In my West-coast hidey hole, I missed all the hullabaloo about the new search engine Cuil’s official launch this week. Everyone was talking, but no one was particularly impressed. The site was down. People didn’t find what they expected. Pictures of the wrong things showed up in the right places. No one can expect to compete with Google. Blah, blah.

The first thing I noticed about Cuil was that it did not conform to the list of things model for displaying information that libraries and famously successful search engines use. Sure, sure, the results screen is ordered somehow, but the searcher can browse the group of things in more than one dimension. Sometimes, the results include the groups to which your search might belong. That’s one or two points for the good guys at least.

Ordering search results in a list not only implies hierarchy and value, it implies a direction that the information-seeker should take. When we search on Google, we click on the first hit or scan the first few results before making a decision. When what we want is not there, we’re often at a loss. After the first few hits, the order of search results might as well be arbitrary because there is no clear direction to follow. The 34th hit is often as good as the 19th.

Google is good for looking things up, then, and I gather that Cuil is trying to fill the niche of what Google is not good at: serendipity and discovery.

Siva Vaidhyanathan puts it a little differently.

One of my biggest problems with Google Web Search is that it presents the illusion of quality via rankings. That linearity undermines the important yet frustrating aspects of research: you don’t always know what you are looking for or know what is best for you.

Siva gets this and so many other things so right that I was considering just writing a love letter. What compels me to write about Cuil instead has been librarians’ reaction to it.

On discussion lists like Web4Lib and Publib, as well as on library blogs, three themes have really emerged.

One theme is, “if you think Cuil is neat, you should try SearchMe”.

SearchMe may fill some of the same discovery and serendipity niche that Cuil is vying for, but its immediate appeal is that it looks a lot like Apples’ “Cover Flow” browse interface for the iPhone/iTouch (and something similar in OS X 10.5). There is an obvious ranked order, which infodoodads Jane points out is helpful, but still means that browsing is an essentially unidirectional journey.

Somewhat dishearteningly, most of the posts I’ve seen about SearchMe get no reaction. No one seems to be interested in analyzing what search engines are good for, only whether or not the interface is nice and if the results are good. The second theme in librarians’ reactions to Cuil is that the main way that we’ve gone about evaluating Cuil is by searching for our own names.

On Web4Lib, Dan Lester wrote (and read the whole thread if you want),

Doing a little vanity searching, I searched

dan lester boise

and found lots of stuff….well, some….almost all of it from old
newsgroup postings from 93 and 94. Well, I guess that’s a rich
source of fodder to build stuff up, but hardly exciting or
convincing.

Connie Jo Ozinga did a similar search and discussed it on the Publib discussion list (again, check out the whole thread if you want), concluding that the search results did not reflect her current position: “…if you search for me on Cuil, don’t think that is me you are seeing.”

Here is a kind of irony; Cuil is claiming the moral high ground by treating user data as completely private, and yet librarians complain that they are too anonymous in Cuil’s results.

I admit it, I searched myself on Cuil also. The initial results tell you a lot more about where and who I’ve been than where and who I am now, and that’s just the opposite of what Google values. Google’s paradigm props up the newest and the most popular web pages and clearly defines a hierarchy of value.

Cuil doesn’t seem to care as much. Librarians should love this.

We are always complaining that people value online information over print, digital over analog. We worry that too many people “satisfice” their information needs with the first few hits from Google. Some of us even worry that ordering results any way but alphabetically implies too much value to the resources at the top of the list and limits patrons’ freedom of inquiry by devaluing everything else.

And when it comes to search engines - at least for those of us blogging and sending e-mail to discussion lists - we won’t have it any other way. The third theme is a defense of Google as a gold standard for search engines.

OhioLINK’s Thomas Dowling wrote,

I must have missed the welling dissatisfaction with Google’s search results that would make anyone think a new search engine is their path to fame and fortune.

I could probably discuss a fourth theme, the comments on downtime, mismatched photos, dead links, and missing pages, but I think this points only to a static expectation of what a search engine should do and how it should perform. If it’s not Google, it’s not Google-enough.

Where the talkative librarians dismiss Cuil on these grounds, Siva Vaidhyanathan raises the question of what search engine results should be. After comparing the prominence of his own self in Google results for “Siva” with the mild lack thereof in Cuil, Vaidhyanathan comments,

There are almost a billion Hindus in the world. Don’t you think that a search for the name of a Hindu god should be dominated by references to that God? Sure, I am one important Hindu! Don’t get me wrong. But I am not as important as any god, Hindu or not.

A few messages down in the Web4Lib archive, Dowling argues that Google does it this way because it’s what people want. True, but maybe not true enough.

We’re ignoring that people want it that way because Google does it that way, and we’re ignoring the fact that other ways to do it produce different results and are good for different purposes.

It’s time we started treating search engines like reference sources: pick them up, feel the weight, read the introduction, look for specific sample entries, even read random ones, gauge the editorial position, check out the index and other appendixes, and finally think about what kinds of things we would use it for.