On Jan 17, 2008 8:33 PM, Tim Starling <tstarling(a)wikimedia.org> wrote:
George Herbert wrote:
Take this up with the en.wp bot auth group.
There are bots with admin bit set, run by a few people, but it's a
small subset of the total (and, predictably, where some of the more
really heinous problems came from).
While every action is basically reversible, not all actions are
practically reversible. While it's unlikely that an admin bot would
be doing stuff which could accidentally lead to some of the nightmare
bot vandalism attack stuff I've gamed out playing red team, I prefer
caution, as do those running en.wp these days.
A properly written bot would be safer than a human. Automation is a good
way to avoid common mistakes. Like when a vandal moves a large article to
a suspicious title and replaces the contents with nonsense, then tags it
for speedy deletion. A human deletes the page and wonders why the servers
crash. A bot could check the page move log.
Yes, I know the human could check the page move log too. But computers can
process data much faster than humans can, and they don't get bored or
lazy. So they can run lots of complex checks every single time and never
risk an error due to a moment of inattention.
I am a large fan of system administrator automation tools... cfengine,
kickstart and jumpstart, (insert long list here). I write scripts and
tools at (insertdayjob).
In my observation, the WP interface and how bots are being coded to do
stuff lead to more errors than I like to see.
If the bots were as reliable as cfengine, or my disk management
scripts and patch audit scripts have been, then there would be no
problem.
--
-george william herbert
george.herbert(a)gmail.com