[WikiEN-l] JoanB is DW

Wed Apr 16 20:36:00 UTC 2003

> Do people still believe that anonymous contribution is critical
> to success, or is it just an attitude left over from the startup
> days?

This has been discussed to death, here's my opinion on what we need to get  
done in the next months:

1) Allow sysops to check the IP of new* contributors
2) Allow sysops to block new* contributors by IP/name
3) Have IP blocks expire after a definable amount of time
4) Use an additional cookie-based "soft blocking" mechanism that should  
fool the technically challenged
5) In order to increase sysop accountability, log all sysop actions in the  
recent_changes (e.g. user:foo banned, page xyz protected)

*new meaning that the user has not made any edits older than n days

This allows us to deal with almost all kinds of pests fairly quickly:

1) Standard vandals get a 2 hour IP block, which hopefully will teach them  
to behave. If they don't, whack them with another one. This is quick and  
easy with little risk to other users of the same ISP.

2) Obnoxious critters like Black Widow can be re-blocked as they re-appear  
under different nicknames and hopefully identified by their behavior and  
their IP address.

3) The cookie-based blocking could work across usernames/IPs if the user  
doesn't figure out how to disable it (cleaning cookies is easy, knowing  
that you can't edit *because* of the cookie is harder to figure out).

These improvements would help us to avoid the ever looming threat of an  
"email signup with confirmation URL" process that would significantly slow  
down the flow of new contributors. Having Wikipedia wide open has always  
been part of its attraction, allowing people coming from search engines to  
quickly fix stuff and maybe get "hooked" by the concept. Erecting barriers  
typically scares people away -- how many passwords do you have to deal  
with already? Even Wikipedia's signup process is easy as pie, and should  
stay that way.

But our current mechanisms are insufficient for dealing with persistent  
problem cases, so we need to improve them. Anyone up to providing code  
should join the wikitech-l mailing list and get going.

Regards,

Erik