Twitter is bad at measuring online influence, and that’s good
Twitter cofounder and board member Ev Williams suggested this week that the microblogging site de-emphasize users’ follower counts in favor of some better algorithm for measuring their influence, according to Buzzfeed. “The thing I think would be more interesting than followers is ... retweets,” Williams said, adding that a simple follower count “doesn’t capture your distribution.” He went on: “The dream metric is how many people saw your tweet.”
But why stop there? After all, people might see your tweet and ignore it. Or they might see one tweet but not the next, because they don’t follow you. Or they might be influenced when you tweet about one thing but ignore you when you tweet about something else. What Twitter is really after, as it tries to woo advertisers, is a way to objectively measure users’ true influence on the site. Unfortunately, that leads to a game of whack-a-mole. Every time you try to measure one thing, something else pops up that you’re not adequately measuring.
Take the example of Klout, a startup that was already trying to do what Twitter might try to do now. After working for more than two years to try to get a bead on people’s online influence via Twitter, then Twitter and Facebook, then Twitter and Facebook and Google+, it remained dissatisfied that Justin Bieber outranked President Obama. So it began trying to incorporate offline influence into its ratings as well, drawing on sites like Wikipedia and LinkedIn as proxies for real-world stature.
Another site called Kred has moved beyond numerical ratings to offer “a visual history of your Social Media Influence,” allowing you to “see your full influence story and zoom in on meaningful moments.” (I’m picturing a washed-up former social-media ace sitting in his living room 20 years from now, surrounded by old Foursquare badges, swilling a beer and reliving his glory days. “Guys, remember that time I wrote that tweet that engaged Pete Cashmore, Paulo Coelho and 374 others?”)
The effect of all of these attempts at objective quantification is to turn social media from a conversation into a competition. Williams observes that Twitter’s existing metric, the follower count, is too easy to game. It has spawned a cottage industry of “social media experts,” many of whom cheat the system by harnessing fake followers for their clients. Williams’ reasoning seems to be that more sophisticated metrics will be harder to cheat, and thus more accurate. Another possibility is that it will simply lead to more sophisticated cheating. But even if it works, it’s hard to see the value here for the average user.
Some bloggers have complained that Twitter’s analytics today are available only to advertisers and not the general public. But maybe we should count this as a blessing. At its best, Twitter is a tool for connecting people around the world in a many-faceted real-time conversation. Attaching a rating to each participant only reinforces the idea that it’s a contest rather than a discussion — and there are already too many people on the site who treat it that way.