway, suddenly
realizing your model is inaccurate doesn't mean the best answer is to delete
to model and tell everyone they really don't need it. That model gave
information on disk health and process status. What we have now gives
nothing. Unless the accuracy become so bad that the model was always
exactly and precisely wrong, we'd still be better off with the old UI
(something is better than nothing).
Secondly, in your description, you talk about why the inaccuracy exists in
terms of the old UI (1 bar represents X, and four colors). If you've come
to the conclusion that the reality has moved beyond the model's abilities,
it's time to change how you think about the model. Vista has this great new
3D graphical interface. Why not use it? The old UI used 2-bit color
(white, red, blue, green). Why not use the whole 24-bit color space? What
I know about defragging can dance on the head of a pin without jostling the
angels. But, how about a model where each "bar" represents one file and the
24-bit color of that bar represents the percentage of fragmentation in that
file? If I remember the numbers from some of my old Norton virus scans,
I've got something on the order of 800K files on this system. That would be
a whole lot easier to represent than the 26M you mentioned. Plus, who says
the UI has to show the whole model all the time? In the old Windows 95
model (I think), you could scroll for page after page to see in excruciating
detail what was going on. What about showing some level of abstraction in
the background and then having a zoom window in the foreground (assuming
it's even needed)?
My point is, there are other ways to give health and process status
information to the user than by using the old models and paradigms. To say
the old way was too inaccurate and justified ripping it out entirely is like
saying that the atlas on the desk doesn't show the local Starbucks,
therefore it's of no use and should be burned (I'd like to point out that
Google Maps (and I assume MS's equivalent) got around that problem by using
a zoom function). Just like it's unlikely that anyone capable of finding
and opening an atlas would be confused by the "inaccuracy" of the model, I
doubt there are many people capable of finding and starting the defragger
who would be confused by its "inaccuracy." Telling people they don't really
need that atlas anymore since the airlines will get them to their
destination without their "obsessing" with nits like geography also doesn't
cut the mustard.