Well the final and long-awaited part of On Wikipedia's series "Who's On Wikipedia" is out (written, unlike the first 5 parts by yours truly). In it we present the results of a survey of BLP subjects, and provide what we think is one of the most common sense solutions to the BLP problem. Please do take a look.

Well, at least we know how WP could improve without pissing off the public. And amazing thing is that the pubic is not more pissed.

The idea of a "contact us if you're the BLP subject and you don't like it" tab, was floated by yours truly, here on WR, some time ago. I'm glad you thought of it independently, Limey. It is an obvious "fix." Provided the person on the other end doesn't screw it up by saying "prove it." Which, of course, they will.

The second recommendation is good as far as it goes, but trying to implement it is only going to expose the BLP problem's root: there are far more BLPs than Wikipedia's intelligent, ethical, and process-aware editors can maintain. It would indeed be a very good thing if all BLP subjects were aware of their articles and had the contact information of one of the aforementioned intelligent, ethical, etc. editors with whom they could raise any problems, but I don't think implementing that project-wide is viable for the same reason that individually checking every one of Wikipedia's BLPs isn't viable: there are too damned many.

For this reason, the first recommendation seems more helpful. In point of fact, something like that recommendation already exists: the BLP talk page banner includes a note reading "If you are connected to the subject of this article and need help with issues related to it, please see this page." The problems are that the notice is somewhere that most subjects aren't going to notice it, and that the page is unhelpful in the sense of being long, convoluted, and, incredibly, does not actually contain the OTRS e-mail address (it does contain a link to this page, which does mention the e-mail address, at the very bottom).

Also Known As — We Who Are About To Be Banned (If Not Already)

I'll leave it to Peter to work up the Latin …

Jon

Won't work. You underestimate the extent that Wkipedia is a revenge engine. Engaging the Wikipedians with ther own internal process will only result in 15 year old's telling the already aggrieved BLP subject that they "deserve"coverage, like it or not. If they persist they might even be pursued off wiki in ther jobs, etc. As always you got to separate the editorial decision away from unilateral Wikipedian control or it is just more wack-a-mole problem moving. This could be done with independent outside dispute resolution made available to concerned BLP subjects. Otherwise people will regret even trying to engage the Wikipedians "constructively."

Sounds like a great scavenger hunt!

Nevermind-- all projects left to be done by programmers alone, come out looking like that. Nested little puzzles with easter egg surprises far down the tree. It's how THEY think.

Engineers, too, sometimes.

===============

Where's the OFF switch??

Down on the lower far back corner, under the flange. Reach back and you can feel it.

WHY?

So you don't turn it off accidently.

But WHAT IF I *&%ING WANT TO TURN IT OFF ON PURPOSE?!

Well, you should have read the manual. It's here on page 34, section on Normal Operations, first paragraph on Deactivation, right there.

The treatment of BLP subjects who complain to OTRS is, in my experience, a couple orders of magnitude better than those who don't. Making it easier for aggrieved subjects to contact OTRS won't resolve the systemic problems, obviously, but I think it will do a lot of good in individual cases.

Plus, who knows - if it increases BLP complaints to OTRS by a large enough volume, Somebody might do something about some of the systemic problems, too - that's probably just mindless optimism rearing its head, though.

Great minds think alike indeed.




I generally respect you, but I must say that this is absolutely the sort of rhetoric that leads to absolutely nothing getting done on Wikipedia ever. "There are too damned many" BLPs, so don't try to fix any of them? Contacting even just one BLP subject makes the problem better for everyone. Contact 10,000, and you've started to do something significant. Sadly, the rhetoric of Wikipedia often seems to go: "Oh that solution's not perfect, therefore we will not attempt it" even when the status quo is completely intolerable. Wikipedia needs to do something, and mounting a grand campaign like the one I have suggested will do a lot of good, even if it eventually runs out of steam.

Perhaps, you were going more in the direction of "there are too damned many" BLPs so we have to delete a sizable chunk of them. If so, I won't disagree with you, but I think we all know that mass deletion of BLPs won't happen any time soon. I think the real beauty of contacting all the BLP subjects is that it's a solution even a radical inclusionist could love.

That interpretation is the correct one. I'm all for contacting as many BLP subjects as is feasible, I'm just saying that as long as the number of BLPs remains where it is, "as many BLP subjects as is feasible" < "a significant proportion of BLP subjects".

Let me illustrate what I mean: if somebody wants to design some kind of tracking system so we know who's already been contacted, I will contact 100 BLP subjects. It will take me quite a while, but I'm a student with a fair amount of free time and god knows I'm unhesitant to waste some of it on Wikipedia. Now all you need is, what, 8,000 more like me? You can say that if Wikipedia has the manpower to *write* all those BLPs, it should have the manpower to contact its subjects, and you'd be right, but "should" doesn't really enter into things, as a practical consideration.

Incidentally, back when I used to create BLPs, I always used to contact the subject, so I agree with your general thrust; it's just that it amounts to treatment of symptoms.

You'd be surprised at how quickly this goes. Generally, you find that very quickly (on a personal webpage, university website, business site, etc.) or not at all. Looking back at my email outbox it took me about 5-7 minutes for each contact I made (I didn't formally track how many people I couldn't find contact info for, though, but I'd guess I found information for about 1 in 4). So, back of the envelope. Take the 460,000 BLPs on Wikipedia. For the sake of prioritizing throw out all of them that don't come from English-speaking countries (if the drive goes well, you can go back and hit those). You're left with 230,000 BLPs. You'll be able to get in touch with about 50,000 of them. On average, that'll take you six minutes each = 5000 man hours of work. Out of those, you'll get responses from 60% = 30,000 responses. Now, it's harder to say how much time it takes to deal with the responses, so we'll leave that out for now and bring it back in later.

So, the 5000 hours needed to contact the subjects. There's no reason that it has to all happen immediately, so let's say you plan to space it out over 30 days. Now you need 167 man hours per day, which is essentially nothing. That's like asking each active admin to give you 15 minutes a day for a month (or we'll say 18, 3 contacts per day each). What about dealing with all the responses, the real answer is I don't know, but let's just imagine that 10 minutes is reasonable. That then comes out to another 5,000 man hours. So, we're asking 800 Wikipedians for half an hour a day for a month. Is that really so hard?

Well, as a statistics buff I'm glad you discuss the limitations of your sampling methods and possible bias issues. I'd be interested in knowing what the breakdown of careers/occupations were.

On your proposal to email all the BLP subjects; difficult to do, especially given the numbers, but not impossible and it would certainly be helpful to people less likely to know they have an article. Personally, I haven't had much luck with contacting BLP subjects. The only BLPs I edit and maintain are for media industry professionals. I've sent multiple notes through emails and PMs where possible, but I've only ever gotten a response from one, who basically said "technically the source you quoted is incorrect, as I don't have that job title. Otherwise it was "remarkably accurate". And then something about a simple man trying to make his way in the galaxy. I'm frankly surprised you got such a high response rate. What email address were you using? I don't think my Gmail and university accounts hold much sway

The first obstacle is finding 800 Wikipedians you trust to interact with article subjects upset with their articles.

Anyway, I note that you've gone from "every BLP subject" to "the approximately 50,000 BLP subjects in English-speaking countries whose contact information is fairly readily available" (I'm not accusing you of changing your position, just pointing out that your clarification of it changes matters). That does make things somewhat easier, but I think you're overestimating most Wikipedians' willingness to devote time to things that don't serve their pet interests, whatever those happen to be.

Well, I think we must acknowledge that it's basically impossible to contact every single BLP subject (if nothing else, a certain percentage of them are almost certainly imagined). I'm not suggesting that we be strictly limited to 50,000 BLP subjects, but rather that the project start with this low-hanging fruit. If it's successful, then continue it to the more difficult ones. If it proves impossible even to contact these folks, then there's nothing to do but let it drop.

I'm open to the idea that I'm "overestimating most Wikipedians' willingness to devote time to things that don't serve their pet interests" but I'm not sure it's true. There are a great number of Wikipedians who spend a great amount of time at AfD, CSD, etc. I have trouble believing any of them are genuinely interested in those areas as such.

Interesting to hear how your experience has differed. As per your request here is the rough breakdown by occupation. University Faculty: 3 (+2 from other categories who happen to serve as adjunct faculty members), 1 Musician, 3 Novelists, 1 Government Official, 1 Actress, 1 Actor/Comedian, 1 Elected Politician, 1 Lawyer, 3 Businessmen.

As long as you're providing additional data, any chance you'd summarize things in a table of some kind? One column for the type of person (as above) and one for each of your questions? I'd be interested in seeing correlations between them (as statistically unsound as they may be). Alternatively, would you report on any interesting correlations?

We made certain promises to the people we talked with about confidentiality, so I don't want to put forward anything that might let you identify someone individually, but I don't really see a problem with this. As for correlations, I really think that anything based on occupation would be ridiculous (since most of those are N=1). So far as correlations go, we were wondering about a correlation between familiarity with Wikipedia and opinions, but found nothing in this regard.

We did find a few things, but again I caution about small sample size. In general, people who did not previously know about their biographies are less likely to find them "fair and accurate". Really, though, I'm loathe to provide such speculative information.

We are still planning to complete the full study, and once we've compiled a larger sample, Fact Man will be writing something up that probes the correlations and such. It's also our plan to eventually launch a website that will sit alongside the On Wikipedia blog. We'll use the website to do things like host our raw data sets, but I can't promise you that will happen any time soon as we both have other priorities right now.

That's a more varied and less boring result than my own BLP cross-section sample, which represents all persons born on exactly the same day (except those where data was missing or used a non-standard date format):

http://wikipediareview.com/index.php?act=findpost&pid=196469

Hey, one of them is a trick cyclist.

I did not attempt to contact any of them. If somebody does try doing so please make note of the response rate in that thread.


However I think if this is going to work WP would need some centralized way to track progress.

Repeated correspondence to one subject eventually will annoy the hell out him or her: "So if I say 'delete it', will you fuckers stop calling me?"

It also would waste what time could be spent contacting other subjects sooner. On the other hand we do not want participants to skip some names because they are lazy or careless, or lie about the results to further some agenda, or exaggerate individual progress while satisfying their new-found coldcallcountitis (?).

Some redundancy would provide a reasonable safe-guard against incidental failure, but I don't know what the best middle-ground would be.

I tend to think that tracking everyone so that contact is made only once is the best way to go. As pointed out, the workload would be significant, so it wouldn't make sense to duplicate until the whole list was done. Yes, some people will probably lie and exaggerate, but there are ways of dealing with it. If I was organizing things, I would set up a system whereby all emails involved are CCed to a central account for record-keeping purposes and then put into a database to which certain trusted users have access. (In general I think emailing people is a much better idea than having hordes of Wikipedians try to get in touch with BLP subjects by phone).

Now, it's quite possible that some users would try to claim that they had sent out more emails than they had, but assuming the central account is always CCed, that would require a more involved act of duplicity than just saying "Oh yeah ... I emailed 60 people this morning". Hopefully, the central database system would also check fraud about the nature of responses; there might well be some people who for whatever reason would want to misrepresent that content. This is not to deny that there would be implementation difficulties in a massive BLP-contact drive, but they're not particularly large obstacles.

Oh dear. The trouble is that you contact them once when their BLP says they are a nice chap, then the following week, someone suggests that they have a massive cock, as I believe the saying goes, or some other minor edit, and all of a sudden your signoff is invalid.

In the real world you would not have this issue. You would privately evolve the changes for a new edition and once a year, if there were something controversial and/or inappropriate that needed checking, the editor would hopefully ensure that the sources were super-reliable or that a representative of the subjects interests could confirm the details.

Also there is the problem that they are unlikely to realise in signing off their biog there, they have unwittingly contributed to an article that can be taken by someone else and grossly distorted unless you are very clear on educating subjects to the vagaries of CC licensing.

A new joke every day here.

Yes, that is the real devil in any plan involving BLPs other than deleting the lot. Of course, it would be best to implement this such that after the subject is contacted the BLP is semiprotected. One of the most common objections to liberal semiprotection is "But how do we know we're protecting the right version? What if an IP actually comes along and fixes something?". The email approach serves to ensure that you are semiprotecting the right version.

Really, though, it's my hope that by making subjects aware of their bios and giving them someone to get in touch with, those subjects will assume part of the monitoring role. If they notice something amiss, they'll know how to deal with it, unlike now where most BLP victims are either ignorant of what's happening or unable to do anything about it. This is certainly an improvement over the current state of affairs.

Get a clue, Limey, all discussions of reforming Wikipedia are a waste of time and electrons.

Jon

This is where flagged revisions comes in. It’s really not all that much different from semi-protection, except the requirements for “registration” (meaning editing the publicly viewable wiki) are far more stringent, and there are two copies of every article, a publicly viewable one and a draft. IP viewers can see only the public one. Nameusers can see that AND the draft, and can edit the draft, but can’t replace the public one with it. Only “reviewers” can do that, and they have to have 500 edits and 6 months of experience or something, so the penalty is VERY high for them to make a vandalistic edit and promote it to public “certified” status. Vandalize once and you lose your reviewer status and have to go though the whole thing again, and you’ll get old doing that.

Now, the BLP problem is simple. All of them simply get set to “draft” status. They disappear from IP view. They can be seen by nameusers, but are marked “draft” so any vulgarity or bad info in them has a label on it. Gradually over time, important BLPs get promoted and approved by nameusers, and then that version can be seen by anybody. The really crappy BLPs stay behind and perhaps many never do make it out of “draft”. So what? Problem solved. So simple, it will never happen.

Now we come to a an observation that I’m not going to bother with on WP, because it’s harder than general relativity, and will warp the brains of all who view it there. And only cause moaning. And do no good. But WR readers will have no problem with it.

Observation #1 is that the Germans, being German, simply instituted all this on all Wikis. I THINK they just make them all draft, and started working to promote the clean ones. This idea freaks en.wiki out, so they’re proposing to do it one-by-one, which of course will mean the default is crap. Typical of WP. BUT the German bulk-solution is great for en.wiki’s pressing BLP problem. Simply flag only the 14% of article that are BLPs. Voila, problem solved. Now you work to promote the 7% of THOSE that are paper-famous, as fast as you can, but it’s a much less serious problem. Everything else disappears from the public, and is marked “draft” even for WP nameusers. But since these are not paper famous, who will notice?

Observation #2 (Really hard) is that sprotection-as-we-know-it and flagged versions are not a binary thing, but two ends of a dial which can be set anyplace, and can be set individually, as is helpful for each Wiki. Sprotection is 10 edits, 4 days. The German flag is 500 edits, 6 months. But it can be anything for any article. What it is, can vary by history of vandalism.

We can set the a priori promotion flag, or bar, very high for BLPs and other classes of heavily vandalized stuff, such as things K-12 kids are forced to study. Also political subjects like abortion and Israel and the presidents. But for advanced subjects that only people in college are likely to see, less. And for some classes of articles, like pop culture junk, we can leave it IP-edit allowed. If it’s really true that the masses need a large sandbox to make farting noises in, and complain about farting noises in, this can be half the en.wiki, if you like. They can vandalize and revert vandalism, to their hearts’ content. Everyone needs a hobby.

How do you know what level of protection is needed? You don’t. After you finish flagging all BLP high, you just start low and work up for the rest of the academic/paper encyclopedic articles which NEED to remain free of vandalism for WP's public reputation. Every time an article is vandalized, you double its protection. You do this till vandalism stops.

The approximate ratchet levels are:

10 edits 4 days
20 edits 10 days
30 edits 20 days
60 edits 45 days
125 edits 90 days
250 edits 3 months
500 edits 6 months

These aren’t quite double each previous one, but I’ve jiggered them a little to make it come out even at “wp sprotection” at LOW and “German flag level” at HIGH. But whatever. As I noted, large patches of wp like Star Trek or Pokemon can be left IP editable, if you like.

Remember, if this eventually happens, Milton Roe thought of it, not Eric Moeller. I was promoting a “draft/public” article system here long ago, before the Germans did it. I suppose it’s obvious, but not when you get to the details.

Yeah.

When Limey did a back-of-the-envelope calculation concluding he would need "only" 800 users with a half hour per day, it should be obvious that it won't work. Volunteers do the kinds of work they want to do; nobody pays them to do necessary work, that's part of the definition. People who are on the site to build Simpsons episodes are unlikely to plow into confirming information for Croatian soccer players.

At any rate, I think most subjects are aware of their entries (the subjects I've talked to certainly are). Awareness isn't the problem. The problem is that it's completely non-obvious who they can complain to. The ORTS email is surprisingly well-hidden.

You may not use "unhesitant" until you can demonstrate it's a valid word. Wiktionary doesn't count. (Bonus points if you can make a reasonable argument for not using "willing.")


Not particularly surprising. It's entirely intentional, after all.

"[Unhesitant] is available in our premium Merriam-Webster Unabridged Dictionary.""Unhesitant" more precisely conveys my intended meaning; "willing" is a much broader word.

I think they just want your money.

I logged into the Oxford English Dictionary to see if it contained a citation for "unhesitant" (because I clearly have some free time at the moment). "Unhesitatingly" and "unhesitating" are the only two related entries. And, for what it's worth, Firefox's dictionary knows which word of the three to underline in red.

If you're into litotes you get double points if you can use "not unhesitant" in a sentence, and not have anybody go:

I did consider that, and so ran this test.

Homer: The food was...not undelicious.
Lisa (transcribing and editing simultaneously):"The food was delicious."
Homer: Brilliant!

If only this is what was going to be done. Even once "flagged revisions" are turned on Wikipedia (and that's becoming an if not a when), the implementation isn't going to look at all like the reasonable version of things.

Yep. The trouble is that flagged revisions are seen as a problem, not a solution and there will be a major effort to make Wikipedia work like it did before - people complaining one minute after an edit that it is not visible on the public site.

As far as I can tell, there is no concept that flagged revisions means what a sensible implementation should mean - a place for editing away from public gaze (after all it is established that editors make a tiny proportion of users of Wikipedia) and when a plausible article is produced it can be released.

As such a sane approach doesn't have the "Look! I did this!" factor, one could see that casual editors would soon walk away. This might be a good thing, but then we'd be left with Slim, Tony WhatsMyNameToday and David Gerrard to it seems unlikely.

On Wikipedia is now linked to from the Strategy Wiki's Task force/Living People thanks to NW.

We'll have to wait and see if the Strategy Wiki task force eventually peters out like the one on Meta did.