Are closed source bots prevalent? Isn't part of
the BRFA process
evaluation of the underlying code? Any admin bot should probably be
relatively slow, and make up for the slowness with long periods of
uptime. Some of the paranoia is a bit farfetched - it shouldn't be
incredibly difficult to get well designed bots that don't screw up,
and notice when they do. It might be exceptional among bots, but it
should still be possible. Bot RfA's have been doomed from the outset
recently, because most of the !voters don't have the technical skills
to evaluate whether or not its well designed (myself included).
On Jan 18, 2008 6:28 AM, Thomas Dalton <thomas.dalton(a)gmail.com> wrote:
On 18/01/2008, David Gerard
<dgerard(a)gmail.com> wrote:
On 18/01/2008, Tim Starling
<tstarling(a)wikimedia.org> wrote:
What's wrong with giving bots sysop access?
Are you worried they
might
rise up and overthrow the human sysops?
More or less. There's lots of paranoia on en:wp about admin bots going
batshit in sorcerer's apprentice mode. Though I don't think it's
warranted, as *anything* an admin can do is easily reversible except
history merges. (Making those *easily* reversible is one for the
wishlist.)
But that's not true when bots are involved. A human can only screw up
at roughly the same speed as another human can fix it, so it's not a
big deal, but a bot can screw up a million times in a few minutes -
that's not practically reversible without using another bot to undo it
all, which takes a lot of preparation (the bot needs to be written,
tested to make sure it's not going to screw things up even more, and
approved - that's likely to take a day or so at least).
Personally, I wouldn't object to open source admin bots ("With enough
eyes, all bugs are shallow." or whatever the quote it), but closed
source ones are too likely to go wrong and are thus too risky (the
chance of them going wrong is still quite small, but the potential
damage is enormous, so the risk is still high). Also, an open source
bot can probably be modified by any programmer to fix its own mistakes
quite easily, doing that with a closed source bot requires the author.
(So a closed source, supervised bot wouldn't be so bad, but I'd still
rather not have them.)
_______________________________________________
WikiEN-l mailing list
WikiEN-l(a)lists.wikimedia.org
To unsubscribe from this mailing list, visit:
http://lists.wikimedia.org/mailman/listinfo/wikien-l
_______________________________________________
WikiEN-l mailing list
WikiEN-l(a)lists.wikimedia.org
To unsubscribe from this mailing list, visit: