I was an administrator at Wikipedia from 2003 to 2008. I kept out of a lot of the politics, but I was in some of the politics of the site. I was neither beloved nor on anyone's most wanted poster. I ruined a few folks nights, I'm sure, with my arguments, and plenty of people disagreed with my "elitist" position. All of that, though, is irrelevant to this other skill I had while I was there: I was fairly good at looking at the system as a system and looking at the psychology of the system as a series of weak points. I was disinterested (in a literal sense; I could be distant).
Now, I do know of a way that Google or some other wealthy organization could replace Wikipedia within six months, if it wanted, but this is because of the structure of Wikipedia, not the content of Wikipedia. The arguments about content are trivial and trite, for the most part, when they're not symptomatic of other arguments (like whether Paul Revere had a six shooter).
In user-generated sites, when there is anything like a flat or layered hierarchy, the discussion we praise is also the argument we hate. This, my children and droogies, is Democracy as the eighteenth century warned about. When any argument erupts, speakers note the trees, not the forest, because forests are really hard to manage, and trees are easy to attack.
What I offer below is not meant to be a critique of Wikipedia, attack on anyone or anything, but merely a set of observations. If they are useful, then let them be used. If they do not apply, then so it may be. However, I believe that the lessons and observations show us things about the process of user-generated and blog reliant sites rather than merely what happened in my five years of Wikipedia.
I organize my observations into "vandal patrol" (Frank Zappa's "Help, I'm a cop!"), "consensus" decisions (Radiohead), administrator ranks (Alice Cooper's "I Wanna Be Elected"), outsider powers -- real and imagined ("They're Coming to Take Me Away"), and the non-transparent rivers of power ("PM"). The last section is about the departures.
This is either entirely "meta" or entirely an analysis of that online encyclopedia, so read on only forewarned.
Oh, and this is going to be long.
Overview: I imagine that everyone knows, at least roughly, what Wikipedia is and how it works. I heard about it on WNYC and a little story about the Quixotic attempt at creating an online encyclopedia with volunteers. Since I was working in technology at the time and was bursting with my humanities knowledge having no outlet, I began to tinker with the project. I then moved, became a librarian, and found that I had even more joy at assembling materials and writing articles. I was teaching, at that time, in an inner city public high school. It boasted, properly, of 94% college bound graduates and 90% African American, inner city population. I thought to myself, "You know what? I want to make sure that when a kid like this Googles some allusion she or he trips over in a book (example: kid asked me about "malleus mallefactorum" because of fantasy novels) or wants to learn about some obscure thing, I want there to be something that isn't for sale or a downloaded term paper. I also felt that the project needed to be an encyclopedia, which means a thesis statement, an interpretation, a guide to research, and an exclusion of material that is extraneous. I did not see the advantage of writing up articles on favorite hotdog stands, nor was I, as an academic, afraid of authorial point of view, because I had had years and years of training in the matter. Over time, all this would come to the fore.
Wikipedia became a success. As it succeeded, it began to hold growth as a goal. (Why, one might ask.) It accumulated vast numbers of accounts. Its use of the GFDL meant that it was mirrored all over the place, so virtually everyone uses Wikipedia, whether he wishes to use Wikipedia or not. That made Wikipedia a master target.
I argued that the absence or presence of an article on one's business could not infer reality on it, and any company, band, or art using Wikipedia to advertise has already failed as a project. Nevertheless, long before the Siegenthaller affair, people began to fight reality through accounts of reality. Again, it was like advertising a non-existent business: they would write reality as it should have been to make it what they wanted it to be. (Stephen Colbert's "truthiness" is a perfect encapsulation of this mentality.) In the 2004 U.S. presidential campaign, "BushCountry.org" advertised, we heard, a link to the Wikipedia article on John Kerry that said that the medals were unwarranted, that he was a traitor, and a cry baby, etc. They did this by having a link to an outdated page, which is easy to do. Back then, the warning banners I recall were in a normal font. Long, long before Siegenthaller, then, there had been warnings. The English language Wikipedia always had polite people with good English skills editing and crafting articles on pieces of European history, too, and European geography. That these articles were part of nationalist causes would not occur to the average teenaged Wikipedia editor or administrator, and, if it did occur, then he would tend to like the "nice one."
Wikipedia was and still is overrun by Stormfront trolls very early on. I have confidence that many are still there. However, American editors could at least sometimes find them. The Polish/Russian, Estonian/Russian, Lithuanian/Russian, Polish/German edit wars also went on. The crypto-zoology, UFOlogy, conspiracy, and "alternate science" edit wars went on as well. Each side had two elements of argument. One was the belief of a zealot that saying it on Wikipedia made it so, and the other was that professionals were realizing that a view in Wikipedia would propagate and could be linked to. Indeed, after propagation, it is possible for self-published "research" to act as the basis of research in enough time.
Enough overview; I should get to practical hints.
"Help, I'm a cop!"
I put this first, because it was the first phenomenon to show up. One reason I have cool quotations at my beck is that I looked them up a long time ago under other circumstances. Thus, I love to cite this passage from Alexander Pope's Essay on Criticism:
"All seems infected that the infected spy
As all seems yellow to the jaundic'd eye."
To put it another way, everything looks like a nail to the man with a hammer, and every man looks like a pervert to a stripper, every citizen looks like a criminal to a cop. No matter the project, any successful project will have detractors, distractors, and attackers.
A site devoted to left wing politics will attract right wingers who want to come and scream at liberals. A site devoted to women will get chauvinists and misogynists. A site devoted to gay rights will attract religious fundamentalists. This is the classic definition of a "troll." (Actually, there are two definitions of 'troll' deriving from two usages. One is a corruption of "trawling," and it refers to people desperately trying to get conversation. The other is a person seeking out what he, usually, hates to scream anger or work frustration, from "troll doll.")
Additionally, all sites with any traffic will get people who want to turn the site's attention to how the site is run. There are people whose purpose seems to be to go to sites and ask how that site ensures complete free speech or gun ownership or something. If not that, then these people will want to bring up by-laws and procedures and question each one and demand polls on them. If not that, there will be a population that says, "Yes, what we're doing here at this encyclopedia is great, but why aren't we giving a guide to the Pokemon cards? There's no reason we can't. After all, we have pages for every stripper!"
Finally, there will be explicit and implicit attacks. The explicit attacks from vandalism, DoS attacks, TOR accounts, and all the rest are incessant and sometimes professional (mostly 'script kiddies'). The implicit attacks are usually professional and only sometimes amateur, and they will be the people writing in, "I'm a lifelong liberal, but I'm disappointed in Obama because he won't cut the capital gains tax." In other words, stalking horses, people who change only a line in an article, people who put in only one little external link to a hate, gambling, Trojan, or porn site, and the like.
Thus, some group of people will need to protect. In user-generated content sites, this means user-generated protection, and that means people whose raison d'script is to find vandals. The business of the site, whether it's getting progressives elected or stopping pollution or writing articles, ceases to exist for these people, because the entire project becomes a hellscape of criminals trying to steal into the fortress, with only them as the noble guardians on the wall. Very, very, very soon, they begin quoting Jack Nicholson and telling everyone else that they "can't handle the truth."
The moment people make vandal protection their jobs or count these efforts as their credentials, things have gone wrong, very wrong. The person is offering up skills that are unrelated, in essence, to the site. They are not user content, are not to the purpose of the site, and are socially corrosive, as they foster a world that must be populated by enemies. A vandal hunter will never accept "all is well" as a verdict. The job may be necessary, but it is necessary as a function of site maintenance, not site function. The function is the cause people assemble around.
Fake Plastic Trees
"Fake Plastic Trees" may not be an obvious choice for talking about the problems of computer mediated communication and consensus. My satirical itch made me do it (consensus arrives in mass production; when people are present, there is no such beast).
I want readers. They make me exist. After all, online there is no reality, and there is no online life in reality, so I "am" only as I am read. I have to type. I have to talk. If, therefore, there is a topic, I want to weigh in on it. If I feel like saying, "Me, too," that's no good, so I'll say "Me too" with a compliment or an insult. The point is, I have to be heard.
This is why, when there are debates with user voters, there is never resolution. Everyone must speak. Each individual opinion has equal weight, theoretically, and persuasion only works if people have good will and open minds. Abridge either of those, and reason and persuasion will not work. At Wikipedia, the call for a king was instantaneous and has never stopped ringing. Even though they got King Log, they needed something above the flat hierarchy simply because of what the flat hierarchy did in the absence of persuasion. (As for why he's King Log, I'm not going into that. If you don't know Jimbo Wales's foibles, then peace to you. If you do, then you don't need me.)
Furthermore, either because of the perverse desire to be heard or just because of how people read and respond to complexity, general issues tend to be discussed in terms of the most objectionable clause, adjective, or phrase.
"Geogre: We should not put in an article on every hot dog stand and traffic light in America. We have to have some notability. I'm sure that Woody Allen stood there once, but that's not notability."
"Objector: Woody Allen? When I say that Hemingway drank at this bar, it's not the same thing as Woody Allen. His movies have fewer viewers than Hemingway's novels any day."
"Obfuscator: Let's look at those sales! I don't think so!"
And away it goes.
The first or last thing said may be the only thing read, and when "everyone" speaks, the discussions grow redundant and infinite and interminable. However, when there is a requirement for consensus over majority, there is a recipe for stasis or a vaccuum that will demand power.
When policy comes up, it's better to do what we do in reality. Get people into working groups to present positions (eliminating the repetition), and then have votes. Otherwise, people want to be heard.
What Wikipedia did/does instead is rely upon increasingly opaque structures of power and decreasingly rational narrowing of the communications medium via IRC channels with fewer members to cut through discussion. That, however, turns the front of the site into an obvious onanistic exercise for "little people" and increases the spur of each to climb.
I Wanna Be Elected
Wikipedia has self-selection. Because it wanted to be flat in its hierarchy, it originally required established users and administrators to sponsor new administrators and then for there to be votes, but that fell away very quickly. The rule quickly became that people put themselves forward. For the really powerful positions, the arbitrators, the board, and the various projects, it is purely self-nomination.
Therefore, instead of anyone being noticed for good work, people prepare a case of their good work. They do this with portfolios of edit numbers (which can be gotten by vandal fighting, by creating -bots that perform 20,000 or 30,000 edits in an evening, or by changing one of the categories). Where once my 200 fully written articles would have been astonishing, for people who port public domain dictionaries directly into Wikipedia, they could say that they had "written" 15,000 articles in a day -- each archaic and inelegant. Others would be certain to be a happy, nice, well known editor. They would place "prizes" on well known, high visibility users' talk pages, jump into policy discussions out of the blue on the side of a powerful administrator, etc.
The only administrators who should be working at user-generated content sites should be the tribe of Cincinnatus. As with American democracy, the story of Cincinnatus is central. The reason is simple: those attackers, distractors, and detractors, and especially the professionals, have no problem becoming administrators. Additionally, people with a strong desire to be important are dangerous to others. People who want to be powerful online are sometimes exorcising demons of their personal lives, and this usually ends in tears.
Any time self-nomination is involved, deception is involved. It might be minimal, but it's present.
They're Coming to Take Me Away
The real forces on the outside of Wikipedia and the imaginary forces on the outside of Wikipedia working to ruin it were always in a tango in the popular imagination and filled up half of the Administrators Notice Board on most days. I will make two quick divisions separate from the attempts to subvert Wikipedia by professional and national organizations.
The 'I hate you' wiki
For mysterious reasons (see below, or above), people would decide that they wanted to start their own wiki, and it would be better than Wikipedia, and it would make fun of Wikipedia, and it would tell all of Wikipedia's secrets. Former and current Wikipedia administrators, users, etc., were in some of them. To my knowledge, the only people in the world who read or visited those projects were Wikipedia contributors. The world yawned, but the febrile and swaddled online world of Wikipedia society went/goes/will explode into Committee of Public Safety mode. Each user is suspect, each suspect is suspicious, etc. The "serious" "I hate you wiki" projects came from like-minded ideologies with money or numbers, and these, amusingly, never generated the hunt for counter-revolutionaries at Wikipedia. Thus, the ones that actually have visitors and contributors are the Conservapedia and the wiki's begun as rebel/cast off/exiles of Wikipedia around particular television shows.
Help, help! I'm being repressed!
The other internal division was the never-ending and constantly shifting segment that was having its "first amendment rights" denied. From people upset that an article on a favorite band got deleted to people upset that an edit on Bigfoot didn't go the way 'latest research' confirms, a segment was always suspicious of conspiracy within the project and always ready to make common cause and conspiracy. The problem here is that it was easy for people to dismiss the complaints and to miss genuine takeovers. This is where the most damage has been done.
Because nuts are plentiful in the dish, people would miss, often, the meats. In fact, haughty and juvenile administrators would sometimes fail to investigate at all and thus create vandals out of users.
The fifth column
Everyone outside of Wikipedia knows that press offices for small nations have people in charge of watching the Wikipedia page. We know that companies monitor and "clean up" their Wikipedia pages. We know that political parties have people to spare for observing and engaging in edits to pages. Given the power of an edit to Wikipedia, it's economical (cost: zero; mirrors: hundreds; speed of propagation: 24 hours). Informed people know that intelligence agencies around the world also work on Wikipedia entries for the same reason. People who can read the newspaper should know that private "security" firms and "reputation defenders" promise to infiltrate and control Wikipedia pages.
In short: Wikipedia is a beanbag chair being shot by a thousand machine guns.
Now, given the self-selection there, and given the fact that a cube warrior is getting paid to get in, and given the fact that an I.O. has studied the organization already, what do you think the odds are that an information agent could become an administrator? Would you be shocked or absolutely aghast to learn that administrators have turned out to be not what they seemed at all?
Yeah, me too.
How about if "arbitrators" who were "personally chosen" by Jimmy Wales turned out to be a bit less than they had appeared?
Forget, if you can, that, and imagine this instead: a Lithuanian contributor is going to give an account of The Second Battle of Khotyn (1673), where Lithuanian and Poles defeated the Ottoman forces, and his English isn't very fluent. Someone, though, is changing the article to say that Lithuanian forces "played a minor role, subordinate to the Poles." Well, that's not true! So an edit battle erupts. The Lithuanian gets frustrated, because there are five editors whose names seem to be Polish who keep trying to erase facts, and he's just trying to say something objectively true. He calls them like he sees them: he feels sure that they're part of a Polish nationalist group that's infamous on the Internet in the area. He's banned. The article goes to the "consensus" form. This is because the sputtering, stuttering, Lithuanian was "not civil," while the organized presence of interests is quite civil, quite polite, quite numerous, and quite professional. This is how Wikipedia can spread propaganda.
So long as administrators and arbitrators are self-selected and selecting on the basis of "I want it," and so long as there are no credentials or expertise, the only barrier to a professional information officer is "civility" and numbers and citation. (David Barton should make obvious to everyone what academics already knew: citation does not mean truth. It only means publication.) A twenty year old American will not be able to tell true from false in European seventeenth century history and so will default to "who's obeying the rules."
PM
In my own estimation, the end of Wikipedia came with the use of IRC channels. The moment there was an "3leets" group or a place where decisions were made that were not recorded or portable over to more open media, the underlying principle of the project was gone. More to the point, though, it exacerbated every other problem.
The "trees instead of the forest" and "consensus" problem are worsened when people are chatting socially and forming friendships on IRC channels. The "vandal hunter Van Helsing" problem is worsened by the fact that an edit can be spotted, investigated, convicted, and banned in seconds, without any attempt at communication with the suspect. Worse, that went on even dealing with administrative issues. Administrators were dismissed, users banned for non-vandalism, and the like with no communication, without asking the person, without hearing or lending credence to a person's words because of the vandal hunting mentality.
Two things deserve special attention, though. The presence of three or four separate IRC channels for Wikipedians (one for mechanics, one for "administrators only, and trusted others" (by whom?), one for vandalism, and another I think), run by unelected individuals with rules set not by or on Wikipedia threw napalm on every conspiracy. The advantage to Wikipedia of these channels was zero, in my view, but the negative was gigantic. Every person who thought there might be a conspiracy against him or her now had the mechanism for one. Furthermore, though, the IRC channels gave the real information officers a perfect way of controlling the encyclopedia and amplifying their presence. Our Natasha Fatale only needs to be chatty and charming and interesting and "1keweldoood" on a chat session 12 hours a day, and she can control the fate of half a dozen articles. Votes on policy matters would appear by the dozen within seconds of each other, and this would correspond to someone going on IRC to wave a flag.
More importantly... much more importantly, the presence of the socializing channels and the socializing made the administrator promotion ridiculous. Is it possible to be an administrator of Wikipedia now without actually writing anything? Yes. Go to the IRC channels and be a friend. Be a pal. Be funny. Agree with people. Compliment people. Ask for help like a rube at a poker game. Then do something that will get "edits" without doing anything. Voila! You're an administrator, and you can delete articles.
Once this social side was in place, it was impossible to reform, because it would require a consensus to remove (fake plastic trees, or, of course, the power group above consensus, but, since that group used IRC to form...). Socializing is fun, as we know. Kos is a community, but it sets out to be. This is the "community site." Socializing and meeting people, especially for the people who go to Wikipedia (the psychological profile is specialized) regularly, is addictive, and so people could hear and understand my argument about where things were going wrong, but they just... couldn't... agree... because... oh!
Concussion
Wikipedia is "not reliable" I hear. That's right.
I know that what I wrote was, but that doesn't mean it is. One of the reasons people leave and left was the imperiousness of socially promoted people trying to tell researchers what things mean. For me, I had to endure idiots telling me about "inline cites." I used parenthetical citation. I feel that citations that don't rely on computer code are more robust, and I also feel that footnotes are far easier to fudge than citations at the end of the line. I also think that parenthetical citations make it clear when someone is effectively paraphrasing a source. However, the people who are programmers and want "edits" with -bots wanted to be able to run programs that would be able to do this or that with tags, and so they organized to make a rule that a work had no citations if they weren't done with a particular footnote code. Thus, it's kind of irritating to know that I wrote the article on Orrmulum there, and I had a citation at the end of every flippin' line that needed one, and then some spiteful jackal removed them and slapped a "citations needed" tag on it.
The poetry blog editor for The Guardian left after IRC society proved more powerful than the site's rules. Oh, and his articles had full citation, too, but now they have magically been stripped of their citations, too. (Wrong coding, you know.) He, too, had to be told patiently that he didn't know anything about anything.
The good authors I know who left, left because of people, not rules, and not the vandals. We left because of the social side, every single time, and because of the little social clusters and talk pages of users and barnstars and smiley faces and elections conducted by chat room. For my part, I could simply write out of the joy of writing for a while, but when the continuing juridification and corruption of the place (self-pleasing), with no hope of having the junkie vote to cut off the heroin, accelerated, and there were insta-trials by private IRC, it was clear that some people wanted to be the boss, and that's not possible at a volunteer project.
If there is a single lesson, it is this: the aggregation of people does work. However, in a flat hierarchy, the first thing to go is the flatness, because in the mass there will always be an urge for power. There has to be a constitution before there can be democracy. We knew that once.