Cuil Is Not So Cool
The new search engine, Cuil, from some ex-Google peeps, is horrid. I see a lot of bloggers fauning because, well, it’s the hip thing to do, Cuil’s new, and because Cuil is challenging The Establishment, i.e. Google and Microsoft. Well fine, praise their boldness and their politics, but don’t let that color your view of how the service actually works, because it works horribly.
First, the 2 and 3 column layouts not only look horrible in a respectably sized browser window, but it’s incredibly difficult to scan to find what you’re looking for, especially when page titles end up on multiple lines. The attempt to associate an image with each result is even worse. Sometimes it will pick a random icon or image from the actual site, and sometimes it will slap down a maybe relevant image from some other random site, giving the impression to the end user that the image is relevant to that search result, when it’s not. Example: my picture is associated with the ExpressionEngine wiki…wha?
Next we have a wrong image associated with the EllisLab blog, and an example of the often funky text handling of the page titles. The ampersands you see below are all in the markup.
Lastly, this stranger is yet again associated with the wrong site, this one. Odd since Cuil obviously figured out my association to ExpressionEngine, but it can’t make that association for my own site? Also notice the URL. For about a month when I was changing domains, that would go to an Apache generated 404. Now it is a permanent redirect to derekderekderek.com. I don’t know when the index was made, but either behavior is incorrect for a search engine, whether Cuil spidered my site when it was a 404 or a 301 code. Definitely not friendly for site owners, and will lead to loads of incorrect content in their index.
I didn’t have too look hard for these examples either; it’s par for the course. Do a few searches on content that you know will bring up sites that you are familiar with, and you’ll readily see the resource matching issues and the page title malformation. And sadly there’s no way to customize the output beyond their classified ad multi-column approach. That might be forgivable if they were liquid and filled your browser based on its width.
The goal of a search engine is to get me from the idea in my head to relevant content as quickly and accurately as possible. Cuil does not fit that bill in any size, shape, or form. At least they don’t have a big "Beta!" badge on the front page, though in this case, perhaps that would be entirely appropriate. I hope they get their act together on the presentation of material, because their goal I must admit, on paper, is noble. Relevance is supposed to be determined solely on content, and they do index more pages than any other engine. Of course, if they’re counting things like associating my content from derekderekderek.com with koruproductions.com as additional pages or sites despite the fact that that domain has always used proper HTTP status codes, their number is wildly inflated.