Re:Fake Friends with Real Benefits
by marmarlade on Monday June 09, 2014 @08:36AM
(#7868113)
>> It's pretty damning these fake followers were able to game Klout's algorithms, since that's what Klout was trying to solve.
Agreed. They must have improved their algorithm since though, surely? The algorithm at PeerIndex (European Klout competitor) was good at ignoring fake followers IIRC. See https://web.archive.org/web/20130426180937/http://www.kernel... & http://blog.peerindex.com/pornfluence-and-the-world-of-buyin... (quite a similar experiment that Gilad conducted)
Yeah, a follower count is a bad metric because it completely ignores the quality of a follower. Compare a high profile, well known, respected and talented person who has contributed a lot as a follower, as well as a bot who signed up a week ago and has no meaningful contributions.
Each follower counts for 1 in the total follower count.
It's pretty damning these fake followers were able to game Klout's algorithms, since that's what Klout was trying to solve.
But "number of followers" is easy for people to understand so people will continue to consider it important.