What percentage of Wikipedia vandalism is done by registered users?
A few wiki bloggers are reporting the misleading statistic that 97% of vandalism is done by “anonymous” users when they actually mean users who are not logged in to Wikipedia. Logging in doesn’t mean you are no longer anonymous since there is no requirement to use real names on Wikipedia.
The small scale Wikipedia Vandalism Study showed that out of 31 instances of obvious vandalism across 100 actively edited articles, 30 of the edits were by users who were not at the time logged in to Wikipedia. I’m not convinced the single occurrence of vandalism by a logged in user was vandalism or an edit made in good faith which may or may not have been accurate. The edit involved changing “Daisy Murdoch Cortlandt” to “Daisy Murdoch Cortlandt Cortlandt” which looks wrong, but is how IDMb have the name.
I carried out my own study looking at vandalism made to my user page. My page has had 182 malicious edits since I joined Wikipedia in February 2003. 47% of the vandalism was done by registered users. See my write up on the wiki for more details.
Even if most vandalism is done by unregistered users, I don’t see this as evidence that a wiki should force editors to register. Forced registration doesn’t prevent vandals logging in. The only difference such a rule would make is that you would then have 100% of vandalism made by registered users, which makes it much harder to spot. Vandal fighters can filter out edits by registered users when they look at recent changes, and anti-vandal bots can be programmed to watch for unregistered users. Removing that distinction only makes it harder to find vandalism - it doesn’t stop it happening. Vandals will sometimes log in if they have to. And genuine users sometimes won’t. I’m assuming vandals have more time to waste than the average Wikipedia reader, so there’s the danger that more vandals than normal people will have time to register, changing the balance of good users to bad.
Nick Jenkins said,
March 30, 2007 at 1:27
182 malicious edits! Wow, that’s a lot….. :-(
I’m curious about why you encountered so much vandalism, so I added a few items to your “ideas for future studies” list…. but if you don’t like them please feel free to revert, and if you really really don’t like them perhaps increment that malicious edit count up to 183 ;-)
– All the best, Nick.
Nick Moreau said,
April 2, 2007 at 15:29
Good to hear you’re perspective on it. I was initially duped into believing it at face value, until I read your post.
Quick typo: IMDb, not IDMb.
Raul654 said,
April 2, 2007 at 19:28
Basing your study on edits done only to your user page biases the study. As a high profile Wikipedian, the people who vandalize your userpage are predictably more sophisticated than ordinary article vandals, so it’s not very surprising that more of them register nicknames.
Angela Beesley said,
April 2, 2007 at 21:16
You’re right. It certainly biases the study, but it also points out that vandalism patterns do differ across different parts of Wikipedia and blocking unregistered users would not solve all of the vandalism problems. I think more studies on “sophisticated” vandalism compared to simple test edits or junk additions would be interesting.
Jon said,
April 11, 2007 at 23:13
Angela,
Your study was originally included in Wikiprojects:Vandalism Studies, and was carefully looked at before beginning the first study. I think the goal is to create several different kinds of studies so we can have multiple perspectives. Perhaps I’m misreading the tone of your post, but I don’t see your study as contradicting the Vandalism Studies, but rather augmenting them.
To that end, your userpage study is a noteworthy perspective. Personally, I’ve never had anyone edit my userpage. If anyone did, I think I would consider it vandalism, or at least presumptuous! But your data considerably expands the understanding of vandalism.
I take your point that “blocking unregistered users would not solve all of the vandalism problems.” Certainly, changing the anon IP policy is a touchy topic on Wikipedia! Personally, I believe the Vandalism studies deserve a warm tip-of-the-hat for being conducted without a partisan agenda, and with a very strict focus on being rigorously objective.
How Wikiproject:Vandalism’s findings are interpreted and acted upon is of course up to the community. But my understanding is we are going to conduct several different types of vandalism studies in the coming months to create a large, diverse, and hopefully objective set of data.
Study 2 is about to commence and I’ll be curious to hear your input as data comes in. Cheers!
Jon
Angela Beesley said,
April 12, 2007 at 0:52
Jon - Thanks for your comments. One correction - I didn’t do this until after the first study. The random sampling approach of the 2nd study should be interesting.
Jon said,
April 12, 2007 at 1:27
You’re right! I don’t know why I remember this differently. But it looks like User:Remember added your study on March 29, and Study 1 was completed by March 27. My mistake.
Jon
Teratornis said,
December 19, 2007 at 7:59
Given Wikipedia’s notion that we should source all our claims, I don’t understand how you seem to be content with mere speculation on the impact registration would have on vandalism. What happened to testing our hypotheses? The only way to know what effect requiring logins to edit will have on Wikipedia is to try it for a few months, and keep careful track of what happens to the rates of destructive and constructive edits.
I find it puzzling that advocates of the current policy seem to think they have no need to run such a test.
Wikipedia has in fact partly tested the hypothesis, by semi-protecting some pages, and by requiring logins to create new articles. What has been the effect of restricting the editing privileges of non-logged-in users in these ways?
Has the level of vandalism to semi-protected pages been the same as it was before the pages became semi-protected? (If that were true, then there would be no point in semi-protecting any pages.) If semi-protection reduces the rate of vandalism, that may tell us something about the willingness of vandals to register accounts and vandalize under them.
Has semi-protecting pages been shown to harmfully reduce the rate of constructive edits to them? Has it been shown to make vandalism to semi-protected pages harder to detect?
Has restricting new page creation to logged-in users reduced the rate of malicious new page creation? Has it harmfully inhibited the creation of needed new articles?
I suspect that requiring users to register accounts and log in before editing is like putting locks on our doors. Locking our doors by no means prevents determined burglars from getting into our houses. Anyone who really wants to get in can easily get in. But the lock creates a small physical barrier, and a psychological barrier, which deters casual thieves. Imagine leaving the doors of your house wide open, and leaving for a trip - the odds of your house being violated would be many times higher than if you locked your doors, assuming your neighbors would ignore the open doors and not close them for you. Another analogy: the ropes for crowd control that museums set around exhibits. Anyone who wants to step over the rope can easily do so, but most people do not. The rope is effective for signaling a boundary.
The argument that requiring logins may discourage positive contributors who find it too burdensome to take a few seconds to register is hard to fathom in light of the stupefying complexity of Wikipedia’s policies, guidelines, and procedures for editing. Anyone who wants to do much more than merely correct typos must spend many hours reading the extensive online manuals, puzzling over other editors’ edits to their edits, possibly asking questions on the Help desk, and so on. (I went through this exercise recently enough to remember it clearly, and I am still going through it.) Take a look at John Broughton’s Editor’s index at WP:EIW with its mind-boggling 2000+ links. The amount of time and effort required to become even modestly competent at Wikipedia editing completely dwarfs the labor of registering an account. If someone lacks the few seconds it takes to create an account, how are they going to find time to learn enough about Wikipedia editing to insure that most of their contributions are constructive?
Even if requiring logins to edit drove away all of the (presumably) millions of unregistered users, the English Wikipedia has more than five million accounts now. Call me optimistic, but I suspect five million registered users are sufficient to write an encyclopedia (even allowing for all the inactives). If five million is not enough, the rate of new account creation remains brisk, and should double again in a few years.
Another concern is the impact on expert contributors. It’s hard to imagine many people who are leading authorities in their fields wishing to continue editing on Wikipedia after seeing their work vandalized by schoolkids under IP addresses. Even though reverting such vandalism is easy enough for people who spend a lot of time doing that, how many people who are actually good enough at something to have become prominent will want to locate and plow through the intricate instructions for reverting vandalism properly, and spend their time repeatedly cleaning up the silly vandalism Wikipedia invites? We have no way to quantify the potentially positive contributors who look at the vandalism and simply leave. This may be a hidden cost of vandalism that has not received its fair consideration in the registration debate, the damage to Wikipedia’s credibility among people of significant intellectual attainment. This is not to say Wikipedia doesn’t attract lots of smart people - it does - but I wonder how many of the world’s experts are also sufficiently long-suffering to live with the current open-edit policy.
Angela Beesley said,
December 19, 2007 at 10:40
> Given Wikipedia’s notion that we should source all our
> claims, I don’t understand how you seem to be content
> with mere speculation on the impact registration would
> have on vandalism. What happened to testing our
> hypotheses?
Good question, which is why I have agreed to such tests being run at Wikia even though I oppose the idea of blocking unregistered users. The Wikia sites on Pushing Daisies and Flash Gordon disallow anonymous edits and are being compared with a number of other TV wikis which don’t have that restriction.
I don’t know if anyone has done any decent research on the effect of barring anon page creation at Wikipedia, but the data is certainly out there if anyone wants to look into it.
> The argument that requiring logins may
> discourage positive contributors who find
> it too burdensome to take a few seconds
> to register is hard to fathom in light of the
> stupefying complexity of Wikipedia’s policies,
> guidelines, and procedures for editing.
Interesting point. Has the English Wikipedia become too hard for casual editors? If so, the effects of forcing registration might be very different to doing the same on a smaller wiki.