Tuesday 29 July 2008

Cuil - not going to cull Google!

For a change I thought I'd give my opinions on a new search engine that's being touted around.

Cuil is a new search engine that claims to have the biggest search index and give better results than Google owing to a methodology that looks at word context rather than page links.

There are already lots of comments on Cuil - for example, Webware's "New Search Engine Cuil takes aim at Google" or Karen Blakemen's "Cuil - not so cool"

I too played with Cuil - for around 5 minutes before I realised that this is very much a "what you see is what you get" effort - and I didn't see very much.

One of the first things I do when I use a search engine is change my preferences - to get 100 hits per page. I find a much more efficient way of looking through pages of results - and the time to look at 10 versus 100 on a single page isn't that much more. So I headed to Cuil's preferences page - and found that there was almost nothing to change. So you're stuck with a page of descriptions - and if they aren't right, you're forced to try the next page or a new search. Not clever! Then what about modifying my search - for specific types of content - title search, filetype search. Nada!

My top test keywords (generally "competitive intelligence" and various permutations of this) came up with the expected sites - but nothing new and not even all I'd expect - plus irritating logos attached to each entry that seemed to be stolen from images that seemed relevant.

My main complaint supports a comment on the Webware blog "Didn't we stop the pissing contest over number of pages searched about 10 years ago?". I concur totally. So what if Cuil has 120 billion pages. It's not size that counts - it's what you do with what you've got that counts. (I'm sure I've heard that somewhere before in a different context ;-) That's why Exalead is so useful - as it's so easy to customise, and refine searches. That's why Google is top-dog - as its interface is so simple and the results tend to be accurate. That's why Ask works - as it gives good results, with options to refine and it highlights news, images, encyclopaedia entries all together making search seem simple.

Finally their purported killer feature - relating search to the words on the page and their context. Isn't that similar or the same as the method Ask (or it's predecessor Teoma) uses, or have I missed something? (Or perhaps it only refers to the actual page rather than related pages which is what Ask does - if so, it's also 10 years out-of-date as just relating content to the actual page rather than linked pages was killed off by Google's linkage innovation).

So - not impressed. I still think that there's scope for a Google Killer out there, but Cuil ain't that Killer!