I'm less of a Wikipedian than just about anybody who contributes here. My only value is in the development of thought about Wikipedia as viewed from outside Wikipedia. Despite this I know more about Wikipedia and have engaged Wikipedia and it's denizens more than anybody I actually know. Not even close. I'm out there in far end of the normal distribution of...well normal people... in terms of experience with Wikipedia. Still, I sometimes watch the parade of things I don't care about in the discussion of personalities and dramas on Wikipedia with bemused detachment.

This has led to a minor insight into the nature of community dysfunction. The received knowledge concerning Wikipedia's community is that once there was a Golden Age in which a true consensus based community evaluated users based on the merit of contribution. This Golden Age was doomed because this utopian state of affairs does not "scale up" to the level now reached by the project.

This never happened. Wikipedia is actually a village of no more than a couple thousand participants on the level that that makes people believe they are part of "the community" Almost all of the millions or tens of millions of "user" Wikipedia have no ongoing relationship or investment with the project. For a "reputational economy" to function among a group pseudonyms a couple of things are required. First the membership must be stable over time. This is mostly not a problem on Wikipedia, although that might change. Second, the number of participants must achieve a level large enough that only the users record of contributions not alliances, vendettas, and ass-kissing matters. Such an economy would work only if each exchange between participants resulted in record of relevant information only. Information not relating to the goals of the project (eg about "social networking aspects" ) will act as a substitute medium of exchange. Like currency in any economy, the bad will drive out the good.

Bootleg social networking currency can only retain value as long as the project scale remains limited. In a project that truly has millions or tens of millions of participants repeated interaction between the same participants becomes too rare to hold value. The only currency that then matters is that relating to the project's manifest goals.

This needed scale has never been achieved on Wikipedia, which functions very much like a small town in which everyone is into everyone else's "beeswax" on a level that is just plain creepy. This might be because only a very small segment of the population is susceptible to the lure of Wikipedia. It may also be because the "Old Timers of the Golden Age" keep raising the bar to ration access to maintain their positions in the project. In either case the result is the same: Wikipedia is not the the bastion of consensus and merit it hoped to be, but instead it is a hopelessly dysfunctional community. The community that does exist is an extremely skewed sample of the general population that likely contains many systematic distortions and even an overabundance of social pathology.

Does anyone have an idea how to give a reasonable figure on that.

One way of getting a rough number of hardcore inner community members would be to count the voters in the elections. Which is 300-400 or so I think?

I had suggested the same a few days ago. About 750 people voted in the ArbCom elections last year. We'll know how many vote this year in a couple of weeks, but I'd guess it'll be about the same. I don't know how many voted in the Board election this year (which would span all projects), but the number is out there.

At the time of the WMF elections the number of users eligible to vote from en-wiki (eg, those with a high enough editcount to be considered "regulars") was around 10,000. As Kato says,beteen 300-500 bother to participate in the elections, so at a rough guesstimate that makes 95% of regulars who don't get sucked into the drama.

Take that with a huge pinch of salt, as "numbers voting" as a measure excludes the ANI regulars who don't bother to vote, blocked editors such as Peter D who nonetheless are (or were) active etc, while that 10,000 figure includes some users who rack up 500 minor edits to their userpage over a few days, then lose interest and never come back.

As a very unscientific experiment, looking down ANI or the Village Pump I recognise about 25% of the names.

So with your 750 hardcore members who voted, and Eve's 10,000 registered-to-vote in the WMF elections (lets say cut that down to 5000 as half of those will be for other language wikis and projects(?)) - we're getting down to the couple of thousand on the English WP Bead is talking about in his opening post?

This is fascinating, and it would be way cool to get some stats on this. I've long believed that a relative handful of editors are doing all the actual mainspace work (how many of the counted "regular editors" actually edit mainspace??). Also, our of 1,600+ admins, there are supposed to be about 500 or so actually "active" at any time (can't remember where I got that from). That leads to an interesting ratio, too.

With 2,641,436 articles on the English WP, and some 3000 active community members, that means that there are 850 articles to every "community member". Which somewhat explains the Seigenthaler Syndrome (meaning the many unwatched vandalized biographies), and the rest of the garbage out there.

Regarding mainspace, it would be easy for someone on-wiki to analyze how many "active editors" there are (defined as, say, more than XX edits in the last month). Indeed, I'm sure these statistics are already around someplace. The problem with this sort of metric is that creating a new high-quality page can be one edit and reverting 50 "poop" vandalisms is 50 edits, so it's not a straightforward task.

Beadgame's village analogy is apt though. With 2-3000 residents. When I first saw Wikipedia a few years ago, I thought the community would be like the genteel village of Painswick. Instead it turned out to be like Tombstone, Arizona (T-H-L-K-D), or that rancid village in the The Wicker Man (1973 film) (T-H-L-K-D).

A good indicator would be to calculate the average edit per person, and then search for about 20 editors who are closest to that average and who have edited the most article in that limited number of edit. With a condition that they have edited a wide area of subjects. (those using bots or revert vandalism will be the best candidates)

Then choosing randomnly a hundred article or so then see how many of those editors you have found in that selection. You then devide the total number of edits made by those editors in thos articles by the total number of edits. You know the total number of edits in namespace of each of those editors so the rest is basic estimation.

After finding that average you compare it with the number of person who vote in the elections.

If anyone has some spare time, do that.

Of the admins, 950 are active by a very lax definition (30+ edits in the past three months). The mainspace edits are very skewed since an editor can rack up 500 vandal-reverts a day on Twinkle or Huggle and they count the same in the editcount score-table as a days-of-research edit like this; plus there are some editors like Shankbone who don't have much of a mainspace history but are very active with images etc.

If someone posted a well-formed request for data on WikiEn-l, and explained in general terms how the data would be useful, I suspect that someone would go ahead and calculate them.

The point I want to emphasis about scale is that WP has not reached the point of critical mass needed (and probably never will) in which a reputation economy can function. A Metropolis is required for the "currency" to be truly fungible. In the small town a person of colour might not get served at a lunch counter or a person "known" as a drunk might not be able to buy whiskey. The motive may be evil or well intended but the result is a distortion and undermining the function of the economy.

An example of a website with a more successful reputational economy (but still very imperfect) would be eBay.

A steady editor on Wikipedia who focuses on a particular subject will develop a reputation in that corner of the project. For example, I would be well-known among administrators and (at least before I took my break this year) among Wikiproject Law and Wikiproject US Congress editors and on the Nero Wolfe pages, but if I edited a music or architecture article no one would know me. I can't think of anyone other than Jimbo Wales who would have a reputation permeating Wikipedia as a whole.

Wales has touched on this in his own inept manner a couple of times and so did Bauder. The idea is that every user has some kind of "cred" based on their contributions (and I would add, although they would deny, social influence and alliances.) This is somehow factored into the Wikipedian highly fluid notion of "consensus" and is why voting is so frowned upon. It is an ill defined and distorted form of a reputational economy. It is contained in a users contribs and every user has one. Being ill defined it is of course all the more subject to distortion and manipulation based on a user's influence and alliances. I believe this results in a kind of gangster thug black market economy.

I was with you until the last sentence. Trading off support on different issues is seriously frowned upon. That doesn't mean that meatpuppetry or horse-trading (overt or tacit) doesn't happen, but you haven't made a case that it's commonplace.

The selectivity effect does not explain everything. Notice boards, village pump are beter indicators, yet it's almost the same bunch of people. Only those who's contribution is more than article writting will be interested to vote, the other's won't, unless they are instructed to do so. (or unless there was an arbitration involving them)

Most users are not known because of their article writting, editors like Tim Vickers are not the norm.

The most relevent point, is why those who write good articles in a healty spot of Wikipedia with good cooperation will even want to deal with wikipolitics like voting, or being part of the 'group', for them all this is just a distruction and a total waste of time.

It would be hard to calculate; I can't think how you'd write a script to differentiate between the addition of a long block of original text to an article, and a five-second "undo" restoring a blanked section to an article. Any solution I can think of (aside from just dip-sampling) would involve comparing the "new" version to every recent previous version, and that would make the servers scream in protest.

I've thought about doing "real" edit counting based on edit complexity, using a local database copy populated from dumps and a fetch engine. It's still something I want to do someday, but I need more computational resources than I currently have at my ready disposal. Might still happen, but I have several other projects of higher priority ahead of that (like an APRS digipeater/gateway).

There are those whose reputations precede them.

But not so different from ordinary politics and fame in the age of electronics, no? Once upon a time fame and the buzz and "seeking the bubble reputation even in the canon's mouth," all had to be done by "hand," as it were. It was all person-to-person gossip and stories and sagas and such.

Now, we know more about our favorite celebs or Barack Obama, than we know about the guy who's lived for years three houses down the street from us.

We're all members of the same-sized communities, but now they are spread oddly over distance, now. Which would be fine if it weren't for all the "fake" connections--- people we feel we "know" because of electronified politics, who we really don't. And there is a lot of fiction clogging our social inputs; perhaps more than at any time in human history. Sure the Greeks all knew the semi-fictional characters of Homer, but how many of the same fictional characters do most of us know from a hundred TV shows? Thousands of names which replace the reps of thousands of real people, since the brain only has room for so many, and no more.

Sure, it's true Wikipedia doesn't scale well. But neither does human society once you get larger than the small village our brains are built to deal with. Beyond that, we get snookered by flackmen. I'm not sure it's that much worse on Wikipedia than it is in the rest of society, except for the fact that Wikipedia, like the rest of Usenet, adds anonymity and portable and changable identities (something that is getting harder for people to do in meatspace). Meatspace has its credentials, letters of recommendation, CVs, and so on. Wikipedia's answer to this is very poor.

Nevertheless, I think you're onto something. I'm just not quite sure what, yet.

Here is an example of Wales utilizes aspects of a "reputational economy from today. Thanks to The Wale Hunter. I will reply here to avoid the passions of the ArbCom election and focus on this apect:




Notice that the factor Wales uses to measure "reputation" is adminship, not anything related to article writing. I will repeat here my standby characterization of an admin as an untalented 15 year old who pisses away a summer holed up in a basement bedroom with a laptop assembling 5,000 edits.

Trudat. There are only a few thousand active admins, and they have a stranglehold on the place.