[Foundation-l] Use of moderation

[Tisza Gergő]

Erik Moeller <erik at ...> writes:
> What do you suggest? Are there models from other mailing list
> communities that we should experiment with to create a healthier, more
> productive discussion culture? What, based on your own experience of
> this list, would you like to see change?

I'll try to gather what I see as a problem, strictly from a technical point of
view (code being law and all that, I think it's still the easier side to attack
the problem from):

- the discussion space is divided by time, not by topic. What little topic-based
division exists (the subject line of the mails) depends on the ability of the
first poster of the thread to choose an informative title, and is hard to fix
afterwards. This, together with the lack of good search, means that there is no
easy way to see whether something was already discussed before (which makes
people reluctant to write about issues that they think might have been raised
before), and it is not easy to make use of insights gathered on this list,
making it a huge sink of time and effort.
- the moderation is not transparent: if someone claims being censured, there is
no way for most people to check whether he just tried to post complete bullshit,
or one of the moderators was indeed overzealous.
- the moderation is binary, and consequently too soft: there is no way to flag
messages as not containing any new information or insight, and this with the
habit of some of the regulars to get into frequent unproductive debates about
semantics and proper etiquette and such makes the signal to noise ratio low.
Also, there is no way to highlight posts, which would make sense in some cases;
e. g. in questions addressed to the foundation, authoritative answers by
board/staff members should stand out.
- topics cannot be raised on multiple lists without splitting the discussion;
there is also no easy way to move a discussion to another place.
- it is hard to include new people (who where not subscribed before) into a
discussion bacause the way replying works. (This is actually solved by gmane,
but it is not widely known, nor 100% reliable.)
- there is no way to see how many people are interested in a thread.
- there is no way to determine consensus (even approximately). With many people
not wanting to spam the list with mails saying only they agree or disagree, it
just devolves into the consensus of the most loud.
- it just doesn't scale well. Already everyone is complaining about the traffic,
and there are scarcely any issues discussed (compare with the number of
proposals on the strategy wiki).

I always found it strange that Wikimedia, being one of the greatest facilitators
of online collaboration, doesn't have its own cutting edge communication tools.
Not only do the mailing lists suck, wiki talk pages are just as bad. I think the
logical thing to do would be to take back most of the meta-project communication
to the wikis, eat our own dogfood, and develop a wiki-based communication system
that works (preferably in reverse order). LiquidThreads was developed for that
purpose, but it seems to have been largely discarded, with no significant
interest from the community, the foundation or the usability team - why?

I think the foundation should invest into reviewing state of the art tools for
large-scale constructive/informative discussion (slashdot, stackoverflow,
ideatorrent, uservoice come to mind) and adding whatever feature needed to
LiquidThreads to make it stick. I think opt-out moderation based on some sort of
collaborative scoring, some sort of voting or at least ranking method, and
thread summaries with a tag or category system are the norm nowadays, and of
course there would be need for a bidirectional email gateway.

That said, a few suggestions that do not require moving away from the current
system, are easy to implement, and might help the situation somewhat:
- set up a clone of foundation-l which is heavily moderated, and where all
letters which do not add new information or insight to the discussion are
discarded. People with a lot of time on their hands could still read the
unmoderated version on the original list.
- make better use of Nabble (or some opensource equivalent), which already
provides a forum interface to foundation-l with abilities to post from the web
interface, postmoderate and collaboratively rank posts and threads:
http://www.nabble.com/WikiMedia-Foundation-f14054.html
- make some of the private lists readable to everyone. If the only reason for
their existence is noise, it is enough to control write access strictly.
- set up a public waste bin where moderated mails can still be read (thus
avoiding the censorship debates) but do not pollute the discussion otherwise.