Patterns in Search Design Re: [Sigia-l] re: the future of search

[Scott Berkun]

By "single set of design choices" all I meant was that these
goals/patterns/user needs pull against each other. The more you try to
satisfy *all* of these requirements in a single interface, the more
constrained you'll be at satisfying any individual one of them. That's
the nature of design - tradeoffs  that's all that I meant. (I wrote
about this in detail here - the myth of optimal web design -
http://www.uiweb.com/issues/issue17.htm).  

The idea of shifting the model (sort of what you describe below), or
adapting the ui to the task has it's own problems. While we certainly
can shift things around to map the design to match the different user
need/behavior at a given time, we'd create an experience that was
non-deterministic, and no matter how smart it was, some percentage of
folks would have trouble managing those transitions (perhaps that's a
worthwhile sacrifice). Or worse, they'll use the wrong model for the
wrong kind of task.  I could imagine automating this completely, and
automatically derive the user's context and intention from the query,
but given the average query length is still pretty short (2 to 3 words),
it's hard to derive much, or with any confidence. 

So if we can't derive enough from the query, we still have the
additional overhead of forcing people to make more upfront choices
before they can type in their query search. The ubiquity of the search
box has trained folks to get going on their search very quickly - We'd
have to compete with the receptiveness of the search box (that
promiscuous little ui :). Even with the offer of better results, many
folks wouldn't bother to invest the time - you'd still need that little
search box.

In the search assistant for Win IE5/6 (click the search button) we tried
a very primitive version of this - the search box was paired with a set
of radio buttons that allowed the user to tell us what kind of search
they wanted to do, and that was mapped to a taxonomy of providers of
those services. We learned that we had to default to general web
searching because of the natural habit described above. Even after
people used the assistant to perform a more specialized search, and
agreed that they received better results that way, they still tended to
just type stuff in and deal with the results. Only when they failed with
the fast/basic method would they give a more involved approach a try -
I suppose this is just human nature. We make the same kind of decisions
all the time in the real world.

-Scott

Scott Berkun
Design & Usability Training Manager
Microsoft Corporation

-----Original Message-----
From: Peter Merholz [mailto:peterme at peterme.com] 
Sent: Sunday, July 28, 2002 10:14 PM
To: sigia-l at asis.org
Subject: Patterns in Search Design Re: [Sigia-l] re: the future of
search


Reading this thread, I'm quite surprised that this community's
(typically
misguided) fascination with patterns hasn't come up.

For one, it's clearly silly to say, "Search will be [this]" or "search
will be [that]." Because the kinds of searching and the contexts for
searching are too variable.

Scott got the closest with his comments on search as chore, fuzzy chore,
serendipity, and research, but the desire for a "single set of design
and engineering decisions" strikes me as an odd conclusion.

One thing I can pretty much guarantee -- the future of semantic
searching (the kind typically done on a place like google) has NOTHING
to do with kartoo-like visualization. Gack. I wish people would stop
wasting time with that shit. (I tend to feel the same way about the
Semantic Web.)

I think the challenge we face is developing a taxonomy of search tasks
and contexts, and match those up with patterns/guidelines/whathaveyou
for search design. SO: searching in a homogenous and large set of
resources? Use facets? Exploring a single concept in a vast set of
different types of information? Something like Google. Want to know what
restaurants are near you? Use this map! Etc. etc.

Sounds like something for an academic institution. I'd advise, oh,
Andrew Dillon to force graduate students to figure this out.

--peter
==========================
Join me on Adaptive Path's Tour 2002 http://adaptivepath.com/events

------------
When replying, please *trim your post* as much as possible.

*Plain text, please; NO Attachments

ASIST Annual Meeting: http://www.asis.org/Conferences/AM02/index.html
ASIST SIG IA website: http://www.asis.org/SIG/SIGIA/index.html
_______________________________________________
Sigia-l mailing list -- post to: Sigia-l at asis.org
Changes to subscription: http://mail.asis.org/mailman/listinfo/sigia-l