Monday, July 28, 2008

Call of Cuilthulhu

I'm going to add to the dog pile already on top of

"cool" dot com wants to be the next Google, but it's not even the next Mahalo. Putting aside the performance problems and outages, the search results are mostly useless. My experience, and those that I have read, indicate that Cuil's "120 billion" (insert Dr. Evil pinky here) index results are a gross over-evaluation.

A typical search result for a person will include:
  1. LinkedIn page, just one, and there will be a picture, probably of someone else.
  2. Blog postings, comments, and other junk pulled from RSS feeds.
That's it.

You won't find any "top level" results like, oh, the person's web site. If you seach for "John Smith" you will not get, even if there is one and it's a very popular site.

The same goes for organizations. I searched for my college marching band: The California Aggie Marching Band-uh from UC Davis in California. There is only one and that's a pretty unique search string. Did it find Nope. It found some of the sub pages, along with a random picture that has nothing to do with the band. Plus it found some other crap that is very tangental to the band.

The image thumbnails that cuil inserts into the search results are all hosted by cuil (that's polite) but there is no way to tell where they came from, and that's bad. If someone does a cuil search on you, and your LinkedIn page comes up with someone else's picture next to it, that's bad. But worse is if YOUR picture comes up next to some one else, say a criminal, whose police blotter is the first search result. I can see the lawsuits now!

Try cuil yourself (if it's up) on something close to you and see how it works for you. I'm sure it works for some things, but how can we have any confidence in this service if the results are so suspect?

No comments: