Hey Klout, One-dimensional Reputation Is Meaningless.

According to Klout,  “Your Klout Score measures your online influence on a scale of 1 to 100.” I just read a blog post (from June 2011) that carefully analyzes all the ways in which the Klout score is flawed, and therefore meaningless (hat tip to Peter Skomoroch). However, there is a more important reason than the flaws in the calculation: what does “online influence” even mean? There are countless topics and fields over which people can exert influence both online and off. I believe the main problem with Klout is the one-dimensional score. Like the excellent “Building Web Reputation Systems” book says, karma is user reputation within a context.

To put it in less abstract terms: some things are easy to measure with a single number, some are not. The faster car is the one with the highest speed. On the other hand, what’s the “usefulness score” of a car? You can rate a vehicle along a number of dimensions, but a single “vehicle score” would be a marketing gimmick because it lacks context.

Going back to online influence for people, it would be interesting to trying to measure it for a specific topic. However, I suspect that there won’t be much data for anybody but the most well known people within that topic. E.g. how influential is Robert Scoble in Distributed Systems? Is he more or less influential than Paul Graham for this topic? Obviously the answer changes if we’re talking about startup advice, although it’s still nuanced. Clearly Barack Obama is more influential in Politics than Brian Schweitzer (governor of Montana). But how does Lionel Messi’s political influence match up with Rafael Nadal’s?

The gist of a service like Klout is that it needs to be popular, and most people are not particularly influential at large. If only a few thousand people had Klout scores, and only for specific topics, then it would just be a  “Who’s Who in X” list. As it stands, it’s simply a game that measures how good you are at it. In other words, your Klout Score measures how good you are at getting a high Klout Score.

As usual, this is all unscientific speculation on my part. Follow me on Twitter for more of that, plus the occasional Foursquare check-in :)

Discuss on Hacker News if you please.

7 thoughts on “Hey Klout, One-dimensional Reputation Is Meaningless.

  1. Diego,You are dead on! Thanks so much for articulating so well something that’s bothered me for a long time! Great post, great insight and nicely said!- drew

  2. It’s interesting you based your opinion off a post from June of last year. From what I can tell, the Klout algorithm has changed considerably since then, making it much harder to game.

  3. @shawnccpr it doesn’t matter whether it’s harder to game, that’s not my point. The question is what’s it good for. Suppose the scores for Paul Graham and Ashton Kutcher (58 and 83 respectively) are perfect indicators of their “online influence.” Which one do you want as an investor? “Online influence” is so generic that a measurement of that variable without context doesn’t really help you. There are types of startups for which I wouldn’t want Ashton Kutcher to come close as an investor. On the other hand, if I were doing something related to the film industry then having the guy who wrote “Kill Hollywod” as an investor may not be the best idea :)

  4. The footer to this post reads a lot like an ironic attempt to obliquely increase your klout score. Not that I would fault you for that, but the interesting question is whether you would fault yourself for that or not.Of course, the far more likely scenario is that you simply want your content shared. After all, who doesn’t?

  5. @AABoyles I don’t care about the Klout score, but I do care about building an audience. Having generic “online influence” obviously means nothing to me. What I like is reaching others who may have interesting thoughts or counterpoints, as well as obtaining professional recognition among my peers. If Klout works as advertised, then an obvious side effect would be increasing that score. I care about it about the same as I do about Foursquare badges :)

  6. First @shawnccpr The Klout algorithm pertaining to subject area expertise remains ridiculously easy to game, and thus rather meaningless. I have a Twitter friend, he knows I am statistician looking for a job, I know he is a mathematician looking for a job. I noticed that he gave me 2 or 3 +K’s in Statistics, so I give him +K’s in Math. We have a mutual Twitter friend who has a job and is good at statistics AND math. I’ve noticed that we’ve both been giving him +K’s. He is now the 10th highest ranking in subject matter expertise in Statistics on Klout. (That is possibly a fair assessment, by the way, but not arrived at too methodically. And yes, I realize that there are other determinants beside being bequeathed +K’s by me and my Math friend!) I am 14th in Statistics. I don’t think this is indicative of my brilliance. Note that none of this is a result of deliberate collusion. I have never spoken to either of these two Twitter friends, about any topic, let alone Klout scores! If someone I follow, who follows me, happens to request +K’s, I don’t question. I gladly provide. Why? Because he (or she) is my friend, RT’s my little Twitter messages, makes finance or info sec jokes that I laugh about. How could I deny granting such a small thing as a few +K’s to someone who brightens my day with humor, warmth and validation?Diego, I checked that post you referenced, on Karma. I initially snorted in derision, then read further, and thought, “Hmmm, this is accurate, wow!” Here is the problem with karma and online reputation quantifiers:Unlike most content reputation, karma is implicit, opaque, and complex.Forgive me for leaving a message here, not the Ycombinator site as you requested. I was perma-banned on Hacker News some time ago, for offenses that were never disclosed. I do believe that you are correct in your conclusions regarding the relative meaninglessness of simple (which they are, of necessity) 1-D reputation metrics.

Leave a Reply

Your email address will not be published. Required fields are marked *