10.22.06
mini-review: The Perfect Search Engine Is Not Enough
This is a mini-review of The Perfect Search Engine Is Not Enough: A Study of Orienteering Behavior in Directed Search (2004).
The authors looked at how users looked for information in electronic documents (Web pages, email messages, or files in a file system) and found that people didn’t always use search (what they called “teleporting”), but frequently took a sequence of smaller steps to get to their goal. For example, a user went first to a math department’s home page, then clicked a few links to get to a specific faculty member’s home page.
They concluded that there are advantages to orienteering:
- keyword searches often fail
- orientation has a lower cognitive load (i.e. you don’t have to figure out what a good search query would be)
- orienteering allows people to use meta-information that keyword search does not (e.g. who a message is from)
- orienteering gave people more context (e.g. they knew they wanted the oldest file in a directory, but didn’t remember the name of the doc)
- orienteering helped people maintain a sense of location during the search (e.g. they could edit a URL directly to get to a slightly different Web page)
They suggest that search tools allow people to search on meta-data, learn which sites a particular user trusts, and steer search results to those pages, clustering information, and/or suggesting refinements.
—
I found the title slightly disingenuous: their users didn’t have the a perfect search engine, and it isn’t fair to extrapolate that the users would still orienteer if they had the perfect search engine.
An observation that I didn’t see them make explicitly is that orienteering can reduce the search space and thus increase precision. I do this frequently; for example, when searching for a file in the file system, I go a directory that I’m pretty sure is above the file I am looking for, then do a find:
find . -name emailCartoons
Instead of navigating to some subdirectory, I could do the find from / or from my home directory, but that would give me way more information than I need.
The authors also don’t say what operating system their test subjects used, nor which email client they used. Different email clients have radically different search capabilities! In particular, I remember that when I was researching my book, people told me that the Outlook search was so slow that it was too painful to use. Gmail, on the other hand, has search so fast that I almost never orienteer to find a message. While I have never tried it, I hear that Google Desktop is a whiz at teleporting.
I also have to think that better training on search engines would help people teleport more. For example, instead of typing the URL for the University of East-Central Illinois at Hoople’s math department, then clicking around to the faculty page to get to Sabrina Aguilar’s home page, someone could instead search Google for
aguilar site:math.ueci-h.edu
but a lot of people don’t know about the site: feature.
The Google Guide by my friend Nancy Blachman has good information on how to search better, including this advanced operators cheat sheet.
