For some time, I’ve been wanting to blog about how the rise of Twitter could have implications for Search. So, when I saw a Tweet today from @socialmedia411 today on this very subject, this inspired me to write this post. The tweet actually referenced an article on Mashable http://mashable.com/2009/05/07/twitter-search-real/ and it touted Twitter’s aspirations to go beyond its current Search offering into the realms of “real” web search, competing with the established search hierarchy of Google, Yahoo, and Microsoft. Is this realistic for Twitter?

Twitter Search currently provides a very powerful way for users to find current topics and discussions happening on the web, but it currently doesn’t allow users to search the entire web - yet. Today’s article on Mashable speaks to the fact that Tweets contain links and so Twitter has a great plethora of link data to potentially crawl. Added to this, is the fact that Twitter could very readily rank the relevance of links, by the reputation of the Twitter user. Reputation can be a function of the number of followers a person has, the number of people they are following, in addition to the number of times a person’s tweet (containing the link) has been retweeted. I actually think, this is the key differentiator that Twitter has over any other search engine.

Traditional search engines make use of “Page Rank” to rank the relevance of web-pages for searches. Page Rank as devised by the Google founder, Larry Page, ranks a page based on the ratio of number of incoming links to outgoing links on a given web page (Very simplified). In a way, this ratio measures the popularity of a page. If your web-site gets a lot of other sites linking to it then the page’s Rank is weighted highly. This has proven to be a good measure of page relevance for web-searches to date.

However, Twitter’s search ranking system could bring an improvement to the traditional Page Rank algorithm. Twitter brings the notion of social or Reputation Rank. By Twitterers posting links in Twitter they are giving a vote to that web-page. And if this person has a high number of followers, then there is a reputation rank you can attach to the ranking of that link. Whereas a Page Rank calculation can be relatively static, a Reputation Rank could be a more real-time metric of how important or relevant a web page is. I say Page Rank can be relatively static, because there exists a lot of static content on the internet for which referring links do not change. Twitter on the other hand, catches all the new and great content, and so potentially has the edge on a main-stream search engine for informing on “what’s relevant now”

But can Twitter really match the scale of searching the entire web, like a standard search engine does?

Currently, Twitter has a snapshot of one segment of the web. Twitter is growing certainly, but it’s a fair assessment that even with all the tweets containing links to date, this would not get anywhere near to all the pages on the web, which is what a  mainstream search engine does. The links made available to Twitter for ranking, are really only a subset of the web. So, Twitter Search in its current form could not provide a reputation rank for all pages on the web. Granted those pages which are available for reputation indexing by Twitter are going to be a great set of relevant content to be consumed by search.

If Twitter is to offer a full scale web search then it would appear then that Reputation Rank should be used in tandem with Page Rank. That way, they could cover the entire web. If a page has a Twitter reputation rank, then this could increase its relevance over a Page Rank alone. Offering a full web search service then presents challenges of scale and other technical complexities, which in my view may defocus them from what Twitter has become.

I guess the real question is, does Twitter want to offer a full-web search, or do they want to offer users a more targeted search service that lights up the web’s most current and popular content. I suspect the latter is more achievable for Twitter on their own, but the former could readily be achieved by a partnership with an established search player. Time will tell.

What do you think?