From the monthly archives: November 2011

Klout’s recent rejigging of their scoring system caused a lot of kvetching by people that care about it. (Some going so far as to saying that they have “pulled a NetFlicks”. Ha!)

Then more fuel was added to the fire this week, regarding their privacy policy. (Someone should name a law that says: “Every social media company will inevitably have a privacy scandal.”)

Although I’m sure Klout wished that the response was less negative, they have to be pleased with the amount of attention they’re getting. A lot of people are hearing about Klout for the first time, and asking “What is this and why does anyone care?”

TechCrunch stepped in with a helpful post titled Nobody Gives a Damn about your Klout score.  It’s worth reading.

 

What is it?

Klout is trying to rank the “influence” of each Twitter user. (They take other social networks into account as well but, predominantly, this is a Twitter tool.) It looks beyond just how many followers you have and considers whether your messages are responded to or shared. (The exact formula is secret).

The approach is flawed because there are too many different factors to consider and, more importantly, too many different ways to use Twitter. The score is often absurdly meaningless, as pointed out in the above article and as various experiments have shown. So, at the moment, I recommend looking at your Klout score for entertainment purposes only.

 

A system like Klout is inevitable

Despite the current flaws, I think Klout – or a system like it – is inevitable for platforms like Twitter. But not for the reasons most people think. I think its main value will be to weed out spam accounts and bots which are going to become a bigger problem as Twitter spreads.

Unlike email, where there is no central authority, all tweets go through Twitter’s servers. Twitter is the judge and jury in deciding which accounts are being too spammy (or otherwise uncool).

Up till now, they seem to be doing an OK job of that. But as the value of twitter increases, the cleverness of bots is going to get better. And as the overall volume grows, will Twitter be able to keep up?

 

Is there a human there?

Furthermore, Twitter’s judgment is binary: an account is either banned or not. We need something that allows for some grey areas. It’s fine to have an account that just automatically tweets based on some algorithm, but it would be good to be able to know when that’s the case. Is there a human behind the wheel or not?

 

Compare with email

Can we just import our decades-worth of experience in dealing with email spam? In that battle, we have sophisticated automated filters and, to supplement them, filters in our own brain for assessing an email’s spaminess. Both the machine- and human-based filters work because email gives us many clues to help with that assessment – the sender, the words, the formatting.

But Twitter presents a different challenge: We have very little content in an individual tweet on which make a judgment, but instead we have persistence of identity. That is, we know with certainty which account a message came from and we can easily see the account’s message history. (Something email can’t do.)

 

The medium dictates the scoring system

When a messaging system gives you minimal per-message content but full message history, effective quality scoring must be based on the latter.

That is essentially what Klout (and others like it) are doing.