Hello, all: I've written up [[Wikipedia:Quasi-protection policy]] [1], a proposal similar to semi-protection that would effectively limit sleeper accounts used to vandalize articles linked from the Main Page. I know that I've written a lot, and at first glance, the proposal may seem daunting. However, I truly believe that this would immensely improve Wikipedia and implore you to read it through and offer your thoughts. Thanks!
Flcelloguy
From Wikipedia, the Free Encyclopedia.
[1] http://en.wikipedia.org/wiki/Wikipedia:Quasi-protection_policy
_________________________________________________________________ Express yourself instantly with MSN Messenger! Download today - it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
On 27/05/06, Fl Celloguy flcelloguy@hotmail.com wrote:
Hello, all: I've written up [[Wikipedia:Quasi-protection policy]] [1], a proposal similar to semi-protection that would effectively limit sleeper accounts used to vandalize articles linked from the Main Page. I know that I've written a lot, and at first glance, the proposal may seem daunting. However, I truly believe that this would immensely improve Wikipedia and implore you to read it through and offer your thoughts. Thanks!
I like it, but there's a big logistical burden lying in wait for us. (On the other hand, it would mean that an admin virtually has to eyeball each new user and leave them a friendly comment, which is a good thing)
On the talk page, Jtdirl has quite a good idea - modify semi-protection to be "accounts where first edit was more than four days ago" rather than "accounts where creation was more than four days ago". Less logistics, but certainly a benefit to the sleeper problem.
Andrew Gray wrote: [...]
Hello, all: I've written up [[Wikipedia:Quasi-protection policy]] [1], a proposal similar to semi-protection that would effectively limit sleeper accounts used to vandalize articles linked from the Main Page.
[...]
I like it, but there's a big logistical burden lying in wait for us.
Do we have any statistics on how many new accounts are created each day?
That would help to determine if the burden of manually checking every new user would be unsustainable -- it would be unacceptable for potentially constructive users to be held back from contributing because of an 'account validation backlog'.
Cheers,
N.
On 5/27/06, Andrew Gray shimgray@gmail.com wrote:
I like it, but there's a big logistical burden lying in wait for us. (On the other hand, it would mean that an admin virtually has to eyeball each new user and leave them a friendly comment, which is a good thing)
I don't see how the time saved not having to block a small number of vandals can possibly outweigh the time spent validating a large number of good users.
On the talk page, Jtdirl has quite a good idea - modify semi-protection to be "accounts where first edit was more than four days ago" rather than "accounts where creation was more than four days ago". Less logistics, but certainly a benefit to the sleeper problem.
You could always combine the two. Auto-validate X days after Y edits (say 2 and 10), or manually validate by an admin at any time.
Of course, there are relatively easy ways to get around this. To stop a well-written bot you have to block the IP address(es), not the usernames. But then you've gotta worry about collateral damage, because IP blocks affect *everyone* who uses that IP (even admins, see [[Wikipedia:Blocking policy proposal]]).
And that's really the fear. Not the vandalism itself as it will almost always get reverted. Not even the waste of time, of admins and regular reverting users, although that's up there. But the real fear is that the vandals are going to create a situation where good editors can't edit, or have to cross very high hurdles to do so. The fear is the realization of what Nicholas Carr called "the death of Wikipedia" (he was premature, but it's only a few well executed vandal attacks away).
Anthony
On 5/28/06, Andrew Gray shimgray@gmail.com wrote:
On the talk page, Jtdirl has quite a good idea - modify semi-protection to be "accounts where first edit was more than four days ago" rather than "accounts where creation was more than four days ago". Less logistics, but certainly a benefit to the sleeper problem.
Unfortunately, that will only lead to bots adding spurious interwiki links, null edits, switching commas, and so on. The vandals we're talking about are highly determined, so they're probably not going to be deterred by algorithmic measures alone.
Having a layer between semi-protection and full protection is likely to lead to a lot of added complexity and does have the "jump through hurdles" problem Anthony described. The "positive" way to implement such systems is through clearly indicated "levels" which users can advance in. Everything2 does that, as do most RPGs.
I don't think we want to go that way. If we create a trust-based model, then I think it should replace the current semi-protection model, rather than being layered above it. The question is, how to best implement it?
I think it could be combined with the culture of welcoming users. A "welcome" tab on the talk page might lead to a new screen, which offers two options:
[x] Trust this user to edit semi-protected pages [x] Send welcome message
Hello, Foo, and welcome to Wikipedia!_________ ___________________________________________ ___________________________________________
This wouldn't be too hard to implement. I'm not sure a model where only admins can set the "trust this user" flag would scale, though.
Erik
Erik Moeller wrote:
On 5/28/06, Andrew Gray shimgray@gmail.com wrote:
On the talk page, Jtdirl has quite a good idea - modify semi-protection to be "accounts where first edit was more than four days ago" rather than "accounts where creation was more than four days ago". Less logistics, but certainly a benefit to the sleeper problem.
Unfortunately, that will only lead to bots adding spurious interwiki links, null edits, switching commas, and so on. The vandals we're talking about are highly determined, so they're probably not going to be deterred by algorithmic measures alone.
Having a layer between semi-protection and full protection is likely to lead to a lot of added complexity and does have the "jump through hurdles" problem Anthony described. The "positive" way to implement such systems is through clearly indicated "levels" which users can advance in. Everything2 does that, as do most RPGs.
I don't think we want to go that way. If we create a trust-based model, then I think it should replace the current semi-protection model, rather than being layered above it. The question is, how to best implement it?
I think it could be combined with the culture of welcoming users. A "welcome" tab on the talk page might lead to a new screen, which offers two options:
[x] Trust this user to edit semi-protected pages [x] Send welcome message Hello, Foo, and welcome to Wikipedia!_________ ___________________________________________ ___________________________________________This wouldn't be too hard to implement. I'm not sure a model where only admins can set the "trust this user" flag would scale, though.
How about the same trust model that Advogato uses?
On 5/28/06, Andrew Gray shimgray@gmail.com wrote:
On the talk page, Jtdirl has quite a good idea - modify semi-protection to be "accounts where first edit was more than four days ago" rather than "accounts where creation was more than four days ago". Less logistics, but certainly a benefit to the sleeper problem.
No, that change could be just as easily gamed as the creation system. How would you define "edit'? A user who shows up and modifies his user page? A user who shows up, presses "random page" and makes one trivial edit? These don't take substantially longer than creating the account.
We probably do need either a genuine trust metric, or a genuine human trust system, whereby people can indicate their trust in other editors.
Steve