Using the internet is first and foremost a quantified experience. Every activity is assigned a number, or a metric; each action, private or public, is stored somewhere, somehow. So it’s surprising that calculating online influence is, at best, an underdeveloped, imprecise, and unreliable science. We’re submitting so much data to the world, but nobody can tell us what it amounts to.
Despite the efforts of companies like Klout and Twitalyzer, the industry that’s appeared around “influence” measurement best resembles the early days of search engine optimization. It’s full of tricks, games, and shady third parties trying to game the system to make a quick buck. Anyone with a few hours to spare can create a Twitter bot that not only appears human, but that, according to the best tools we have right now, is a more influential entity than actual people.
Earlier this summer, a study out of the Federal University of Ouro Preto in Brazil found just that. After closely observing the behavior of bot accounts, Fabricio Benevenuto, an assistant professor and co-author of the study, created a bot account disguised as a journalist for the Brazilian news channel Globo. After following the maximum allowable 2,000 accounts, the bot — @Scarina91 — was ordered to tweet out links (at random intervals, as to appear human) to major news stories off Globo’s site, as well as retweet posts from influential users with similar keywords. After 90 days, @Scarinia91 managed to pull in retweets and favorites from verified users, and secure a Klout score of 37.5 and an impressive Twitalyzer score of 86 (for reference, the study lists Barack Obama’s Twitalyzer score at 100).
“It really wasn’t difficult to create the bot account,” Benevenuto told BuzzFeed. “For a computer science student, this is really just one day’s work. If we can do this, really, anyone can do it, and others who spend more time will do it better.” He argues as well that it’s not just bots that are reason for concern. “Many real people are trying to manipulate numbers and appear influential as well, and there’s so much more that could be done to make influence metrics more reliable.”
Lithium Technologies chief scientist, Dr. Michael Wu, who has been trying to understand social influence for some time, agrees. “The vendors out there have no good measures to accurately determine true influence. What they’re finding is a kind of influence, but it’s often not the kind they want.”
In order for sites like Klout and Twitalyzer to succeed, Wu suggests they develop an adaptive algorithm that, rather than using set parameters, learns from human behavior and adapts as bots, fake followers, and unsavory “gurus” try to game the system. “Right now it seems that whoever tweets the most gets a higher score, but we see that doesn’t quite work, so you can have a human that goes through and puts less weight on things like volume, since everyone is tweeting a lot. Basically, mix the algorithms so whatever people do to game the system, you damp that metric down and weigh it less.”
But this kind of system is labor intensive and difficult to implement. “There are thousands of ways to cheat the system, and you’d have to track all those behaviors,” Wu said, noting that a system like Google’s adaptive search algorithms could bring credibility to influence measurement — essentially, a PageRank for your social presence. Just as Google’s changing algorithms helped to stamp out the rampant, shady search engine games of the mid-to-late 2000s, a social PageRank could bring an end to the rash of scammers, fake followers, and bot accounts.
But for companies like Klout, the window to become the trusted industry standard might be narrow. “I don’t think these companies have enough computing infrastructure in place,” Wu said. “But as these companies get bigger and the demand for these metrics grows, they’ll need to add it quickly. If they don’t, people will start to believe that influence measurement is meaningless and there will be less money coming their way.” According to CrunchBase, Klout has raised roughly $40 million dollars in venture capital money to date.
For now, the ability to accurately calculate and display influence seems to exist only in theory, though it’s easy to see how transformative a reliable metric could be. For personal accounts it could mean the kind of agency reserved in the real world for those with celebrity status. It could mean some free swag (Klout tries to mimic this by offering “perks” to users with higher scores) or greater attention from customer service (say good-bye to inattentive Time Warner employees and 45-minute hold times!). For companies, it’d be a chance to locate and vet hundreds, if not thousands of free spokespersons, each with his or her own audience.
Nowhere is our digital paper trail more useful and easy to gather than in the social realm. As ad dollars continue to flow into Twitter, Facebook, and other upcoming platforms, the desire to attach a definitive value to a person’s or brand’s influence is growing by the day.
By all accounts, the companies that calculate a user’s online influence should be among the most hyped and promising companies on the internet. Instead, sites like Klout and Twitalyzer are the frequent butt of jokes for their veiled, seemingly arbitrary calculations. It’s not hard to imagine these sites going the way of SEO, where once-promising companies are reduced to dilapidated, spammy enterprises, panhandling for clicks off flashing banner ads. They’re certainly facing a tougher fight. SEO was about quantifying and understanding the habits and desires of a complex machine. The influence measurers are trying to quantify, and sell, the habits and desires of the most complex machine of all.