The Signpost

WikiProject report

Counter-Vandalism Unit

WikiProject news
News in brief
Submit your project's news and announcements for next week's WikiProject Report at the Signpost's WikiProject Desk.
Seal of the Counter-Vandalism Unit

This week, we interviewed the Counter-Vandalism Unit. Unique among WikiProjects, the Counter-Vandalism Unit was created as a bot tied to an IRC channel tasked with identifying and reverting vandalism. Over time, the bot morphed into a WikiProject and gained the trappings of a para-military unit as a result of new members misunderstanding the origin of the CVU's name. After tense discussions about the project's scope and motivations, including several attempts to delete the project in 2006 and 2007, efforts were undertaken to reduce the militant language on the project's page and make the project more inviting to ordinary Wikipedians. Today, the project provides a variety of resources to editors seeking to curb vandalism on Wikipedia. Among these are an academy to teach strategies for detecting vandalism and dealing with vandals, various studies of vandalism, a list of tools and scripts, and a think tank which serves as a forum for ideas on how editors can better protect Wikipedia from vandalism. Included in this week's interview are Dan653, Waggers, Achowat, and the project's founder とある白い猫 (To Aru Shiroi Neko).

What motivated you to join the Counter-Vandalism Unit? How much of your time on Wikipedia is spent cleaning up vandalism?

Dan653: Achowat actually motivated me to join the Counter-Vandalism Unit (CVU). About 80% of my time is spent cleaning up vandalism, with the rest of my time spent working on the CVU, doing a little work on articles, archiving talk pages, adding coordinates to articles, and new page patrolling.
Waggers: A few years ago, before things like Cluebot had really got to grips with vandalism, it was pretty rife on Wikipedia. After trying to tackle it via pure recent pages patrolling it became clear to me that we needed a place to coordinate our efforts on tackling vandalism and I stumbled across the CVU. As well as providing a great place to discuss ideas and list useful tools, the CVU brand helped to make what is fundamentally a pretty dull cleaning-up task into something a bit more exciting.
Achowat: I'd like to think that my experience with Counter Vandalism is very similar to most editors whose first real contributions are anti-vandalism in nature: I used Wikipedia as a resource, realized how useful a tool it can be, and wanted to help. It seems easy enough "See something that doesn't belong, revert it". Over the next few months I had more experienced users direct me to WP:VANDAL, to the user talk templates at WP:WARN, and to WP:AIV. Other users helped me make the most of my time reverting vandalism, to make sure I had the most impact. To answer your second question, I don't spend as much time as I used to looking for vandalism. I've found other ways to be useful to the Encyclopedia, including helping other, newer counter-vandals "learn the ropes", as it were.
とある白い猫: This brings back memories. Counter Vandalism-Unit (CVU) grew out of my userspace. The name was originally intended to be the name of the script/bot that detected the vandalism where the "Unit" was the bot. I suppose this question can be rephrased to "What prompted CVU to be something people can actually join?". A lot of people asked if they could "Join" the unit hence the small user subpage list grew into the Wikiproject it is today. What this achieved is established a median to crowdsource individual users in dealing with vandalism rather than uncoordinated individual effort which is far more efficient. CVU has even spread to other projects.

How useful are the new pages and recent changes feeds in detecting and fighting vandalism? Has the project ever teamed up with any patrols or other WikiProjects?

Dan653: The new pages feed is not that useful in detecting and fighting vandalism as most vandalism that I come across is to an already established article. The recent changes feed was very useful in fighting and detecting vandalism until I received rollback rights and started to use Huggle. We are associated with the Recent changes and New Page patrol and we have absorbed Vandalism Studies.
Waggers: They're both very helpful, although as counter vandalism tools have become more advanced I tend to use the native special pages less. They're especially useful when coupled with other tools – most of my early counter vandalism efforts were a simple combination of Special:RecentChanges and WP:Popups. Like Wikipedia as a whole CVU is very much about collaboration, so we're proud to work alongside other wikiprojects and encourage our members to do so.
Achowat: New Page Patrolling, while they do see a fair amount of vandalism, is such a larger set of skills that, frankly, the New Page Patrollers are better at than we are. You need to be able to know when to speedy, when to PROD, when to AFD...when and which tags to add, if Googling would be useful. New Page Patrollers need to take articles that may not be perfect and work to perfect them. Countering vandalism is a whole different ball of wax. We need to be able to recognize that 1 edit, one change to one line, is either vandalism or not. It's less comparing apples to apples than it is comparing apples to suspension bridges. As for RCP: All RC Patrollers are counter-vandals, and most counter-vandals Patrol recent changes. The CVU has a niche, but it can't simply be a group of people who watch over new changes (that would be redundant to the work done at RCP). If CVU is to be successful, if it means to be a useful tool for the Encyclopedia, it needs to focus on something else. Since the CVU redesign and reorganization (which coincided with the Vandalism Studies merge), CVU has a number of 'divisions' ("Divisions", of course chosen because it's 'snappier' than 'Task Force', in much the same way 'Unit' is snappier than 'WikiProject'). The Think Tank, which aims to be a place of collaboration, a place where CVU members can discuss new ways to do what we do. The Academy, that works to train new counter-vandals. The Task Force, is still being tailored to the new vision of CVU, but hopefully it will serve as a "mailing list" for CVU users to, somehow (we're still working on the technical details), of high levels of vandalism. Vandalism Studies, which will be discussed below. We also maintain a Division for listing and keeping up with the new Tools and Scripts that may be developed to help us out with what we do. I realize I may have gone off on a bit of a tangent, so let me return to your final question. "Do we work with other WikiProjects?" The answer is "No, how could we?". Unlike other projects, we don't claim articles, so we don't have an article that we 'share' with another Project (in the way that, say, WPCanada and WPTrains could share "Railroading in Canada"). We'd like to work with as many people as possible, but our job (like RCP) is generally 1-editor;1-keyboard kind of work.
とある白い猫: I think NPP and RC patrol are a valuable asset to Wikipedia but I feel we as a community have grown too dependent and attached to them. NPP and RC patrol is a small fraction of the issues that needs to be addressed daily. And for the other point, I do not believe CVU ever was an independent entity. The scope of most WikiProjects are limited where they can afford to remain independent of other wikiprojects. The moment people were able to join CVU its scope was increased to include every page hence cooperation with other projects is by very nature unavoidable. Specific examples have bee given by other people so I needn't repeat what they said.

Do you use any tools or programs in your vandal-fighting duties? Are there any functions you wish were built directly into Wikipedia's interface to help in detecting and fixing vandalism?

Dan653: I mostly use Huggle and Twinkle, but I have tried Igloo, and STiki. I think the current functions do a good job in detecting and fixing vandalism, but an iOS version of Huggle would be great.
Waggers: I do use a number of scripts and tools; things like Popups and Twinkle are essential. Huggle is brilliant and I used to use it a lot, but I moved away from the Windows OS a number of years ago and so haven't been able to use it. I now prefer things that work within the browser rather than needing to be installed as separate applications – I've recently started using Igloo, it has a lot of potential, and I sometimes use WP:LUPIN. What we really need now is something similar to Huggle or Igloo that can work on a smartphone.
Achowat: I do things by hand and, while that's slower, I think there are times where counter-vandals get a bad name by working too quickly. I don't mean to disparage the fine work semi-automated counter-vandals do, but I personally like the slower pace of undo and rollback allows me to better "Make one edit right, and then repeat it 45 times an hour"; even if it does mean that it's harder to "Make 45 edits an hour". The one technical function I'd like to see (a pipe-dream, I understand, since it would likely cost far more than its worth) is replacing "Show anonymous users"/"Show logged-in users" with a system that breaks it down by Confirmed or Autoconfirmed and "New users and IPs". That will help us take care of vandalism-only accounts much quicker than the current system does.
とある白い猫: I was going to present something about this on Wikimania 2012 but there just weren't enough slots. We live in an era of Artificial Intelligence (AI) and AI can be utilized and perhaps even integrated to assists users find problematic edits. There is an international conference called CLEF which has a PAN workshop that has cutting edge research competing to be the best in Automated Vandalism Detection, Automated Author Identification (could be utilized to identify long-term abuse), Automated Plagiarism Detection (could be utilized to identify copyright violations).

How often do you encounter vandalism from IP users? How difficult is it to fight vandalism without biting newcomers? Have you ever been able to convert a vandal into a productive part of the community?

Dan653: Most of the vandalism I encounter is from IP users; "those that vandalized pages were overwhelmingly unregistered editors, who accounted for 96.77% of all vandalism edits" (Vandalism Studies). It is not difficult to fight vandalism without biting newcomers. I have not been able to convert a vandal into a productive part of the community, though I have tried.
Waggers: Certainly most vandalism comes from IP users; Educational IPs particularly seem to be a problem. At some point I think there may be a case for pre-emptive blocking of such addresses, but that seems a very harsh approach. Fighting vandalism without biting newcomers is easy – it's really just a case of using the user warning templates in the right order and engaging with the user if they try to explain what they're trying to do – but I'm frequently dismayed how often new users are reported to WP:AIV without having been given so much as a level 1 warning, or when their edits are clearly a misunderstanding rather than deliberate vandalism. I know I dropped a bit of a clanger when I first started editing and if I had been treated as harshly as some editors treat newcomers now, I probably would have given up back then.
Achowat: To add to Dan's comments, "Most vandalism comes from IPs, but most IP edits are not vandalism". We do a disservice to ourselves, to the editors behind the IPs, and to the encyclopedia itself when we paint all IPs as vandals. That being said, the current community-accepted User Talk templates at WP:WARN, I believe, do an amazing job at combating both vandals and keeping us from being too BITE-y. They're all of the format "Hey, welcome, it looks like your edits might be vandalism, please stop" – "Your edits appear to be vandalism, use the Sandbox if you want to experiment" – "If you continue your vandalism, we may block you" – "That's it, one more and we block you!". The benefit of the Templates (despite making sure that each counter-vandal says exactly what the community wants them to) is that each template allows time for the templated User to engage in a conversation. The IP-vandalism problem also makes it quite difficult to "see" that an editor has gone from vandal to editor. But there's a difference between a vandal and someone who has vandalized a page. And vandals are entirely incompatible with being productive. It's an issue of mindset. Wikipedia is a project that "fails in theory but works in practice". I'm typing this right now because I think that the Signpost promoting CVU will lead to greater participation in CVU and improve an encyclopedia that could care less if I lived or died. It takes a special kind of someone to do all the work that Wikipedians do, especially given that our only reward might be a picture of a star anchor. Vandals are people who maliciously and thoughtfully try to disrupt that goal, the goal of a free repository of all the world's information. No one like that could ever be turned. However, when issuing a {{uw-vandalism1}}, it's really quite hard to know what the person who made the edit you thought was vandalism was thinking. And we have rules about that sort of thing.
とある白い猫: It is a statistical fact that most vandalism comes from IP edits. This, however, does not mean all IP edits are problematic. It is important to remember that the vast majority of IP edits are intended to improve the wiki and are not vandalism. Patrolling IP edits requires care. Your vandals typically do not care about the site and will not be hurt no matter what you tell them. However, newbies do care a lot about the project and will quickly be completely alienated from the project if they are bitten. Bitten IPs can turn vandals as well.

How often do you deal with trolls intentionally causing disruptions? What is the best way to handle a troll? Have you been involved in the process of blocking or banning such a user?

Dan653: I rarely deal with trolls, mostly vandals. The best way to deal with a troll is to let them know that what they are doing is wrong and can lead to a block. If they continue to troll I let an admin know. I have only dealt with three trolls, two of them involved in an edit war (they both got warnings) and one who was continually removing content (he was blocked indef). Including hir I have reported 35 users to AIV and 3 from RFPP, all of which have been blocked.
Waggers: Genuine trolls are a lot rarer than people think. A lot of people are labelled as trolls when they cause disruption – but often their intentions are pure; they're not being intentionally disruptive. The odd thing about trolls is they can come from anywhere – quite often from good, established users who get a bee in their bonnet and vent their frustration by disrupting the project somehow. I think that's a lot more common than someone coming along new to the project whose sole intention is to disrupt it – usually such users' edits are simple vandalism and easily dealt with. The important thing is always to start off by assuming good faith and asking the user to explain what they're doing. Give them a fair chance. But once it's clear they're a genuine troll, don't let them disrupt the project further by discussing them to death – block/ban swiftly and move on. As an admin I find the sheer length of discussions at WP:AN and WP:ANI can be quite frustrating when the conclusion seems fairly obvious.
Achowat: Waggers is 100% correct. Trolls aren't the kinds of people who disrupt Article space. Not anymore, they're too smart for that. A vandal will make his 4 vandalism edits, take and ignore each template, and get reported and blocked. A troll has bigger goals, and you find them on the Talk pages of controversial subjects: religion, politics, etc. They know that they can cause more anguish and torment by defending some fringe theory and demanding inclusion thereof than by starting an edit war over it or vandalizing the page. The community really needs to set a standard for how much Talk page trolling and how much incivility we're willing to accept (like we have with the "sufficient warning-AIV" system for vandalism). That, unfortunately, seems wholly out of the scope of CVU.
とある白い猫: While trolls are outside the scope of CVU, my advice would be resilience. Trolls may find ways of getting under your skin as much as you try to prevent it but in the end all that matters is resilience. Do not let trolls decide how you edit the wiki.

The Counter-Vandalism Unit recently absorbed the vandalism studies project. What is the intent of this project and how has its research been used to fight vandalism?

Dan653: The intent of the project is to conduct research related to unconstructive edits.
Waggers: It's a case of "know your enemy". The more we know about how, where, when and why vandalism happens, the easier it becomes to clean it up and, hopefully, prevent it happening in the first place. It's important we do that, but equally important that we deny recognition, so it's a tricky balance to get right.
Achowat: Waggers again makes a compelling point, but along with WP:DENY, we don't want to give them any ideas. It's a delicate balance, but it's important to have hard data when we talk about how best to handle disruption. Otherwise we rely on ol' wives tales about vandalism, which is never preferred to the actual numbers. (Addendum) Now that the Academy seems relatively self-sustaining, and since serious technical work is being put into revitalizing the Task Force, I imagine the next "Division" that we'll hope to revive is Vandal Studies. The last studies gave us good numbers, but they're frankly too old to be of use now. Wikipedia is significantly different than it was in 2007. I suspect by September a full study should be in the works, followed shorty (or, perhaps, preceded shortly) by a re-hash of the Obama article study with, perhaps, a Mitt Romney study, or the like. Really, we need to look to see if our information is still accurate. I would also like to see those studies expanded to include 1. What, if any, tools or scripts are anti-vandals using? 2. How effective is ClueBot NG at combating vandalism? 3. How effectively are vandals warned and reported? 4. Do recognized shared-IPs (including schools) have a dramatically higher or lower instance of vandalism than IPs in general?
とある白い猫: Knowledge comes from studying past behavior. Understanding and detecting the pattern will improve detection. AI-assisted tools may be a good improvement in detecting such patterns.

There is an academy at the Counter-Vandalism Unit intended to educate and coach vandal-fighters. How long has this initiative been around and what does the project hope to achieve with the academy? What are some basic lessons about vandalism every Wikipedian should know?

Dan653: The Academy has been around for a little over a month. The project hopes to give budding vandal fighters one-on-one coaching with an experienced vandal fighter. The project also teaches all enrollees to identify vandalism, recognize other kinds of disruptive editing, and how to warn and report those who persist in vandalism. Reading WP:Vandalism will give most Wikipedians a general knowledge about vandalism.
Waggers: I think the basic lessons about vandalism are really the same as the basic lessons about Wikipedia in general. Assume good faith, be bold, be civil, and maintain a neutral point of view. That last one may seem strange in a counter-vandalism context, but a lot of people report editors for vandalism when really it's just a difference of opinions that they need to resolve between them using reliable sources.
Achowat: WT:CVUA should give you the 'history' of the Academy, from the first suggestion on 25 January until its opening on 20 April. I'm not going to rehash what was said there, but if there are any questions about it, feel free to drop me a line. The Four Steps to counter-vandalism are the biggest thing that anyone who'd like to deal with vandals is.
  1. Identify that an edit is vandalism
  2. Restore the page to the last not-vandalized edition
  3. Inform the editor that the edits were perceived as vandalism
  4. Report the editor if vandalism persists.
And, of course, the four policies, help pages, and how-to-guides WP:Vandalism, Help:Reverting, Wikipedia:Template messages/User talk namespace and Wikipedia:Guide to administrator intervention against vandalism are must-reads.
とある白い猫: The intention in forming CVU was to be more of a learning grounds, particularly for users unfamiliar with dealing with vandalism. Anybody can revert vandalism, but efficiency can be greatly improved if people can be taught how to more effectively use tools and how to avoid biting newbies.

What are the Counter-Vandalism Unit's most urgent needs? How can a new member help today?

Dan653: To my knowledge the CVU has no urgent needs. A new member can help today by reverting vandalism, becoming an instructor, or enrolling in the academy.
Waggers: The CVU is one of those organisations that wishes it didn't have to exist. And thanks to things like Cluebot NG, thankfully the need for active (human) vandal-fighters and patrollers is waning. So the needs now are really about the long-term view. We're pretty good at handling vandalism now when it happens, but need to do much more about preventing it happening in the first place. Oh, and that smartphone version of Huggle/Igloo would be great if someone has the ability!
Achowat: The Biggest needs are those that aren't directly countering vandalism. There are enough people, and ClueBot, working counter-vandalism that if someone read this article and thought "How can I help?", there is diminishing returns on that person working the Recent Changes Queue. The biggest need right now is Vandalism Studies. We haven't had a study in over 5 years and, frankly, our information is likely to be wrong. It's hard to talk about taking that long-term view until we know what we're up against.
とある白い猫: What Wikipedia and CVU alike need is better tools that better analyze and help deal with vandalism and other unwanted content (spam, copyvios, etc). Breakthroughs in Artificial Intelligence in the past 5 years in particular can lead to tools that can significantly improve efforts.

Anything else you'd like to add?

Waggers: In the past the CVU has had a somewhat mixed reputation, with accusations of people "playing cops and robbers" with vandals and the like. I really hope that this Signpost article serves to rectify some of that.
Achowat: The biggest initial thing you'll notice about the new CVU is the stripping of the paramilitary pretense we once assumed. We don't talk about "going into combat" and "blasting away vandals with our M1-Abrams Scripts". There have been 4 MFDs of CVU, and the paramilitary trappings always come up. Even people who don't mind the idea of a Counter-Vandalism WikiProject think that pretending we're fighting back the roving hordes that wish to do harm to our precious village is unnecessary. I, personally, think it is not only unnecessary, but also dangerous. We BITE, we assume bad faith, we come to think of any strings of numbers like an enemies flag. CVU is just a group of Wikipedians who want to see disruption kept to a minimum and give every Wikipedian the tools they need to be successful in that aim. As such, as part of the redesign of the CVU homepage (which was more about cleaning clutter, honestly) I personally chose to tone-down the language (you will no longer find references to fighting, combating, or eradicating vandals) and chose not to list some CVU UBXs that supported that mindset.
とある白い猫: CVU from its formation had been the target of significant negative sentiment due to the belief that it had a para-military structure. I think this pretense is unfair, as CVU never operated like a military body. To be blunt, I was never the commander-general, nor was anyone else for that matter. I wish people assumed more good faith towards CVU and I hope this Signpost coverage may alleviate some of the negative sentiment.


Next week, we'll interview some punks. Until then, rock out in the archives.



















Wikipedia:Wikipedia Signpost/2012-06-11/WikiProject_report