Oxygenating the flame in threads was Re: [extropy-chat] a futurist prediction

[Eugen Leitl]

On Fri, Aug 26, 2005 at 01:52:46PM +0100, BillK wrote:
> On 8/26/05, Eugen Leitl wrote:
> > A way would be to patch Mailman to attach voting URLs at the bottom of
> > each message with
> > unique tracking ID for each user and message, and a table
> > to keep scores user vs. user, and a settable score threshold
> > (zero=no filtering) a poster needs to be over in the table
> > for messages to come through.
> > 
> 
> Over the years, I don't remember extropy-chat having a programmer
> available to start patching Mailman, and doing all the testing,
> debugging, conversions, etc. that would be required.

I wasn't describing a particular project related to ExI, just 
a basic functionality required to scratch the itch. 

We have at least two stellar system analysts, and many coders 
here, actually.
 
> But I don't think this would work anyway. The majority on any list are
> non-participants. They read (sometimes) the mail and delete it or file

It doesn't matter: the machinery described allows each user to rank
all messages according to her quality metric. It already works for 
a single participating user.

> it. That's all. They won't start scoring posts up or down. The scoring
> would only be done by the 20 or 30 keen members. And they would have

20 or 30 would be actually enough to start doing some clustering. 
With some added cogs and wheels, a person could subscribe to a particular
cluster's filtering metric.

> different opinions and would all have varying amounts of time
> available. So an assortment of posts would be scored by a small number
> of people of differing opinions. Might as well have a random number
> generator.

The metric isn't a scalar but a vector. A single scalar would
be a special case (a single cluster).
 
> If we have a few keen members, why not make them list administrators
> and use the existing Mailman procedures?

Because a commitee breaks the execution. You'd get Slashdot, where
idiots grade morons. 
 
> The present list moderation is done on a casual, after-the-event
> basis. When threads are over 50 messages long, smoke and flames are
> drifting over the battlefield and the field hospital is full of
> casualties, then the moderator will appear and tell everyone that they
> have been naughty boys and if they don't stop it, he will take their
> toys away.

This is a) too slow b) not fascist enough. 
 
> Fine, but all the violent tirades are already in the archives for
> googling journalists to find and quote back at us.

Tirades are useless. Perpetratos need to be warned offlist, and issued
a temporary suspension, or, in hopeless, chronic cases, a permanent ban.
 
> Mailman has rules for messages to be put in the 'held' queue. By email
> address and/or keyword filters. Moderators are notified and they can

Keyword filters are worse than useless.

> either reject or accept each message. If we have enough moderators and

New users on moderation by default is a good basic rule.

> the rules are loose enough that only an occasional message is held, it
> would not be too much of a burden on them. The rules can be gradually
> tightened up if too much rubbish still gets through to the list.
> 
> List moderation is not a new discovery. The best brains on the web

Complete moderation will kill a list quickly and surely, until
moderation worst case event is a few hours. Exceptions validate
the rule.

> have been talking about it and trying different options for many
> years. But ignoring the problem won't make it go away.

-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07100, 11.36820            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20050828/012221de/attachment.bin>