Cuil takes on Google - can the link madness be stopped?

So CNN tells us that Cuil does not rely upon link analysis to order its search results the way Google does. Unfortunately, at the time of this writing Cuil’s philosophy page does not appear to exist, so I cannot review what they themselves have revealed of their technological approach.

My first few queries did provide impressive results: nearly everything shown was relevant to the query and, unlike Google, Cuil listed very thorough, extensive content instead of popular and untrustworthy destinations like Wikipedia or YouTube videos. That’s not to say that Cuil pulled the best available content. Finding pages that deal extensively with a topic is not nearly as easy as knowing whether those pages are accurate and informative or just propaganda.

I don’t think Cuil has tried to get completely away from link analysis because, frankly, some of the images embedded in my search results were in no way associated with the content Cuil seemed to think they were. A search on my name, for example, included a page from my old press release archive with a picture of a guy I’d never seen before. Maybe his name was Michael Martinez. Another image in the results was also for someone other than me.

While I think the whole magazine-layout presentation is a good idea (it’s much nicer to look at than Google’s sterile SERPs page), embedding pictures of the wrong people with listing excerpts won’t provide a great user experience, in my opinion.

I was surprised to find that a relatively low content page from Xenite.Org ranked better than the SEO Theory blog for the “seo theory” search. This blog dominates on other search engines (as it should since this blog represents the presently largest repository of discussion on the topic). Flattered though I am to be so closely associated with the topic of SEO Theory, I think the blog needs to be found and ranked higher than a blurb page.

When I tried to stack the deck in Cuil’s favor by searching for “seo theory blog”, I found that there is apparently more to their algorithm than just title tags, URLs, inbound link anchor text, and on-page factors. Spammers will be pleased to note that, yes, Cuil’s first day index does include some RSS-feed sites.

I ran a query for a very popular actress about whom I have created some content. Cuil showed absolutely no results. Now, some actresses are better known than others, but this particular actress has a shared name space with another celebrity AND a building. Cuil should have found at least SOMETHING for that name, but apparently could not. Was the conflicted name space the issue?

I tried another name search, for an actor in one of this year’s most popular movies. He happens to share his name space with another high profile individual and several low profile artisans and business folks. Again, no results. There are at least two news makers in the shared name space, so I was very surprised to find no results.

They covered another of my bellwhether searches pretty well: Britney Spears. However, a search for “Thomas Jefferson” produced no results. When I searched for “George Washington” I got more information about GWU than about the former U.S. President. Well, you have to make some choices, I think.

Cuil does offer an “Explore By Category” section in its search results. This reminds me of the old Northern Light search categories (as well as Ask’s categorization), but when I clicked on “Washington Family” one notable member’s name was missing from the list (hint: He was a U.S. President). NOTE: The results sometimes change if you move the mouse cursor over another category and then return to the first category. I’m not sure if that will prove to be a strength or a weakness for Cuil.

I’m not sure Cuil is ready to be used for school research projects, but this is only the first day the service is actually live. It does seem like they’ll make an effort to promote something other than the cheap schlocky content that Google and Microsoft are satisfied with. If I never see another Wikipedia listing in search results again, that will be too soon.

I’d prefer my search engines to provide me with useful, relevant, and trustworthy results. In many topics that is almost virtually impossible — there’s just no way we can know who the real authorities are since any topic can have multiple communities of opinion and research. However, link analysis has brought us past the limitations of meta tags and Cuil now proposes taking us back to the advantages of emphasizing the actual content rather than what people say about it.

What is Cuil actually doing to determine relevance and value for the user? It does appear to me that they are giving serious weight to on-page content. That is a major plus and it gives them a HUGE advantage over Google and the other major search engines, but Cuil has to improve its actual coverage. I don’t care how many pages they claim to have indexed — they need to provide me with some content about Thomas Jefferson. I think more people have heard of HIM than have heard of ME.

Going forward, I hope we’ll see some interesting developments at Cuil. Gigablast was the last search engine to offer any real challenge to Google’s search technology but Gigablast was never able to put adequate resources to the task of building a major search engine. The search community grew tired of waiting for Gigablast to become an effective player in the market.

Cuil probably has a 1-2 year window in which to build some serious market share. Despite the nonsense you read about search market share in the media, Microsoft is actually GAINING on Google in terms of number of estimated monthly visitors. That is where Cuil needs to focus its efforts. If they fall into the “number of queries performed” trap they’ll lose the race before they get far down the road.

I am sure there are many reasons for why Google gets so many queries: ranking checks, poor quality search results, user boredom and curiosity, good depth of coverage for less optimized topics, and a sustained desire to get away from Wikipedia (I don’t know of anyone outside of Google and the Wikipedia who actually considers Wikipedia to be a reliable source of information, other than as a suggestion tool for what to find on other sites).