[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Voting on messages: a way to resolve the mailing list problems

On Sat, Dec 20, 2008 at 10:35:14AM +0000, Jurij Smakov wrote:
> Now, I know that for a bunch of geeks like us it is very tempting to 
> start discussing the technical details and how the scoring is  
> going to be implemented, and how the results are going to be used, 
> and so on. The way I would like to see this idea developing is that it 
> starts as an unofficial project, with very simple rules (like, "you 
> can vote once for each message ID"), which simply collects the data 
> and makes it publicly available in some way. Interested parties and 
> individuals can then use the data to provide their own metrics (and 
> try to convince others that their way of calculating the mailing list 
> "karma" is the right one). Eventually, we should be able to settle on 
> one authoritative way of calculating it, which can become "official", 
> and used to develop procedures for warning the offensive posters that 
> their behaviour is considered disruptive, for example.

To reiterate my point from IRC Friday, I don't think the described system
is at all useful *unless* we agree on a means by which strong community
disapproval of the poster's mails has consequences for that person's posting
privileges.  I think by this point, the people whose mails I disapprove of
know it ;), so what's the benefit to me in spending time ranking their mails
in a system that's advisory?

Furthermore, there are built-in rewards for people to continue posting,
despite knowing that some number of developers disapprove of their posts:

- it gives them an opportunity to try to persuade the "audience" of
- it increases the chances that the disapprovers will give up and go away,
  which can be seen as a "win" if the disapproval is mutual
- it gives them a chance to have the last word

Since we're discussing this in the first place, I think it's a given that we
think peer pressure is not adequately offsetting these rewards, so to
improve the quality of our list discussions, some other deterrent seems to
be required.

Finding a deterrent that's both effective and agreeable to the project is
the hard part, so I think that's where we need to focus our attention, not
on implementing the raw infrastructure that enables users to vote mails up
or down.

AJ's suggestion of a "cap and trade" system for mailing list posts is an
interesting one, but do we really have a clear idea what the monthly quotas
should be?  Further, if everyone has fixed quotas, that implies abusers /
malcontents get to send a fixed number of messages per month before they're
cut off; and that a minority of two is enough to keep a noxious thread
going.  Is there really a constant threshold that is going to be high enough
to let relevant discussions take place, while also being low enough to
dampen unhealthy discussions?

The alternative I still favor is to allow up/down votes on individual mails
to be fed into an overall "score" for the poster, taking into consideration
such factors as: the number of people disapproving of individual mails, the
breadth of disapproval of the poster's mails overall (i.e.: 10 "down" votes
from a single developer should weight less than 10 "down" votes from 10
different developers), the age of the messages (probably by clock time, with
exponential dropoff), and the amount of /approval/ of the messages in
question.  Then, below a certain threshold, we block mails from that poster.
Effectively this is a rate limit, since the score will change over time due
to aging of the data.  It does leave open the possibility of a crapflood
from one person posting a huge number of messages before anyone has a chance
to respond with scoring, but on the flipside, doing this gives us more
source data to use as a basis for banning them even longer since there are
more messages available to use in scoring.

If this seems too fiddly, I'm happy for us to at least give AJ's solution a
try first.

Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
Ubuntu Developer                                    http://www.debian.org/
slangasek@ubuntu.com                                     vorlon@debian.org

Reply to: