A few posts ago, I wrote Why I have doubt about @Klout. That post actually led to the problems mentioned being fixed by Klout although they never got back at me, not even when being politely asked to do so - which is their loss
Klout announced their Perks a few weeks ago, which basically means that they determine who is influential enough to be able to become eligible (!) for third party goodies and gifts. Or, in their words:
Every individual now has a network value based on their ability to leverage word of mouth and social media to influence their network and spread the word about a product or service.
Ambitious? I'd say so! Nonetheless, their service is still very, very shaky and they have lost my trust for sure. They are absolutely not up to the road lying ahead, simply because their statistics and calculations are erratic at best
I have collected a lot of screenshots, before you know it they'll fix their problems behind my back again without giving me credit for it, which is Lesson 1: Communicate openly and honestly
Note: I have not edited any of the screenshots. All I've done is cut out one part and saved that as JPEG, as to minimise its size (thus increasing the performance of this blog post page)
First, a few statistically impossible deviations. I really don't have to say much, the graphs speak for themselves. Influence is something you gradually build over a period of months or even years, and none of the parameters can significantly drop in a short period
Brian Solis seeing his network score drop by 50% within one week
Me seeing my Inbound-outbound message ratio increase from 0.5 to 7.5 within one week
Loic seeing his Klout score drop from 78 to 7 within one month
Lesson 2: if there are statistically impossible errors in your program, something is very, very wrong
This is all bad enough as it is. However, there are worse examples: conflicting scores in one and the same instance
Here's Tim Kastelle showing a score of 44
Here's Tim Kastelle showing a score of 54 (fifty-four)
Notice that the graphs are exactly the same (just like the score in the graph itself) but that both the general Klout score on the left, as well as the Klout description, are different. Scary? Heck yes!
Also notice that Tim's score wasn't updated in the meantime: this is just a screenshot from this week of his score from April 18th
Tim's not alone there, he's getting company from Mitch Lieberman
45 or 51, which is it? It can't be both!
Lesson 3: if your program behaves like an artifical lifeform and you're not a distinguished MIT or University scholar, your program code is probably just simply very messed up
Now, last but not least, a simple obvious error
2 List Inclusions? Meaning Dennis is on 2 lists? Last time I checked it was 342, not 2. This error scares me most, because it appears to be a valid score (not being 0 or -1 or anything else out of the ordinary), but isn't - by far
Lesson 4: If you're a crappy designer and / or programmer, there's always a last resort to get your program to function correctly: test, test, and test again
I suspect Klout is suffering from bad programming all over. Bad design, bad code, bad test, bad quality control. Collecting scores like these is a chain-of-events of course, where each step has to be taken very carefully, and succesfully completed, before continuing. In IT this is called a transaction, consisting of multiple checkpoints or savepoints. If any of those fails, the entire transaction should be considered to have failed
Update June 29th 10:43 CET: over the past days I've helped people explain what happened to their Klout score. There are two reasons IMO for the huge drop in score for nearly everyone: first, Klout 2.0 has had its impact on recalculations. Secondly, I suspect Klout is missing a good part of the history I suspect them to keep for individual users (or somehow magically Twitter is refusing them access to those Tweets). My advice to all: just keep recalculating your Klout score and you'll see it slowly creep up there again. It's worked for me, crawling back from 37 to 52 in four weeks time - and I haven't changed my Twitter ways reallyApart from that, there's always the good measure of sampling your own dogfood every now and then. And if all else fails, you can always take me seriously