Another thing to consider when looking at things like Quicksilver,
Enso, etc. is how they (or variants on them) might support voice as a
command channel. I think many of us "old hands" expected that voice-
based interfaces would be far more common by 2008 than they are.
Maybe one of the reasons we don't see mainstream voice-based
interfaces yet is that they don't splice neatly on the GUI / WIMP
paradigm, which is (unlike CLIs and QS) oriented toward recognizing,
not recalling, commands and items.
Granted, we still have the problems of ambient noise, not-quite-
accurate speech recognition and social permissibility (do you want to
be in the cube next to someone saying "Excel. New. Template. Time
log." all day?), but an interface like Quicksilver's, where the size
of the recognizable vocabulary is constrained by context, may lend
itself well to augmentation by voice.
Imagine that in addition to a QS hot key, you had a "hot word" that
you could say to invoke the QS UI, and then you could speak the items
that you would otherwise be invoking by typing. Of course the UI would
allow you to mix keystrokes and voice, so if you said X and its first
match for X wasn't the one you wanted, you could use the arrows to
select another one (or, perhaps, say "no" until the one you wanted was
I don't know whether you'd be better off splicing voice onto an
existing launcher or starting fresh, borrowing QS / Enso / etc. ideas
into a new UI that had QS's responsiveness and expressive power, but
was more optimized for voice. Of course, OS X already has some voice
command ability, but I don't get the impression that many people use
it... would a different, QS-like approach be more eagerly accepted, or
is the problem with voice more basic?