Om Malik has posted a pre-review of Wink, a new search service that ranks web sites based on how often users bookmark web pages using their Wink accounts. Using bookmarks as a measure of a page’s importance is an interesting metric. It seems reasonable to assume that good content gets bookmarked more often than bad, so good content should float to the top of results over time.

A big challenge for this search metric is the lack of participation in bookmarking networks. Without a LOT of votes, the quality of the rankings will likely be inconsistent at best.

Malik shapes Wink’s model and similar sites as being part of a “people vs. Google” movement, where a networks of people attempt to out-do Google Search by using votes, bookmarks, or other human behavior rather than a faceless algorithm. Malik expresses his pessimism about the potential for this type of ranking system based on the lack of mainstream interest in social bookmarking and tagging. However, this really isn’t - or shouldn’t be - a mutually exclusive strategy for ranking content. For example, a site like Wink would be worthless if it ONLY displayed content that has been bookmarked. People would reach dead ends or marginal results most of the time if Wink didn’t supplement their search results with results from . . . Google.

The “people vs Google” argument concept bugs me because it ignores the main factor that makes Google Google: link popularity generated by people. As John Battelle explained in his best selling book on Google, The Search, Google’s PageRank system is designed to measure web page citations. The more times a web page is cited on other web pages, the more important that web page becomes. Additionally, citations from important pages are more valuable than citations from less important web pages.

Can a relatively small network of web users tagging and voting on web pages create more relevant search results than Google’s calculation of the citations on billions of web pages? If the the goal is to provide the most relevant results on anything someone could type into a search engine, the answer is clearly no. However, there may be some interesting exceptions for certain types of searches. For example, Digg.com’s voting system does seem to make it easier to find news that’s worth reading within the topics they’ve tackled so far, and it seems like it will effectively scale to other topics. Del.icio.us does a relatively good job identifying must-read content in specific subjects using shared bookmarks.

Will Wink.com’s combination of search, tagging, and voting take search to the next level? If bookmarking and tagging increases the relevancy of search results, will we see Yahoo using metrics from their newly acquired del.icio.us to rank results? Will Google use aggregated Bookmarks data they gather in user’s Personalized Search History as a metric in their algorithm? Are they already?

UPDATE: Wink’s founder and CEO, Michael Tanne has responded to Malik’s original post. Definitely worth a read.

Share This