Wikipedia: charting the decline in participation

[Poetlister]

Well, I have a hunch that the fit of that line may not prove to be so good in the out years...

Is that your way of saying "I haven't got anything better"?

Projecting a trend indefinitely into the future is not rational. The curve will, of course, flatten out.

RfB

Why not plot it with a log scale on the Y axis? Then if it's an exponential decline it would appear to be a straight line.

[Peter Damian]

Why not plot it with a log scale on the Y axis? Then if it's an exponential decline it would appear to be a straight line.

Too much statistics. I've said this before: when a chart is in a sense under the control of the people it is about (the administrators themselves, who can choose to stay or leave, and who can also choose the eligibility requirements), and if any 'trend' is also visible to them, they will take action of some sort. This will interfere with the trend, of course. If they decide that 500 is the absolute minimum, and if they are effective in changing eligibility, by e.g. setting a quota, then it will level off. If they realise that it is is impossible to stem the trend, and if they are rational, then it is clearly a lost cause, and they may all decide to desert en masse. Trying to infer the future from the past is impossible when the future is wholly or partly determined by rational expectations based on the same inference (and inferences about the inference, and so on).

Keynesian beauty contest (T-H-L)

[Textnyymi]

What about this?

https://www.google.com/trends/explore#q=wikipedia

Wikipedia went from 97 (march 2010) to 41 (march 2014).

[neved]

Why not plot it with a log scale on the Y axis? Then if it's an exponential decline it would appear to be a straight line.

Too much statistics. I've said this before: when a chart is in a sense under the control of the people it is about (the administrators themselves, who can choose to stay or leave, and who can also choose the eligibility requirements), and if any 'trend' is also visible to them, they will take action of some sort. This will interfere with the trend, of course. If they decide that 500 is the absolute minimum, and if they are effective in changing eligibility, by e.g. setting a quota, then it will level off. If they realise that it is is impossible to stem the trend, and if they are rational, then it is clearly a lost cause, and they may all decide to desert en masse. Trying to infer the future from the past is impossible when the future is wholly or partly determined by rational expectations based on the same inference (and inferences about the inference, and so on).

Keynesian beauty contest (T-H-L)

I agree. The WMF has more than enough free money to prevent their sites from going into abyss where they belong. I think that the only way to deal with Wikipedia is to come up with something much better, much more decent, much more sane than wikipedia is. Larry Sanger tried once, but failed. Hopefully the site he's working on now, would do a better job, although I believe it is hard to impossible to have something big and at the same time sane and decent on the NET.

[Hex]

There has been a massive exodus of admins from the site of the last year or so and very few new admins being created.

As you can see from the figures I've compiled at Wikipedia:Desysoppings by month (T-H-L), this is not entirely true. In the years following the enactment of the inactivity policy:

2011: 278 out, 52 in, net change -226 (large cull of inactive admins)
2012: 117 out, 28 in, net change -89
2013: 81 out, 34 in, net change -47
2014 so far: 30 out, 10 in, net change -20

The rate at which the total number of admins is decreasing is itself decreasing. This is entirely unsurprising as the huge bulge in admins created in the boom period of 2005-7 moves into the past. A couple of years from now the in and out numbers are probably going to hit parity.

[Randy from Boise]

There has been a massive exodus of admins from the site of the last year or so and very few new admins being created.

As you can see from the figures I've compiled at Wikipedia:Desysoppings by month (T-H-L), this is not entirely true. In the years following the enactment of the inactivity policy:

2011: 278 out, 52 in, net change -226 (large cull of inactive admins)
2012: 117 out, 28 in, net change -89
2013: 81 out, 34 in, net change -47
2014 so far: 30 out, 10 in, net change -20

The rate at which the total number of admins is decreasing is itself decreasing. This is entirely unsurprising as the huge bulge in admins created in the boom period of 2005-7 moves into the past. A couple of years from now the in and out numbers are probably going to hit parity.

I believe this is accurate.

tim

[TungstenCarbide]

There has been a massive exodus of admins from the site of the last year or so and very few new admins being created.

As you can see from the figures I've compiled at Wikipedia:Desysoppings by month (T-H-L), this is not entirely true. In the years following the enactment of the inactivity policy:

2011: 278 out, 52 in, net change -226 (large cull of inactive admins)
2012: 117 out, 28 in, net change -89
2013: 81 out, 34 in, net change -47
2014 so far: 30 out, 10 in, net change -20

The rate at which the total number of admins is decreasing is itself decreasing. This is entirely unsurprising as the huge bulge in admins created in the boom period of 2005-7 moves into the past. A couple of years from now the in and out numbers are probably going to hit parity.

I believe this is accurate.

That's predicated on no viable alternatives to Wikipedia popping up.

[Poetlister]

There has been a massive exodus of admins from the site of the last year or so and very few new admins being created.

As you can see from the figures I've compiled at Wikipedia:Desysoppings by month (T-H-L), this is not entirely true. In the years following the enactment of the inactivity policy:

2011: 278 out, 52 in, net change -226 (large cull of inactive admins)
2012: 117 out, 28 in, net change -89
2013: 81 out, 34 in, net change -47
2014 so far: 30 out, 10 in, net change -20

The rate at which the total number of admins is decreasing is itself decreasing. This is entirely unsurprising as the huge bulge in admins created in the boom period of 2005-7 moves into the past. A couple of years from now the in and out numbers are probably going to hit parity.

We're just under 1/3 of the way through 2014, so a best guess of the final figures for 2014 would be

2014: 90 out, 30 in, net change -60

[Kelly Martin]

The rate at which the total number of admins is decreasing is itself decreasing.

Is it decreasing faster, slower, or at the same rate at which the number of admins is decreasing? If administrative resignations are independent processes (like, say, radioactive decay events), we'd expect the number of admins who resign in any given period to be proportional to the (average) number of admins during that period (just as the rate of decay events from a given sample decreases over time). Therefore, a decrease in the rate at which admins resign does not in itself indicate that the process is approaching a non-zero steady state.

In practice, of course, administrative resignations are probably not independent events, but exactly what they cluster about is not obvious to me.

My expectation is that the slow quasi-linear decline will continue until either a major incident occurs (which could cause it to go either way depending on the nature of the incident) or a viable alternative to Wikipedia pops up (which will lead to a precipitous mass exodus).

[Kumioko]

The rate at which the total number of admins is decreasing is itself decreasing.

Is it decreasing faster, slower, or at the same rate at which the number of admins is decreasing? If administrative resignations are independent processes (like, say, radioactive decay events), we'd expect the number of admins who resign in any given period to be proportional to the (average) number of admins during that period (just as the rate of decay events from a given sample decreases over time). Therefore, a decrease in the rate at which admins resign does not in itself indicate that the process is approaching a non-zero steady state.

In practice, of course, administrative resignations are probably not independent events, but exactly what they cluster about is not obvious to me.

My expectation is that the slow quasi-linear decline will continue until either a major incident occurs (which could cause it to go either way depending on the nature of the incident) or a viable alternative to Wikipedia pops up (which will lead to a precipitous mass exodus).

I agree, everything we say at this point is really just speculation because a lot of things could change it to improve or decline. Implementing Flow or reimplementing Visual Editor could affect the outcome. The WMF or the community could do a number of things to drop or increase. At this point though, IMO, the project will continue to decline because the site is overrun by admins who are abusive and don't care about anything else except stroking their own ego and making themselves feel important.

[Peter Damian]

Here is a chart of the average net change in number of active admins arriving (positive) or departing (negative) each month, from 2007-2013. That is, if the net change is up, regardless of quantity, you give a 1 for the month of that year. If net change down, give -1, with 0 for no change. Then average it for that same month for all the years.

Data was taken from this page.

It confirms the seasonality I had always suspected. For some reason, February sees an increase in active admins. Likewise December. Slight increase in January. Net negative for all others months. Suggests that administrators quite like gnoming in the winter months?

The data is to the end of 2013 only, but this year saw the same effect. A slight uptick in January, an increase in February, then massive falls throughout March and April.

[neved]

The second is the beloved Giano (T-C-L). Not edited since April 4. Giano always leaves some kind of message when taking a break, or when ‘retiring’. Not so.

There could be other explanations.

Here it is:

It's not a wikibreak. Thanks for all the emails - for the record and to answer a few of the questions: I am not ill (that I know of); I have not been sent to prison, nor am I in an asylum for the insane (pity; I could have organised a wiki-meet). I'm just sick to death of this self-destructing project (eg: just today, and all the morons who never write a word, but want to enforce the letters of the law and make pointless rule, and then become daft little Admins on the back of their negative 'work.' I have not retired or stormed off in a huff (No, if it quacks like a duck, it's not necessarily a fucking duck); I am just taking time to enjoy the summer and assess where I want to be with this project, which at the moment is a long way from it. Giano (talk) 17:57, 1 May 2014 (UTC)

[Freddy]

The second is the beloved Giano (T-C-L). Not edited since April 4. Giano always leaves some kind of message when taking a break, or when ‘retiring’. Not so.

There could be other explanations.

Here it is:

It's not a wikibreak. Thanks for all the emails - for the record and to answer a few of the questions: I am not ill (that I know of); I have not been sent to prison, nor am I in an asylum for the insane (pity; I could have organised a wiki-meet). I'm just sick to death of this self-destructing project (eg: just today, and all the morons who never write a word, but want to enforce the letters of the law and make pointless rule, and then become daft little Admins on the back of their negative 'work.' I have not retired or stormed off in a huff (No, if it quacks like a duck, it's not necessarily a fucking duck); I am just taking time to enjoy the summer and assess where I want to be with this project, which at the moment is a long way from it. Giano (talk) 17:57, 1 May 2014 (UTC)

I hope he decides to stay.

[Kumioko]

The second is the beloved Giano (T-C-L). Not edited since April 4. Giano always leaves some kind of message when taking a break, or when ‘retiring’. Not so.

There could be other explanations.

Here it is:

It's not a wikibreak. Thanks for all the emails - for the record and to answer a few of the questions: I am not ill (that I know of); I have not been sent to prison, nor am I in an asylum for the insane (pity; I could have organised a wiki-meet). I'm just sick to death of this self-destructing project (eg: just today, and all the morons who never write a word, but want to enforce the letters of the law and make pointless rule, and then become daft little Admins on the back of their negative 'work.' I have not retired or stormed off in a huff (No, if it quacks like a duck, it's not necessarily a fucking duck); I am just taking time to enjoy the summer and assess where I want to be with this project, which at the moment is a long way from it. Giano (talk) 17:57, 1 May 2014 (UTC)

I for one have come to realize how shallow the friends I thought I had in Wikipedia were. Sometimes in life, we align ourselves with "friends" that do nothing but bring us down and hold us back from our true potential. At some point it becomes necessary, if we are too succeed in life, to cut that dead weight loose and find new friends who are thoughtful and intelligent and do not bring us down to their level. Wikipedia is very very much like this. The so called friends in Wikipedia are a lot like scaffolding apparently...just temporary and not meant to last, to be torn down and hidden away when not needed and only brought back out when they are needed for something.

[Volunteer Marek]

Well, I have a hunch that the fit of that line may not prove to be so good in the out years...

Is that your way of saying "I haven't got anything better"?

Projecting a trend indefinitely into the future is not rational. The curve will, of course, flatten out.

RfB

well, yes, it of course will "flatten out", unless someone invents negative editors.

[Kevin]

...unless someone invents negative editors.

Surely this describes a large part of the active editing population?

[Kumioko]

There has been a massive exodus of admins from the site of the last year or so and very few new admins being created.

As you can see from the figures I've compiled at Wikipedia:Desysoppings by month (T-H-L), this is not entirely true. In the years following the enactment of the inactivity policy:

2011: 278 out, 52 in, net change -226 (large cull of inactive admins)
2012: 117 out, 28 in, net change -89
2013: 81 out, 34 in, net change -47
2014 so far: 30 out, 10 in, net change -20

The rate at which the total number of admins is decreasing is itself decreasing. This is entirely unsurprising as the huge bulge in admins created in the boom period of 2005-7 moves into the past. A couple of years from now the in and out numbers are probably going to hit parity.

We're just under 1/3 of the way through 2014, so a best guess of the final figures for 2014 would be

2014: 90 out, 30 in, net change -60

Contrary to what some have said above a net loss of 60 admins on the site, even as an estimate, which I think might be pretty accurate is significant. Lets also remember that it comes in waves. Major changes like visual editor and certain times of the year show a higher churn rate than other times. It may seem steady now but all it would take to change that would be a major change such as the WMF making changes to try and quell the toxic editing environment and you would see a spike. Personally I am to the point where the WMF needs to wipe the slate clean. They need to lift the vast majority of blocks and bans, they need to eliminate all the admins except for a select few and make them go back through the RFA process. The RFA process is bad because there is no incentive to change it, if everyone has to go back through, it will force change. And that is assuming that the WMF wouldn't change that. Adminship shouldn't be a big deal, it never should have been. It was a select group of bigheaded admins that made it that way and that needs to be undone.

[EricBarbour]

I have updated my chart of successful/failed RFAs. Posted on the WO wiki. (Do you see any "improvement" in there?)


Also note that since 2005, the percentages of failed vs. successful RFAs has been very roughly the same: 66% fail, 33% succeed.
That was when the paranoia really started to set in.

[Volunteer Marek]

I have updated my chart of successful/failed RFAs. Posted on the WO wiki. (Do you see any "improvement" in there?)


Also note that since 2005, the percentages of failed vs. successful RFAs has been very roughly the same: 66% fail, 33% succeed.
That was when the paranoia really started to set in.

You got a graph of the share that was successful?

[Poetlister]

Interesting how few repeat attempts there are.

[thekohser]

Interesting how few repeat attempts there are.

The graph only shows the repeat attempters who were stupid enough not to vanish and return under a new account name.

[EricBarbour]

Interesting how few repeat attempts there are.

The graph only shows the repeat attempters who were stupid enough not to vanish and return under a new account name.

Correct. A good number of them failed at RFA and changed their names. It's really difficult to track all of those instances. WP:Changing username does not list all of them. Some were made with the connivance of admin friends, many times they just created a new account and said nothing about their old one. Impossible to study.

You got a graph of the share that was successful?

?? That's what the blue area is.

If you want to make your own chart, email me and I will send you the LibreOffice spreadsheet.

[Volunteer Marek]

Interesting how few repeat attempts there are.

?? That's what the blue area is.

If you want to make your own chart, email me and I will send you the LibreOffice spreadsheet.

I know, what I want to see is the ratio of the blue area to the orange area. I'll email you.

[Volunteer Marek]

The thing is, however you slice it, there's a big crazy break in 2007 in terms of editor participation. The break in editors trying out for RfA happens a bit later (eyeballing Eric's chart, early 2008?). This may simply have to do with the fact that you got to run up your edit count before you try for RfA so the latter variable is naturally gonna lag editor participation. Interpretation could be a little different if the % of people accepted as admins changed too.

But the basic question is still the same. Why 2007?

If the explanation is that "slowly people realized that Wikipedia is sort of stupid" then you'd expect a smooth trend which gradually reverses itself. But here what we have is a trend which suggests everyone thinking "Wikipedia is teh awesome encyclopedia!" one day, and then waking up the next day and thinking "damn that site sort of sucks, I feel embarrassed about having been part of it". It's a "break" not a "gradual reversal in trend".

Did Essjay really kill Wikipedia? Or was it some kind of critical mass (meaning, given how dysfunctional the site is, it was bound to happen sooner or later and 2007 just happened to be the year it did, and Essjay is just a spurious correlation)? In a way it would be really really funny if it turned out in that just one phony poseur was responsible for essentially killing the attempt to bring the "sum of all human knowledge" to the "starving girl in Africa". With Jimbo's approval.

[Kelly Martin]

Did Essjay really kill Wikipedia? Or was it some kind of critical mass (meaning, given how dysfunctional the site is, it was bound to happen sooner or later and 2007 just happened to be the year it did, and Essjay is just a spurious correlation)? In a way it would be really really funny if it turned out in that just one phony poseur was responsible for essentially killing the attempt to bring the "sum of all human knowledge" to the "starving girl in Africa". With Jimbo's approval.

Personally, I think it has to do with the fall 2006 change prohibiting IP editors from creating articles, but I don't understand why it lags so much.

[EricBarbour]

Did Essjay really kill Wikipedia? Or was it some kind of critical mass (meaning, given how dysfunctional the site is, it was bound to happen sooner or later and 2007 just happened to be the year it did, and Essjay is just a spurious correlation)? In a way it would be really really funny if it turned out in that just one phony poseur was responsible for essentially killing the attempt to bring the "sum of all human knowledge" to the "starving girl in Africa". With Jimbo's approval.

Personally, I think it has to do with the fall 2006 change prohibiting IP editors from creating articles, but I don't understand why it lags so much.

I suspect there were many contributing factors, but Essjay was the first really big internal scandal that ended up in the news media. Remember back to all those old WR threads about SlimVirgin, Gary Weiss, Jayjg etc. etc. acting like assholes -- that stuff was ignored outside WP circles. But then you had Essjay, which had been building up for two years prior. Not everyone loved Essjay.

In quick succession, the big media stories during the next year:

• *Tim Pierce assigns WP vandalism to his students, and gets threats from WP admins
  *Essjay
  *Taner Akçam is arrested at Montreal airport based on crap in his Wikipedia article
  *Search Wikia (and Jimbo on the covers of business magazines, being hailed as a "genius" )
  *Fuzzy Zoeller's lawsuit
  *Danny Wool and Brad Patrick resigning
  *the staggering idiocy of BADSITES
  *Carolyn Doran fired/arrested
  *Sue Gardner and Mike Godwin are hired
  *Andrew Keen's book comes out
  *the de Braeckeleer article about SlimVirgin
  *Richard Dawkins criticizes Wikipedia on his TV show (almost forgotten today)
  *Mantanmoreland and Judd Bagley
  *the secret mailing list Durova was running
  *WMF moves from Tampa (cheap living) to San Francisco (one of the most expensive cities in the world)
  *Rachel Marsden
  *Jimbo's expense account flap and extramarital affairs
  *Jimbo is photographed on vacation with Richard Branson and Tony Blair

February 2007-March 2008 was an "eventful" year. And all of this was reported in the media. It added up.

The Siegenthaler mess in 2005 should have harmed participation, but it had the exact opposite effect. I've often wondered if unscrupulous people saw Siegenthaler happen, and though, "gee, if you can become powerful on Wikipedia, you can rewrite history" or some such idea. Which turns out to be somewhat true, many years later.

[Vigilant]

You forgot the Merkey-madness.

[EricBarbour]

You forgot the Merkey-madness.

http://news.bbc.co.uk/1/hi/technology/7291382.stm
Merkey was not seen as "reliable", for obvious reasons. But that's another possible item.
And shortly thereafter, RFA requests began to plummet.

[Vigilant]

You forgot the Merkey-madness.

http://news.bbc.co.uk/1/hi/technology/7291382.stm
Merkey was not seen as "reliable", for obvious reasons. But that's another possible item.
And shortly thereafter, RFA requests began to plummet.

I was in the middle of that shit.
Danny Wool, Kat Walsh, Jimmy Wales, Nicholas Turnbull (who quit over Danny calling him on a private line in the middle of the night to tell him not to mess with Merkey), Tony Sidaway, Duk, Proabviouac aka Timothy Usher, Guy Chapman

An utter cluster fuck.
Merkey is an unreliable witness for almost everything, but he made some sort of dealio with Danny and/or Jimmy to get free reign on en.wp. That much is incontrovertible.

He was indeffed by NichT.
Unblocked by a minion of the insiders and NichT told to afford special rights to Merkey, upon which NichT righteously quit. He later diva flounced back in.

He even apologized to the foundation, et al
http://en.wikipedia.org/w/index.php?tit ... d=58099750

Can't find the original post where he told them all to go fuck themselves though.

Edit: There we go
http://en.wikipedia.org/w/index.php?tit ... d=49092834

Revision as of 01:26, 19 April 2006

I have just been informed by Danny Wool of the Wikimedia Foundation, via #wikipedia-en-admins on Freenode, that I am expected to give preferential treatment to Jeffrey Vernon Merkey here on Wikipedia (User:Gadugi, User:Waya sahoni, and newer socks) by not applying admin actions to him - simply because, apparently, he's promised the Foundation money. Well, that's it, I'm not standing for this. I will not participate in a project where users can pay to have their editing privileges restored after they've been consistently unpleasant, made aggressive vanity editing of articles, launched legal threats against Wikipedians, etc. I have no further interest on continuing here. --NicholasTurnbull | (talk) 01:26, 19 April 2006 (UTC)

[everyking]

The thing is, however you slice it, there's a big crazy break in 2007 in terms of editor participation. The break in editors trying out for RfA happens a bit later (eyeballing Eric's chart, early 2008?). This may simply have to do with the fact that you got to run up your edit count before you try for RfA so the latter variable is naturally gonna lag editor participation. Interpretation could be a little different if the % of people accepted as admins changed too.

But the basic question is still the same. Why 2007?

If the explanation is that "slowly people realized that Wikipedia is sort of stupid" then you'd expect a smooth trend which gradually reverses itself. But here what we have is a trend which suggests everyone thinking "Wikipedia is teh awesome encyclopedia!" one day, and then waking up the next day and thinking "damn that site sort of sucks, I feel embarrassed about having been part of it". It's a "break" not a "gradual reversal in trend".

Did Essjay really kill Wikipedia? Or was it some kind of critical mass (meaning, given how dysfunctional the site is, it was bound to happen sooner or later and 2007 just happened to be the year it did, and Essjay is just a spurious correlation)? In a way it would be really really funny if it turned out in that just one phony poseur was responsible for essentially killing the attempt to bring the "sum of all human knowledge" to the "starving girl in Africa". With Jimbo's approval.

That was a bad time. I'd say that was about the time that the abusive, toxic atmosphere was at its worst. I think that the escalating abuses and controversies and especially the relentless banning and running off of contributors simply outstripped the ability of the site to attract new, replacement editors around that time.

[Johnny Au]

Remember Veropedia (T-H-L)? It was founded by Danny Wool. He was from Toronto's north end near Wilson (TTC) (T-H-L).

I was a member of Veropedia.

[Kelly Martin]

Remember Veropedia (T-H-L)? It was founded by Danny Wool. He was from Toronto's north end near Wilson (TTC) (T-H-L).

I was a member of Veropedia.

I was involved in Veropedia but backed away when it became clear that Danny was simply designating his friends as "experts" instead of doing any sort of systematic expert review, and when the original plan to develop automated methods to identify quality editors (e.g. using WikiBlame techniques to determine who contributed content to articles identified as quality, then search for other articles written by the same editors) went by the wayside as "too difficult to implement".

Veropedia was always mainly an effort for Danny to get a meal ticket.

[Peter Damian]

But the basic question is still the same. Why 2007?

This paper agrees with my own experience of that period. Wikipedia was growing massively throughout late 2005, 2006 and then 2007. Before that it was still a sort of village atmosphere where it was OK that anyone can edit, and there was also still a taboo against banning people. But then there was this tidal wave of vandalism, as well as a whole bunch of people who were clearly crazy. There were people like that before, but it was manageable.

They couldn’t drop the policy that ‘anyone can edit’ because that was one of the pillars that supposedly made Wikipedia great (remember the creation myth that warned of a Nupedia approach to vetting contributors). So instead they prevented anyone from editing by blocking them, and by the use of automated tools like Huggle and Rollback which were designed and developed in the face of the tsunami of vandalism.

That explains the immediate decline after 2007. Why the decline continued is a matter of conjecture.

[Poetlister]

Interesting how few repeat attempts there are.

The graph only shows the repeat attempters who were stupid enough not to vanish and return under a new account name.

If you've only just missed and had several people saying "I'll vote for you next time", it must be worth a second attempt.

[Kelly Martin]

There was certainly an upswing in automated tools in 2006. We had a handful of them in 2005, mostly just things that made it easier to access the recent changes stream and filter it for "likely problematic edits". But they didn't make edits for you, just displayed them for you; you still had to hit all the buttons yourself to revert or whatever. Autowikibrowser showed up in 2006 (I think) and actually did whole edits for you without you having to actually make each one by hand. I used AWB a lot to fix stupid spelling errors, which is doofus work but work that needs doing. AWB wasn't much use for vandalism because it didn't access the Recent Changes stream. Later in 2006 we started seeing scripts like the predecessors to Huggle and Twinkle, which not only edited articles but also, critically, inserted templated messages onto user talk pages. I think this is where Wikipedia went off the rails. Prior to the Huggle/Twinkle era, messages on talk pages were mostly composed for the instant: you'd have something to say, and you'd type what it was and say it. You might have a wording you tended to use, but you'd actually type it out yourself and the messages were personal and personalized. The "automated caretaker" era changed that. Editor to editor communication became impersonal. For new editors on Wikipedia, starting in late 2006, the first message, and often the first several messages, that a new editor would get would be an obviously boilerplated template, often carrying veiled threats (a lot of "welcome templates" have veiled threats in them). I'm sure many new editors were intimidated, rather than welcomed by these.

I created a sockpuppet in 2007 (I think it was) and started editing with it. After only a few edits, I got a "welcome template" from an administrator by the name of Kukini (T-C-L). I removed it from my talk page after reading it (an option that the "user talk page policy" clearly allows to editors). Kukini reinstated it and threatened me with a block for deleting it.

I'm of the opinion that this is the change that fostered the bend in the curve in the participation graph in early 2007: the change in Wikipedia culture that was fostered in by the introduction and widespread use of "antivandalism scripts" operated mainly by incompetent, occasionally malicious imbeciles.

[TungstenCarbide]

Remember Veropedia (T-H-L)? It was founded by Danny Wool. He was from Toronto's north end near Wilson (TTC) (T-H-L).

I was a member of Veropedia.

I was involved in Veropedia but backed away when it became clear that Danny was simply designating his friends as "experts" instead of doing any sort of systematic expert review...

Heh, same thing with the admins irc channel, just Danny's buddies at first.

[thekohser]

Veropedia was always mainly an effort for Danny to get a meal ticket.

Ironic, isn't it then, that some of the editors that he owed prize money for their work, had to wait months and months (or was it years?) to finally get paid by him?

[Vigilant]

Veropedia was always mainly an effort for Danny to get a meal ticket.

Ironic, isn't it then, that some of the editors that he owed prize money for their work, had to wait months and months (or was it years?) to finally get paid by him?

Heh.

A tie for second place goes to Cwmhiraeth (talk · contribs) for her wonderful work in Poultry

[Kelly Martin]

Veropedia was always mainly an effort for Danny to get a meal ticket.

Ironic, isn't it then, that some of the editors that he owed prize money for their work, had to wait months and months (or was it years?) to finally get paid by him?

After Danny left the WMF he was without meaningful employ for quite a long time, so I'm not really all that surprised.

[Volunteer Marek]

Rfa Fail.png

Here's the % of attempted RfAs that ended in failure, based on data Eric sent me. I've put in two smoothing moving averages, 11 month and 21 months, to give the sense of the trend. The failure rate increased from April 2004 (before that, pretty much anyone who tried, got it) up to... again, sometime in 2007. Then it stays more or less constant, though the volatility increases quite a bit, most likely due to the lower number of attempts (sample size down, variance up)

Th 11 month moving average may make it seem like there was some kind of trend reversal in 2013 but that's really only due to a single month when there were very few attempts and they all passed (that zero on the blue line).

So yeah, right now it looks stabilized at about 60%. To the extent that it may be true that "the atmosphere at RfA has been getting worse", there's probably a good amount of self selection going on with only people even trying either out of touch hopeless cases or people who are shoe-ins.

[EricBarbour]

There was certainly an upswing in automated tools in 2006. We had a handful of them in 2005, mostly just things that made it easier to access the recent changes stream and filter it for "likely problematic edits". But they didn't make edits for you, just displayed them for you; you still had to hit all the buttons yourself to revert or whatever. Autowikibrowser showed up in 2006 (I think) and actually did whole edits for you without you having to actually make each one by hand. I used AWB a lot to fix stupid spelling errors, which is doofus work but work that needs doing.

Something that should be built into MediaWiki's editor. But isn't, along with a load of other things.

Later in 2006 we started seeing scripts like the predecessors to Huggle and Twinkle, which not only edited articles but also, critically, inserted templated messages onto user talk pages. I think this is where Wikipedia went off the rails. Prior to the Huggle/Twinkle era, messages on talk pages were mostly composed for the instant: you'd have something to say, and you'd type what it was and say it. You might have a wording you tended to use, but you'd actually type it out yourself and the messages were personal and personalized. The "automated caretaker" era changed that. Editor to editor communication became impersonal. For new editors on Wikipedia, starting in late 2006, the first message, and often the first several messages, that a new editor would get would be an obviously boilerplated template, often carrying veiled threats (a lot of "welcome templates" have veiled threats in them). I'm sure many new editors were intimidated, rather than welcomed by these.

You make yet another good point. The "golden age" of bot creation/editing was late 2006 thru 2007, and that was when the decline started. That was also when the snotty "welcome templates" started to appear.

[EricBarbour]

Rfa Fail.png

So yeah, right now it looks stabilized at about 60%. To the extent that it may be true that "the atmosphere at RfA has been getting worse", there's probably a good amount of self selection going on with only people even trying either out of touch hopeless cases or people who are shoe-ins.

Great chart, thanks! Wikipedia's admin "scene" is so deeply inbred and paranoid by now I'm amazed anyone even bothers to apply for admin, other than people with years of IRC butt-kissing behind them.

[HRIP7]

In March 2014, new account registrations in the German Wikipedia stood at 694, the lowest figure since February 2005. (The February figure has been adjusted upwards slightly since this thread started, to 761.)

In the English Wikipedia, new account registrations stood at 6,374 (compared to 6,802 in March 2013, and a high of 13,845 in March 2007). Editors making more than 100 edits per month were down from 3,327 in March 2013 to 3,148 in March 2014. (In that table, too, various figures have changed a little since the start of this thread.)

[HRIP7]

According to the April figures currently displayed* on stats.wikimedia.org new account registrations in the English Wikipedia have dropped below 6,000 for only the second time since November 2005 (the first time was September 2013; September figures are always low, presumably because this is the month when school starts after the summer break in many countries). Editors making > 100 edits per month were again below 3,000 (2995 vs. 3282 in April 2013).

In the German Wikipedia, editors making > 100 edits per month dropped below 900 for the first time since April 2006. New Wikipedians were below 600 for the first time since August 2004.

* I say "currently displayed" because the figures often change a little in the first weeks after being posted, usually (but not always) going up.

[Silent Editor]

But the basic question is still the same. Why 2007?

Could it be simply be that that's when Facebook really started to take off?

[everyking]

According to the April figures currently displayed* on stats.wikimedia.org new account registrations in the English Wikipedia have dropped below 6,000 for only the second time since November 2005 (the first time was September 2013; September figures are always low, presumably because this is the month when school starts after the summer break in many countries). Editors making > 100 edits per month were again below 3,000 (2995 vs. 3282 in April 2013).

In the German Wikipedia, editors making > 100 edits per month dropped below 900 for the first time since April 2006. New Wikipedians were below 600 for the first time since August 2004.

* I say "currently displayed" because the figures often change a little in the first weeks after being posted, usually (but not always) going up.

This is sad and frustrating. The trend ought to be going in the other direction. The simple fact of the matter is that Wikipedia cannot attract or retain people anymore because the atmosphere is toxic and unbearable, and many people are not allowed to contribute even if they want to. The main page still says "anyone can edit", but that has long since become an empty slogan, not an accurate description. If the project gets back to the spirit of "anyone can edit", the trend will reverse.

[Johnny Au]

Approximately 0.05% to 0.1% of all English Wikipedia articles (many of which happen to be some of the most popular and/or controversial articles) are semi-protected or full-protected at least once in their history.

Is it me or is it that when Fakebook and TWITter become mainstream, Wikipedia becomes less popular?

Fakebook and TWITter show up everywhere. Wikipedia, not so much. In fact, more people believe that WikiLeaks is a part of WMF than edit Wikipedia articles at least once a day.

[sparkzilla]

But the basic question is still the same. Why 2007?

It's not because of Facebook or Twitter, which have different user profiles to the average Wikipedian. It's because by 2007 most of the boring encyclopedia stuff was done. Once the pages had been built editors who had become invested in their work changed from creation to defense. It's a lot easier to delete someone else's work than to write new text, a vicious circle that leads to the current situation, where new editors who have enthusiasm to write are attacked by established, lazy, editors who only need to revert.

The main areas where people still think it is worth contributing are 1) very obscure topics where they wont get edited (not interesting to most new writers), news-based topics where they have a chance of actually getting in there fast (but everyone is chasing the same few topics), and controversial topics (Israel, Global warming) that attract zealots, who are naturally aggressive. None of these are conductive to new editors.

The problem could have been mitigated by not allowing editors to feel they have ownership over the topics they write, and to have every edit judged purely on its merits. In practice this would have meant the separation of the writing and editing roles and more structure to the data on the page.

This is sad and frustrating. The trend ought to be going in the other direction.

It is an inevitable result of the way the wiki system works. Wikipedia's success and failure are wrapped in the same idea: "anyone can edit". I have argued elsewhere that the software creates and maintains the toxic culture, not the other way around.

[EricBarbour]

But the basic question is still the same. Why 2007?

It's not because of Facebook or Twitter, which have different user profiles to the average Wikipedian. It's because by 2007 most of the boring encyclopedia stuff was done. Once the pages had been built editors who had become invested in their work changed from creation to defense.

That's part of it. Another part was clearly Essjay, which triggered a LOT of anger for Jimbo and his supporters. Thus, content writers started leaving.

[Johnny Au]

...and how, despite averaging ten edits a day, I manage to be in the top 2100 in terms of most edits made by a human, as well as the top 1500 in edits made per month.

Most of my edits are WikiGnomish with very little substantial content creation and approximately 99.9% of my mainspace edits are Toronto-related (and surround regions).