[maemo-community] Reforming Karma
From: Simon Pickering S.G.Pickering at bath.ac.ukDate: Sat Oct 4 21:33:18 EEST 2008
- Previous message: Reforming Karma
- Next message: Why_the_closed_packages
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
>> For IRC it was mentioned that both the number of lines written by a >> user as well as the number of lines addressed to a user might be a >> useful metric to rate both a user's direct involvement (the number of >> lines they write and that are written to them shortly after they have >> spoken) and their overall contribution to the community (how much >> people talk about them, when they have not typed anything for a while, >> for instance). > > Again, I really can't see how this would increase the reliability. I mean we > maybe could find an algorithm that applies some heuristics but in the end > something like a log or root scale is much easier to implement and is still > evenhanded enough as (simple) heuristics are usually equally error prone. > Also an direct algorithm makes it more transparent and easier to understand. Yes, I agree, it would be far easier to go with the sqrt() suggestion, I was just throwing in some starting point ideas as to how to make the metric more meaningful. Unfortunately developing metrics (and their extraction) generally requires commitment and time and I'm not sure there's enough interest to go for anything other than a rough estimate in this case. OTOH, as the raw data is available online, some people might be interested in looking at it; I can't believe there isn't already research available on this sort of metric calculation, which could be used to provide pointers to some more complex methods. Cheers, Si
- Previous message: Reforming Karma
- Next message: Why_the_closed_packages
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]