[Sigia-l] People thinking like computers (was ROI/Value of Search Engine...)

[Christopher Fahey [askrom]]

(Jay Linden wrote:)
> > it will always be infeasible for a good search 
> > engine to read the minds of end users who aren't
> > sure of what they seek. 

(lisa colvin replied:)
> Interesting point. Mind-reading = understanding query
> intent. Formal semanticists and philosophers work on
> issues of intentionality, yet cs implementations of
> these theories have been of minimal success. 


For all-purpose search engines like Google.com, I agree that "mind
reading" is technologically out of the question (beyond perhaps Google's
great spelling-correcting algorithm).

But for limited-application search engines like Amazon.com's product
search, it would not be terribly difficult for the designer of the
search engine to give "hints" to the algorithm to make
common-but-useless searches (or phrases that resemble common-but-useless
searches) and take the user to specialized "Guide" pages instead of to
useless piles of search results. For example, in the case of the user
looking for "Digital Camera Film", the designers might program (i.e.,
hard-code) the search engine to detect/intercept the 50 most common
searches that *ought* to give the user the option to go to a "Digital
Camera Memory Guide" page. Searches like "digital camera film", or
"camera memory", "digicam ram", and maybe dozens of others. The results
page of such a search would most prominently feature a link to the Guide
page, plus the normal pile of links to stuff like books with those words
in the title, etc.

This "Guide" page, by the way, already exists at Amazon:
  http://www.amazon.com/exec/obidos/tg/browse/-/697752/

I'm suggesting here that for the time being, a specialized, hard-coded
(that is, a cheating) AI is way better than a generalized (i.e., real)
thinking AI. Anyone who doubts this can talk to my stupid bot
(http://rhizome.org/ada1852/).

The phenomenon we're discussing (the inability for most users to make
the leap to becoming power-searchers) is why I think that Derek R's
dismissal of Findability (meaning, his dismissal of the idea that an
information architect should pre-structure content for the end user) is
problematic. Empowering users with versatile tools is great, but there
are a lot of people who are simply unable to use that power.

An anecdote:
I have a friend who recently dissolved his file/folder structure in
Microsoft Outlook, preferring to store *all* emails in a single gigantic
folder. He says that the search and sorting tools in Outlook are quite
sufficient for him to find an email he is looking for, and that this
method is much easier than trying to remember if he put an email from
his sister about a cool web site in the "Cool Web Sites" folder or the
"Emails from my Sister" folder. For someone like him, a real power user,
using search is probably superior. But for someone who can't figure out
what search phrase to type in to produce the right results, the folder
system may be a better choice.

-Cf

[christopher eli fahey]
art: http://www.graphpaper.com
sci: http://www.askrom.com
biz: http://www.behaviordesign.com