Posted by Sam Battin, Senior Search Strategist
How do search engines choose which tweets to show?
Last week, we talked about how some search engine users didn’t even know that Google had added a “Latest Updates” section to search results. This week, we’ll talk about some of the problems that search engines will need to solve before “real time” results can be made meaningful.
The interesting question is: Does incorporating Twitter into search results help search engine users?
If it does help, then that’s a win for search engines. If search engines can keep people on their site longer, then they can serve more ads and expose more folks to marketing, which helps pay the bills. (This may even become a part of Twitter’s business plan, to have search engines pay to receive Twitter data.)
Incorporating the Twitter data stream is quite difficult to do. Suppose you’re a search engine and you want to show Twitter posts in your result page. How would you do it? Would you show “Tweets” only from people who have 1,000 or more followers? I suppose you could, but having 1,000+ followers is not a guarantee that everything in the feed is going to be relevant to the search request. For example, Ashton Kutcher has 4.7 million followers; do you think his tweets are going to help you out?
Another way to filter out the spam from Twitter might be to only show results from people whose “Tweets” have been “Re-Tweeted,” that is, re-posted on different accounts. If something is re-tweeted, it’s a signal that the words are meaningful in some way to a broad variety of people, and potentially has more meaning to a natural search query.
Twitter’s “Suggested Users” feature, however, reduces the value of the “re-tweet” signal for natural search. The “Re-Tweeting” rate is tied to the number of followers of the original poster, or in other words, the more followers an individual has, the more likely something they write will be “re-tweeted.”
Twitter will suggest particular accounts to new members based on their popularity and other measurements. Being a “Suggested User” can greatly inflate the number of followers, and thereby influence the “re-tweeting” rate for their posts. For example, San Francisco mayor Gavin Newsom’s appeared on Twitter’s “Suggested User” list, and it’s likely that at least some of this man’s 1,370,000 followers got the idea from Twitter’s suggestion. If Mr. Newsom posts something about the iPad, should it show up in your search engine results? What if 500,000 of those users only log into Twitter once every six weeks?
Search engines will be working on this question for a while, no doubt. Twitter has somewhere between 6 and 20 million users, and search engines believe there may be something relevant in what they have to say. Perhaps so, but it’ll take some major computing power to consistently deliver important results on search engine pages.
Thanks to local Twitter celebrity and funnyman @joeschmitt for his insights on Twitter.