Taking the Yahoo challenge

| | Comments (3) | TrackBacks (0)

I recently spent a month using Yahoo instead of Google as my default search engine. (Incidentally, the "Google's better, because the Yahoo home page is too busy" argument is bunk--Yahoo's search page at http://search.yahoo.com/ is every bit as clean and simple as Google's.)

I was surprised at how decent Yahoo's search was. Though Google's ranking still seemed to make more intuitive sense, Yahoo did a reasonably good job throughout. Unfortunately,I never felt like Yahoo's search was actually doing a noticeably better job than Google's. It was just a series of minor disappointments when I noticed results that Google could have done better with.

Yahoo's typo detection is very poor, compared to Google's:

  • When I searched for "eage book mod_perl" (with "eagle" mispelled "eage"), Yahoo didnt catch my typo. Google caught the error, and gave me exactly what I wanted--information on Practical mod_perl, a.k.a. the "Eagle" book.
  • When I searched for "ighthouseapp", Yahoo didn't figure out that I wanted to find out about the Lighthouse product, located at www.lighthouseapp.com. Google did.

Some of the results were just bizarre, but have since been fixed by a reindex:

  • I searched for "yahoo india news" and the fourth results down on the search results was a nonexistent page on Yahoo's own site, leading to a 404
  • For some reason, results from Target.com were coming up absurdly high for a wide variety of results. It looks like this has been addressed now.
I was also surprised at how tech-centric Google is, vs. Yahoo:

  • When I searched for "puppet" on Yahoo, the first page of results referred to, well, puppets. Over on Google, the first two hits were for the open source system administration software called Puppet. (Of course, in this case, I really was looking for the Puppet software, so the point goes to Google, but the bias is interesting to note.)

My biggest practical irritation with Yahoo Search wasn't with the web search itself, but the lack of an integrated blog search. I frequently jump to Google's blog search when I'm trying to find out what people are saying about something. Not having that be just a click away changed the way I interacted with the web, very much for the worse.

The Yahoo challenge was fun, but it actually made me appreciate Google even more. Next up, I'm looking forward to trying Bing for a month. When doing side-by-side tests, it seems to return significantly more relevant results than Yahoo Search (which is likely one good reason for the recent deal). I may end up back at Google, but I want to know that I'm using it for the right reasons, and not just laziness-induced lock-in.

0 TrackBacks

Listed below are links to blogs that reference this entry: Taking the Yahoo challenge.

TrackBack URL for this entry: http://www.chatterjee.net/mt/mt-tb.cgi/98


I've noticed Google's tech slant in the past as well, though I assumed it came from a different source. I use several of Google's services, and stay logged in to them on almost any browser I have open. Based on that, they should have been able to build an extensive profile on me by now. Any analysis of that data would quickly tell them two things:

1) I am a hopeless technology geek.
2) Almost 80% of the time the correct search result for me is one related to a technological term.

It's not hard to imagine that they might start skewing my search results towards tech-related articles, given that I have such a clear bias. However, this appears to not be the case.

Siezing upon your example of "puppet", I ran that query through Google first on my normal browser (Linux/Firefox/lots of google.com cookies), as well as on a squeaky clean IE 8 browser. I got the exact same page of results. Either Google's search engine just tends to have a slight tech bias, or it has frighteningly deep user identification capabilities.

I remember reading an article that said Google uses AI to take into account previous search terms you've entered when returning search results, so if that's the case, it might have (in this case, correctly) inferred that, since you'd done other searches on tech-related topics, you probably wanted to know about the sysadmin Puppet rather than the things that go over your hands :)

Referring to Michael's comment, I also remember reading that it looks at information other than browser ID (perhaps machine ID or that in combination with IP address?) to identify users. It would be interesting to try these same searches on different machines connected to different IP blocks.

I've tried running Google searches for "puppet" from different IP blocks, and got back the same results. That doesn't mean that they don't do any IP address-based targeting, but just that it appeared not to have been an issue in that case.

Leave a comment