Theresa Knott wrote:

On Tue, 08 Mar 2005 13:55:04 +0000, David Gerard dgerard@gmail.com wrote:

...

...

Yeah. Possibly the isNewbie() function needs to be set to the last 2% or whatever, to minimise collateral inability to move pages.

The thing is that's not a definite measure, but advertising the metric on the move page would only encourage Willy on Wheels! to work around it with account creations and leaving them dormant until it's time for a move attack.

Although it's a prominent PITA, the page move vandalism is probably not presently bad enough to disable the last 10% of accounts from page moves as an emergency measure.

What if we lowered it to 1%. how many people would that affect and for how long roughly? ( I take it that we are still growing exponentially?)

Theresa

About a month per percent, given the current statistics.

I also think it's reasonable to work on the assumption that WoW is reading this list. We will still need to rely on soft security in the end. Currently, the technical advantage WoW has is the ability to user a tabbed browser to generate vandalism "bursts" that defeat the normal human processes of reverting and user blocking. (I've tried it, all apart from actually making the edits, and it's very quick to set up a batch of page moves, ready to commit by clicking one after another). Page-move vandals know what they are doing, and deliberately choose that form of vandalism for maximum annoyance: even when done manually, page moves take longer to fix and tidy up after than to commit, so the advantage is to the vandal.

Putting in a 2% block will hold off WoW and imitators for at least a month or two, which has got to be a good thing, and gives us time to set up better tools to await the return of the page move vandals. In the longer run, what we need are three things:

* one-click page-move reverting for admins * a page move log, working on the same principles as the deletion log * add page-move rate limiting for non-sysops, and you more or less have drawn the teeth of page-move vandals.

With these tools, page move vandalism need be no more annoying than any other trivial edit vandalism: simply block the user, call up their page-move history, and click revert as many times as needed.

And I agree: we should reserve the right to redefine the heuristics for is_newbie() without notice, to resist any attempts to "game" it. (I can think of several ways right now, but I see no reason to make vandals' lives any easier).

-- Neil