I have a proof-of-concept alias thanks to testing and discussions with mouser.
However, now that I have some interactive experience and I've had some time to reflect, I wonder about the value of this idea is in its current form.
To back up a bit, I think what many of us are interested in is having quick/efficient access to searches that are stored/managed in our browsers. In Firefox (and some other browsers), there is this idea:
"let the user specify -- via the location bar -- a keyword (which identifies a search) followed by search terms"
I think it's probably not too uncommon for people to edit the content of their location bars at least occasionally, there are shortcut keys to access the location bar, and may be the shortcuts are even some of the better known ones (anyone done a survey?). The keyword idea doesn't seem too bad in the context of a web browser if you only have a few searches. I don't think it scales that well though -- coming up with new keywords (and remembering them!) seems uninteresting as an activity in a way similar to coming up with shortcut keys that have not already been used (and remembering them!).
Luckily in FARR, it doesn't seem necessary to use the keywords defined in Firefox -- if you have some already that your fingers memorized, may be it makes some sense to use them -- but why not distinguish searches in FARR using FARR's searching capabilities applied to the names/titles of the searches? I think this scales better and is kinder to your brain -- perhaps you'll even use the time you're not dreaming up non-conflicting keywords (or not remembering them!) in a good way
This is essentially the approach that
Keyser takes -- but it doesn't handle searches associated with bookmark keywords (at least not yet) -- it only handles search bar searches.
So in summary, my current inclination is to focus less on keywords (possibly not even bother) and more on names/titles of searches. The proof-of-concept mentioned above doesn't extract any name/title information so there is nothing implemented toward the revised idea yet.
Any thoughts?
P.S. If there is anyone who'd like to try the proof-of-concept,
here it is. (Contained within should be a README.txt.)