People powered search vs. Algorithm powered search

For quite some time I have been sanity checking some of my searches in delicious. Basically everytime I do a search in Google, I replicate it in Delicious.

I’ve been trying to see which of the two provide me with the most accurate search results, across a variety of search terms.

Google launching Search Wiki is a sign that they might, in the future, incorporate reoccurring trends from self-organised SERPs to influence their overall ranking algorithm.

There are quite a few self organised search engines on the web such as Wikia Search, and of course Delicious, which isn’t a search engine as such, but does provide folksonomy driven search results.

So here are some of my recent search queries (extracted from my Google Account):

“gap liberty shirt”
“fabric live 44”
“Chase and status”
“5D mkII equivalent”
“free enterprise wiki”
“volvo v70 diesel”
“SRAW”
“italian chef gennaro”

I queried all three of the terms above and compared results between Wikia Search, Google and Delicious. To keep things fair I only analysed the first 10 results I was delivered and I compared the positions at which I was delivered the most accurate information.

In my Daytum account I started to give a point to each service which gave me the best search results:

So as my little test shows, Google is still the search daddy and we’re still some way off people powered search catching up. Will it ever catch up? I’m not quite sure, what am I sure about is that when I have found information through Delicious it has been very accurate indeed.

What am I getting at?

The small sets of search results, which have been organised by groups of people deliver real value. If the information is valuable to me, it must be valuable to the all mighty search powers. I strongly believe that in the near future we will see algorithm based search engines use this information in some way to improve the information users get delivered.

Whilst results are relative according to what information an individual may or may not find useful, it would be good to see more extensive (and accurate) tests and results.