Ok, so in order to make this point I have to introduce you to someone.
World, meet my friend Hugh:
He’s sleeping. He’s also a great guy. And according to Klout, he’s an expert in vampires. How do I know? Well it popped up in my Facebook feed:
Now, I know Hugh real well, so I have a bit of a lead on you all. But I can bet that your once-over on Hugh suggests that, in fact, he is not an expert in Vampires. He doesn’t write Vampire fan fiction, he probably didn’t even see the Twilight movies (much less read the series), and he’s probably pretty ambivalent about the whole thing. Why is he an expert? Because he’s:
1. An active Twitter user
2. Tweeted about Vampires once or twice
That’s about all I can tell you, because Klout’s system is proprietary, and I can’t point to the exact line of code that decided he was an expert. All I can do is say that the special sauce deemed him to be one. And that’s the problem: in our collective effort to distill the utility of social media, we have adopted a reductionist view of the whole thing, where you’re defined by a single integer, your Klout score, or you’re deemed to be a “mover and shaker” because you show up on lots of user lists on Twitter. And then we’re left with situations where a single Tweet like this seems to make you an expert:
I got you this giant teddy bear wearing a vampire costume
— Hugh (@hughelton) February 15, 2012
Sure, you can make these things smart. Maybe there’s an algorithm that can do this reliably – maybe we COULD reduce everyone to a single set of digits, and determine who matters in a set of data and who doesn’t. Clearly, though, this isn’t it. We have a long way to go until we can reduce human social activities into a program. What we see right now is not an effort to understand these systems, its an effort to make money off of feel good pseudo-analysis that you can sell to some higher-up. And it doesn’t stop at Klout. Wait, hold on.
Ok, notification noted. Anyways. The point is, the people who run 140kit, Devin Gaffney and Ian Pearce come from an angle where we are humble about the business of understanding social media. We don’t know the answer to everything, and its very likely we can’t deliver a simple single metric. When someone says they can do it, a major indicator that they can’t is when they completely obfuscate the system that their numbers come out of.
Instead, we want to give you freedom to determine what matters. We don’t know the answer, but we do know how to build an engine that can help you figure out what matters to you. If you want to know the most influential accounts calculated by the raw number of times they appeared in a data set, we got you. If you want eigenvector centralities from the retweet network, we got you too. If you want to see what stated languages they speak in, great. If you don’t trust Twitter’s language settings (and you shouldn’t), we’ll give you an analytic that recodes the tweets to the best algorithm we can find. If you don’t see the analytical process you want, you can write one and submit it to us, because 100% of our system is open source (ok, we left out the database passwords, but thats it). If you don’t know how to write these things, tell us and we’ll write it if we think its important.
We don’t know the answers to your dataset, because no one knows the answers to your dataset. What we’re in the business of doing is giving you the tools, real, transparent, and unforgiving though they may be, to figure out what really matters, if it matters at all.