User talk:Yaris678/Deny automated recognition

Bite

You have to look at WP:DENY in the context of WP:Bite. Not naively, because vandalism should never be permitted, but optimistically, insofar as many people initially vandalize out of sheer curiosity--as a way to see that editing is possible--and then out of genuine interest--to see that bad edits are actually caught. This is part of the way users come to believe the encyclopedia is both open to them and reliable enough to constructively contribute to. However we jigger the system, it has to be designed with those facets in mind. To many, this is still a new and weird concept, familiar but unfamiliar, frequented but mysterious. That learning curve can take a while, and the cost of catching vandals a few edits early needs to be weighed against each one who might not become constructive in time to avoid a block that will persuade them it's not worth it. That said, I think two warnings is enough--the first welcoming and the second clear but direct--the 3rd should be the last--making clear that the next one will result in a block--if there is a third one at all. Maybe after 2 vandalisms, a user should be mini-blocked, for 2-6 hours or something. Then they can always try again. Ocaasi (talk) 14:41, 10 March 2011 (UTC)[reply]

Thanks for the input. I agree that WP:Bite is important. I wasn't totally thinking along the same lines as you, but that is roughly what I was getting at with "if an otherwise-helpful user has made a small number of edits that are construed as vandalism, then moving swiftly to a ban is not helping Wikipedia. A polite warning, indicating how their edits can be perceived may be just what such an editor needs." I guess the key difference is that you would be prepared to give a bit more slack to very new users than the 0 warnings that I put in the recommendations section.

The essay, and especially the contents of the recommendations section, are totally open to input by others. I note that you think that four warnings seem too many. Do you agree with my thought that the one-month reprieve is too short?

How would you feel about a table like this:

Number of edits	Number of warnings before AIV
0-10	1
11-100	2
101-1000	3
1001+	4

Yaris678 (talk) 16:09, 10 March 2011 (UTC)[reply]

P.S. I quite like your mini-block idea... but I am thinking of this essay as being about what happens before AIV. I guess we could widen the scope. Yaris678 (talk) 16:13, 10 March 2011 (UTC)[reply]

I think for under 100 edits, 2 warnings would have to be standard. Once 100 is hit, they're basically constructive, so the need for 3 blocks isn't really there. I don't know why 2 blocks then AIV isn't standard across the board. Maybe 3 blocks for under 100, then 2 blocks for over 100. I guess part of the issue is that with CBNG, Huggle, and STiki doing so much great work, no one really cares about the scourge of vandalism anymore. It (almost) all gets caught and as the cost to correct it goes down, the relative value of deterrence does as well. I have a feeling most of this discussion will be a non-starter, just because Blocks are much more invasive than warnings. But it'd be interesting to see what others think. There's def. room for some improvement in the whole anti-vandal picture. Ocaasi (talk) 19:18, 10 March 2011 (UTC)[reply]

Cool. I have edited the essay along the lines you describe.

I think your analysis of the complacency on this issue may be correct. Unfortunately some vandalism will stay in the encyclopedia for considerably longer than we would like. And even if we had perfect anti-vandal tools, encouraging vandals is generating work for admis. I have added some words on generating work for admins.

I am considering adding something about the edit summaries. For example, the default message from STiki refers to "test/vandalism" which might be seen as giving the issue too much attention. Obviously, if we are giving a warning then we have mentioned the fact that it might be considered vandalism already, but I think the less the word is mentioned the better.

Yaris678 (talk) 07:54, 11 March 2011 (UTC)[reply]

rename the essay?

I don't think that this is so much about wp:Deny as it is a proposal to sharply reduce the default number of warnings before issuing blocks. I see a few problems here, firstly not every warning is correct, and moving blocking to more of a hair trigger will mean more incorrect blocks. Secondly many of our best and worst edits are by IPs, moving to more of a hair trigger means blocking more good edits by IPs. Thirdly our current warning system with four escalating levels is designed to deter and to rehabilitate. I don't know what proportion of warnings succeed in either respect, but I think that any proposal to reduce the number of warnings should be accompanied by some stats about the efficacy of the various current warning templates. That isn't to say the system can't be reformed, but I think we need info as to whether 3, 4 or even 5 warning levels would work best. Ϣere SpielChequers 17:18, 12 March 2011 (UTC)[reply]

As the template at the top of the recommendations section says, that bit is tentative. I don't really want to call the page a proposal for anything in particular because what it is recommending may change.

Why do you think the page has not much to do with WP:DENY?

Yaris678 (talk) 17:29, 12 March 2011 (UTC)[reply]

P.S. I agree that IPs complicate the situation. I have been thinking about adding some IP-specific recommendations, but I am not sure what to add. I think templating shared IP addresses can be counter-productive. In many cases it would make sense to just quietly revert the vandalism and allow the good stuff to continue.

Before I noticed your comment, I asked a question at the village pump asking if there is any current policy in the area. Yaris678 (talk) 17:38, 12 March 2011 (UTC)[reply]

Because I don't see the number of warnings on a page as giving significantly greater recognition to the vandal. A proposal to delete vandals user and talkpages and revdelete their edits would fit in with wp:Deny, - not that I'm suggesting that as I believe the cost would outweigh the benefits. But the proposal as currently worded is more about implementing a hair trigger and blocking vandals after far fewer warnings than it is about wp:Deny. Ϣere SpielChequers 17:44, 12 March 2011 (UTC)[reply]

OK. Try not to see this as a proposal. I know I mentioned it at Wikipedia:Village pump (proposals), but that was my mistake.

As I said, the recommendations are tentative. The real point I am making is that the current system is likely to give the vandals the impression that they are being disruptive. WP:DENY is all about how vandals want to be disruptive and about how sometimes we can end up encouraging them, which is what this essay is about.

Yaris678 (talk) 17:57, 12 March 2011 (UTC)[reply]

Vandals are disruptive, that's why we block them. At the moment your motive for the essay may be to alter deny but the effect of what you propose would be to move blocking onto more of a hair trigger. That is a very different thing. Ϣere SpielChequers 15:24, 14 March 2011 (UTC)[reply]

Alter deny?

Also, if you think that 2 warnings before posting at AIV is hair trigger I can't do much about that. However, I would point out that:

Posting at AIV is not the same as blocking. Administrators have the final say. The real question is whether or not we would be generating a load of false positives for the administrators to assess. The argument put forward here is that in some cases, additional warnings actually generate more work for the admins, by encouraging vandals.
The essay does not recommend going to AIV after two warnings in all cases. In the case of shared IPs, a lot more vandalism may take place before AIV is requested, more even than the four acts in the current system. However, the essay does recommend that we don't give a warning in each case. Such warnings will only goad people to further vandalism and/or confuse people who are sharing the same IP address.

Yaris678 (talk) 18:48, 14 March 2011 (UTC)[reply]

Currently the default is the four level warning system, I read your essay as proposing a truncation of that default from 4 to 2 - hence my hairtrigger comment. If that sort of change isn't your intent then you might want to tweak the essay accordingly. Ϣere SpielChequers 14:45, 15 March 2011 (UTC)[reply]

I have done quite a rewrite of the essay. I think the intention is now much clearer. Yaris678 (talk) 20:13, 3 April 2011 (UTC)[reply]

No improvement

I fail to see how this is a big improvement over our current system. The four warnings is a suggestion, not a requirement, hence templates like {{vandalism4im}}. Depending on the type of vandalism it may be appropriate to give someone a higher or lower warning (e.g. vandalism which is obviously malicious and difficult to catch would warrant a higher warning). All of the solutions you are proposing would introduce a lot more complexity to the software. It's not that it couldn't be done, it's just it would require quite a bit of work and reduce the performance of the tools. As of right now there is no unified way other than the talk page to track warnings for a user. A major problem with the talk page is that things can, and often are, blanked. Making the only way to verify that nothing was removed is to pull every revision of the page. This introduces another problem, things which are blanked because they were added accidentally and handwritten warnings. --nn123645 (talk) 21:27, 12 March 2011 (UTC)[reply]

I agree - four warnings is a suggestion, not a requirement. The problem is that a tool like WP:STiki automatically gives four warnings. STiki is great, but it obviously misses some of the subtleties of a person doing anti-vandalism manually. This essay is about the problems that causes and recommends some remedies. Yaris678 (talk) 14:43, 14 March 2011 (UTC)[reply]

More on IPs

I am considering adding more on IPs. I don't think any of the tools do anything that is suggested in WP:R Van in the bit starting "For repeated vandalism by an IP user..."

It may also be worth saying something about Wikipedia:Abuse response, but I am not sure what.

Yaris678 (talk) 15:06, 14 March 2011 (UTC)[reply]

Need a holistic approach

The scope implied by the title is too narrow. We need to look at all of our tools as a coordinated response. A revert by Cluebot says one thing about an editor. A non-vandalism revert by a STiki users says something else (STiki needs to add a separate button for this). We need to consider the types of editors involved and tailor our our response for accepting edits, giving warnings and issuing blocks accordingly. For example:

Grade schools: Frequent test edits, occasional minor vandalism, shared IP addresses and shared computers.
- Have edit filter detect test edits and automatically redirect such users to the sandbox.
- Zero tolerance policy for vandalism. On each instance of vandalism, perform automatic 2 hour block of IP editing, but allow account creation. If consecutive blocks are required, increase the block time. If account creation is abused, then block that as well.
- Have edit filter silently reject gross vandalism.
- No talk page warnings. We only do it today because we feel helpless. It's time to find a better way.
Other types of shared IP addresses:
- Similar to grade schools, but perhaps start with longer term editing blocks.
- More tolerance on permitting account creation.
- Subject edits on new accounts to pending changes for a probation period of 24 hours (assuming that anything sensible can ever come of pending changes like supporting edit by edit protection). Increase probation period if accounts are getting blocked after probation expires.
New accounts from non-shared IP addresses:
- Except in cases of clear vandalism, leave a helpful note instead of a warning.
Auto-confirmed accounts:
- Escalating warnings over a rolling time period.
- Tailor response based on automated analysis of edit history.
- Block if a repeat edit is made after a warning has been issued.
Accounts with over 250 edits:
- Disable automatic warnings. Instead, warn vandalism patroller that the edit is from an established account and require that a personalized message be left if really warranted.

The above isn't intended as a final list. I have tried to cover all types of responses and show in what cases they may be most useful. I proposed redirecting grade school users to the sandbox due to the high number of '''bold text''' edits that we see in such cases. It would be great to redirect all test edits to the sandbox, but we would probably get too many false positives for other editors. I have left very little use for the current series of escalating warnings because I have found that they don't do anything. Good faith editors will soon complain on my talk page while bad faith editors will just keep on going. The exceptions are the single issue warnings for things like WP:MOS that can be valuable, but I would rather classify these as helpful notes rather than warnings. —UncleDouggie (talk) 02:35, 15 March 2011 (UTC)[reply]

Hi UncleDouggie,

Thanks for the Barnstar! Never had one of those before. I am chuffed.

I am content with the current essay name and the scope that implies. That said, I like some of your ideas and I can see a lot of value in your holistic approach. Perhaps it could be expanded upon in a different user essay. Logically, the two essays would link to each other...

Yaris678 (talk) 14:31, 15 March 2011 (UTC)[reply]

Fair enough. Before I go create such a thing, is anyone watching this page aware of an existing essay that covers the same ground? —UncleDouggie (talk) 00:37, 16 March 2011 (UTC)[reply]

Not that I am aware of... I suppose you could ask at Wikipedia talk:WikiProject Essays. Yaris678 (talk) 10:35, 18 March 2011 (UTC)[reply]

Welcome versus warnings

I've looked it over, and I have ambivalent feelings. I understand the idea, but I'm not really persuaded that it's the central problem. I definitely do think that uncritical placement of "Welcome to Wikipedia" templates on vandal talk pages boomerangs on us in the same way. But I'm not convinced that user warning templates are reinforcing to most vandals. Some are genuinely convinced by them to stop vandalizing, and they do stop before a block becomes necessary. That's a good thing, and I've seen it happen numerous times. Of course, the dyed-in-the-dirty-wool trolls may be encouraged by it, but then they get blocked, and who cares. Would it be more productive to focus on welcome templates instead of warning templates? --Tryptofish (talk) 00:26, 5 April 2011 (UTC)[reply]

Thanks for the link. "Minimize talk page templates" is a good principle in general, especially when dealing with new users... and that principle needs to allow for some templates to be used in appropriate circumstances. I agree with the general thrust of what you are saying.

From that perspective, the system we have for dealing with vandalism could be seen as one aspect of a wider problem. However, I still think its an important aspect. Take a look at User talk:71.245.232.34 - surely all those talk page messages are counter productive. Undoubtedly a lot of people will use that IP address and be completely unaware that the talk page exists. But those that see it will be more likely to see Wikipedia as "the enemy" - someone to outwit, rather than a community to work with. And some of the people who would never even consider vandalism might just be scared off. A definite case of biting the newcomers.

I agree that some people are genuinely convinced by talk-page messages to stop vandalising. That is why the essay says "Of course, there is the potential for a polite message to encourage someone to stop vandalising. However, a polite message is not the same thing as using a talk page as a vandalism-recording system." Do you think that it should expand on that point?

Yaris678 (talk) 11:48, 6 April 2011 (UTC)[reply]

I don't claim to know the answer to that. Yes, that IP talk page is quite an example in futility! Maybe a reason to archive old threads on such pages? I guess a case could be made that such talk pages are not so much a recording system (we don't really need them for that purpose, since there's a record anyway) as a system for communicating (or at least trying to communicate) with the vandal. Another thing that occurs to me is that it is often silly to wait out the entire four warnings before a block. I'd never support a block without a warning, probably two, but sometimes there comes a point where WP:DUCK outweighs WP:BITE. --Tryptofish (talk) 18:03, 6 April 2011 (UTC)[reply]

I agree that there are cases where WP:DUCK outweighs WP:BITE. However, I think the biggest scope for change in that regard is when the vandal has a username. Admins are reluctant to block IP addresses because they could be used by someone else who is more productive. That is why the essay argues for editor differentiation.

Can I ask you about the statement "I guess a case could be made that such talk pages are not so much a recording system (we don't really need them for that purpose, since there's a record anyway) as a system for communicating (or at least trying to communicate) with the vandal."? Are you saying that the current system isn't using the warnings as a recording system? Or are you describing how it should work? What is this record that you refer to? Is it the users contributions? The essay is arguing that this be used as a record when it asks for user-contributions analysis. However, at the moment, tools such as ClueBot NG and STiki just count the number of warnings and escalate by one... or request AIV if four warnings have been made that month. That's what I mean by using the talk page as a vandalism-recording system.

Yaris678 (talk) 18:34, 6 April 2011 (UTC)[reply]

Well, I guess you have thought more about what I said than I did! I was thinking about the fact that vandals sometimes blank their talk pages but admins at AIV aren't fooled by it, which wasn't a very deep insight. Actually, I'm not really convinced that there is a problem that will be solved by the suggestions here. If you go forward with bringing this to the community as a more formal proposal, you'll need to have evidence in the form of examples of vandals who were encouraged to vandalize more by getting recognition, and show that they are a bigger problem for the project than would be vandalism that can be prevented when vandals do respond positively to warnings. --Tryptofish (talk) 21:31, 7 April 2011 (UTC)[reply]

That's an interesting issue. I have added some stuff to the essay indicating why it could be important - because of the shear number of edits that anti-vandalism tools are responsible for. However, if people disagree with the assertions on vandal motivation made by WP:DENY then they are unlikely to be convinced by this essay. I guess one could implement some of its recommendations and see if they have any effect... but obviously you are talking about before any of the recommendations are implemented... and even then people would argue about the meaning of the results. Unfortunately:

You are unlikely to get a vandal responding to a questionnaire on motivation.
It is very easy for vandals to user multiple user names and IP addresses, so it is difficult to connect acts (or omissions) of vandalism to warnings given.

Yaris678 (talk) 14:09, 13 April 2011 (UTC)[reply]