Sprint task: Refine the karma system (community input requested)
 

A message was recently sent out on the Community Mailing list. This is the type of topic that can be highly volatile if things get decided in a bubble. Please read through the proposal, view my comments and follow the links to view what others are saying. I would like to see some talk folks get their opinions heard also.


This was the message I sent to the Community Council mailing, it did not go out to the community mailing list.


This prompted VDVsx to include the following addition to the proposal when sent out on the mailing list.

*TMO 'gold posts'(new, need to be implemented): Posts with +40/+50 thanks receive additional karma (10 ? 20 ?)




While I am not sure about any of these numbers in any part of this proposal, this gives a good starting point for discussions. Well community what do YOU think?



Link to the posts on the mailing list

Maemo Mailing list Archives
http://lists.maemo.org/pipermail/mae...read.html#3705

This is a easier way to read them in my opinion...
http://www.gossamer-threads.com/list..._view_threaded


Another interesting thread you may want to check out....
http://www.gossamer-threads.com/list...ommunity/57207

Re: Sprint task: Refine the karma system (community input requested)
 

May I add the following:

As far as I remember, Karma was introduced to measure how deeply somebody is involved. Writing 100 posts a month sure is involvement.

The reason why one can gather Karma for pretty much everything that can be counted here is that Karma doesn't judge. It doesn't (in theory) know about "useful" and "useless" activity. It only knows "activity". Because "activity" is what it should measure. It's a very generic thing. It doesn't measure somebody's skills as a developer. It doesn't tell anything about the quality of your contributions. High Karma values only say: This person comes to sites that somehow belong to the maemo.org empire and do something there. Whatever that may be.

The reason why this isn't really problematic is that Karma only counts when we try to draw a line at elections: We want people to vote who probably know something about this community and invest some of their time. Karma is a good way to tell just that. For this purpose, again, it doesn't matter at all what these people do at maemo.org. They are here, they maybe vote for applications or participate in mailing lists, whatever... No reason to know if their votes are reasonable or their posts on the mailing lists are useful.

This is what Karma is.

Trying to introduce an iron curtain between the "good doers" and the "bad talkers" doesn't do any good. First time I read about this doers/talkers thing I felt personally insulted. (Actually, I still am.) I mean, come on, the great doers write lousy apps that drain people's batteries and us dirty talkers then need to support noobs at t.m.o. who fell into that trap... No, I don't really mean what I wrote in bold before, it's just a provocation... and a vision of what could happen. Do we want to start this kind of "us vs. them" discussion in our community? Really? Doers vs. talkers? Is that what we want? For a number on your maemo.org profile, FCS?!?!

Let's be honest, the thing that started the Karma discussion was the community device program for the N900. Developers thought they should have received a device and couldn't understand why bloggers got one. Got "theirs".

If this is the reason why we're discussing this, stop discussing Karma and start discussing criteria for the next device program. You can count anything. You can count the number of apps in Extras somebody uploaded if you want to cancel the community device program and make it a developer only program. Go ahead, that's a reasonable thing to discuss. (I would advise against it, but at least I see a point here and can understand why this is an issue.)

Tuning Karma as such: maybe, if there's obvious problems. (Such as: How does Karma collected 3 years ago reflect current community involvement?)

Tuning Karma to make it prefer "good" activities over "bad" activities: no. Especially when the good/bad line runs between the doers and the talkers. We found out during the device program discussion that there actually are developers who have little karma because they just don't come here. They develop and release, but aren't, you know, "part of the community". I wouldn't want to change Karma calculations so that people can get high Karma without really being here.

Re: Sprint task: Refine the karma system (community input requested)
 

He said talkers and *doers*, not *coders*. Filing a bug, editing a wiki page, writing a Talk post getting plenty of thanks, a Brainstorm proposal geting plenty of votes, a blog post getting plenty of thumbs up... fall clearly in the category of *doing* even if you do it with words and not code. and this is why the karma system tries to praise that.

Actually I think it was me who started confronting the talkers/doers positions (that in reality are intermixed, of course) putting myself as an example: it could not be possible that I was heading the karma list when in the meantima John Costigan (at the time developer of the most popular Maemo app) was not to be seen in the charts).

I also believe the karma thing goes far beyond device programs. But the discussion here is not wether this system is appropriate or not for the Maemo community but whether it counts karma in the way the community wants.

Re: Sprint task: Refine the karma system (community input requested)
 

"good" and "bad" are so judgmental. Old moral scenarios that wrestle with things like "when was it appropriate to kill Adolph Hitler" are a good (heh) example (although extreme).

I am for a karma system that is "good enough". It doesn't need to be perfect; it just needs to quantify some form of contribution. And once the metrics are gathered, they create a smaller set of the population that is then easier to judge if need be.

So I'm once again with benny1967 on the subject. Let's hash out a system that is good enough, expand/refine it as need be, and exercise judgment against the metrics. I think it's safe to say there will always be context that lets us immediately say "oh yes, Member A with her 1000 posts did indeed contribute positively, while Member A with his 500 posts was an obvious troll."

Some worry about a spammed karma system, but spammers and gamers become self-obvious and easy to weed. ;)

Some in the discussion keep trying to make it about that very thing.

Re: Sprint task: Refine the karma system (community input requested)
 

I'm at all with Benny. Nokia plans device programs the way they need and like, according to the targets they intend to reach (more developers coding programs, more buzz in blogs and social networks, maybe testimonials, who knows?). Karma has been the chosen independent variable this time, but wasn't so for the previous device and things went quite well nonetheless: there has been plenty of good work from people blessed by a Nokia subsidized N8x0, which made up a large base of knowledge, expertise and code for the N900.

If at all differentiating between coders (I don't appreciate doers either, like people not coding in C were unuseful and lazy) and users (why talkers?) is absolutely necessary, I propose two different karma variables, one earned for releasing software and getting plauds for it, and the other for community involvement. At least, if Nokia decides to use again karma points for device programs, they can include a mix of top coders and top users, at the ratio they decide to choose.

Also the elegibility to the council or other community positions could be bases on a function of these two karmas, possibly with a greater weight for involvement.

Re: Sprint task: Refine the karma system (community input requested)
 

I don't think differentiating should be the goal. Rather, to abstract different types of contributions onto a level playing field where commonality can be found. It can't be made perfect, but I think it can be made practical for the uses discussed.

Re: Sprint task: Refine the karma system (community input requested)
 

I just checked my karma: Holy Sh*t! How did I get that much? the system is distorted, IMHO, if someone like me can get that sort of score.

Re: Sprint task: Refine the karma system (community input requested)
 

Wrong! If anything it validates the system and process. You=helpful. ;)

Re: Sprint task: Refine the karma system (community input requested)
 

The problem introduced by VDVsx is that commenters on maemo.org are getting ridiculously high karma for posting a handful of comments, while app devs are getting next to nothing for popular apps.

I also think the mailing lists are out of line with the forums. A post on the mailing lists shouldn't be worth so much compared to a post on the forums.

Look at my maemo.org profile for example.

Comments: 70 karma
(I've made a few comments here and there. Not very many)
Products: 35 karma
(That's for Easy Debian, a 3.5+ star* app with more than 9500 downloads)
Discussion: 92 karma
(Again, I've made a handful of posts to the mailing lists. Not very many)
Forum posts: 75 karma
(That's for 5660 posts! :eek:)
Forum thanks: 550
(That seems OK to me, but I have a lot of thanks)

You see how disproportionately high "comments" and "discussion" are rated? And how low "products" and "Forum posts" are rated?

* As for the rating of apps, I don't like that any random person who has installation difficulties can rate my app with one star and hurt my app's rating... but that's another thread...

Re: Sprint task: Refine the karma system (community input requested)
 

So you received 35 karma for 1 app, and 75 karma for 5600 posts. And you think thats the app didnt receive enough karma compared to the posts? I guarantee people would agree your 5600 comments are worth far more than easy debian. Well at least half of the 5600 anyway :D:D:D

Then comes the thanks, you have 550 karma for thanks which you well earned in this forum.

I personally don't see a huge issue, but maybe its just me.

Activity
Favourites 4.25
Blogs 64
Groups 15
Bugzilla reported 8
Mediawiki edits 3
Discussion 5
Itt thanks 282
Itt posts 54
Bugzilla comments 2

I have 2965 posts, which gave me 54 karma, and 1251 thanks for a total 282 karma. This is based on 2.04 posts per day average since January 2006.

The question becomes how many karma should one app be able to generate, and are all apps equal? Should one app be able to give more karma than 2900 TMO posts? I would say no way. You got 35 for the easy debian, how many do you think you should have received?

Perhaps a way to earn additional karma when updates are published for a app, or additional for porting to new maemo version.