As we could expect, Google engineer Amit Singhal didn’t explain all the factors that influence tweet rankings, but what he disclosed confirms some of the reverse engineering assumptions about the algorithm, that shares some elements with PageRank.

The first similarity is the equation followers = links, they both are recommendations in their own ecosystem. The more high quality links a page has, the better will be for its ranking. The more “reputable” followers a Twitter user has, the better his tweets will rank.

What makes a Twitter user reputable? The number of followers is one of the factors, but it can’t be the only one. How Twitter reputation is elaborated in the algorithm is still a secret and we can assume that other factors like the number of times a tweet is retweeted or favorited come into play.

Tweet uniqueness is another factor that influences ranking: Spammers and bots don’t take the time to customise retweets and the same exact message being repeated could be interpreted as a spam signal.

Hashtag abuse can also be treated as a spam signal, but considering their conversational value it would be a mistake to confuse topically relevant hashtags with spam.

Google engineers have worked on modeling not only the hashtagging behavior, but perhaps more interestingly the way to extract signal from the noise by looking for combinations of words that can create clusters with high topical relevance, like the words “Obama” and “Twitter” yesterday (US President Obama made his first tweet supporting the Red Cross).

Related posts:

1. Is Twitter Reputation the solution for Google R...

   By Daniel Peiser - 23 Dec 09 - SEO

2. Bing Wave 2: A Move to Social Search

   By Richard Bramer - 23 Oct 09 - Social Media

3. Twitter new advertising platform: Promoted Tweets

   By Daniel Peiser - 13 Apr 10 - Social Media