[maemo-community] [Proposal] - Developers should get karma based on the relevance of their software

[From: Andrew Flegg]

On Tue, Dec 8, 2009 at 17:33, Valerio Valerio <vdv100 at gmail.com> wrote:
>>>
>>> There are 3 solutions with significant votes:
>>>
>>> Solution #1: Take downloads into account
>
> Using 100 downloads = 1 karma point.
> OMW has 5204 downloads so 52 points.
>
>>> Solution #2: Take valoration into account
>
> If [stars] => 3 then karma = [stars] * [votes]
> 3*52 = 156 karma points.
>
>>> Solution #6: Measure existing userbase via updates
>
> Using 100 downloads = 1 karma point.
> Assuming a user base around ~1000, in the best case 10 points.
>
> Any of these solutions seems better than the current one (max. karma of
> 42p), but #2 can be unfair for some apps that for some reason the users
> don't like, so the best option IMHO is a mixture of #2 and #6.

AIUI, the point of changing the metric is to award applications that
continue to be popular. Given the effort made on *application* karma,
shouldn't it be considered as a factor somehow?

The other thing to bear in mind is that any mechanism has to be
instantly computable from raw data due to the way the karma
calculation works (unless another table for historic data could be
used, presumably that exists for /application/ karma, unless it is
recalculated each time?)

How about:

   developer karma = average vote * k * application karma

where k = some constant (5-10?). The average vote will be between 0
and 5, and the application karma takes into account downloads, but
diminishes over time. This means that an application which has updates
which are downloaded will be awarded more than an application which
has updates which aren't downloaded.

Cheers,

Andrew

PS. Shouldn't maemo-developers be brought into the loop?
PPS. I've got a limited Internet connection, so apologies if this has
been raised at tmo already. But the discussion on application karma is
here, so the people who understand it most are also here :)

-- 
Andrew Flegg -- mailto:andrew at bleb.org  |  http://www.bleb.org/