How Wikipedia's reaction to sudden popularity is causing its decline
www-users.cs.umn.edu
www-users.cs.umn.edu
I once saw a great example when a notable athlete was traded from one team to another... I think it was a famous basketball player, but can't remember who. On a whim I checked the article history page and saw the same pattern: a new user to Wikipedia would make the edit, and it was quickly reverted by one established editor.
When I looked at the talk page, the WP editor made it clear that it was Official Wikipedia Policy that even though both clubs had announced the trade, and news was being carried on multiple outlets, the trade wasn't official enough until the NBA released a daily transfer list with his name on it. All five new editors were threatened with WP:3RR and WP:yaddayadda instead of encouraged. But hey, when the list was officially released twenty-five minutes later the editor who got the page locked got to make the edit. Great job guys! Way to add value to the project!
This same behavior happens on every site where authority is meted out based on activity, not ability. I once tried to correct an elisp entry on a Stack Overflow answer: my edit was autorejected because it had less than five characters difference. So I added a fairly pointless comment to get past the automatic rejection, only to have someone reject it because the comment wasn't sufficiently useful (go figure). I was finally able to get someone to make a correction by going into the Stack Overflow chat and trying to explain things... except the user, who freely admitted to never using emacs before, messed up and made things worse. Whatever. My give-a-fucks were all out. It's probably still wrong!
Overall it seems to be pretty fiercely territorial, so no surprise why they have trouble attracting new contributors.
So stackoverflow went from "having a useful answer" to "having no answer at all" because folks abused their use-granted power.
I am personally fond of the "king for a day" form of crowd-sourced moderation, where folks who are granted moderation say don't get to keep it long enough to get used to it, and learn how to abuse it.
Slashdot was far from perfect, but I still think the system of temporary moderators followed up by temporary metamoderators is pretty elegant compared to most of the alternatives I've seen.
The Problem with Wikipedia is that Wikipedia does not want to reflect truth, or reality or facts or objectivity however one might call it, but verifiability.
That is why they revert to false statements, even if they know so.
Insisting on verifiability is a necessary evil if Wikipedia wants to have even a modicum of credibility. Maybe the standards are far too stringent, but verifiability is tangible, and the 'truth' isn't.
And once one accepts that one can drop this stupid principle. For me it seems the only reason it exists is to pander the academic community. But it's a hopeless and one sided love.
It does not improve quality, it does not improve credibility, the best thing about it that is produces funny feedback loops.
Articles about NBA players are not a good example of this, but I can easily think of plenty of other topics which would involve extremely loud and opinionated fringe communities who are very motivated to broadcast their views to the world. I can't think of any way to limit the influence of these communities except by insisting that every article have impeccable sources.
The cost of this insistence on sources is that sometimes utterly obvious edits may be reverted unless they can be properly sourced, but I think it is an acceptable price for maintaining consistent standards of quality across the wiki.
Also, I don't agree that the system only exists to pander to academia. Every encyclopedia in the world has standards of quality to which their articles must conform. The only difference between these other encyclopedias and Wikipedia is the speed with which they are updated.
Now I don't know about you, but I don't go to Wikipedia to get the latest news. I go there to learn stuff with the belief that this stuff will be correct. I don't care if players' teams are updated hours instead of seconds after a trade. I care much more about whether this information is correct. Verifiability builds trust, and I don't see how Wikipedia could continue to exist without it.
Maybe WPs process (speaking as someone that's not edited WP here) need to change more towards the distributed VCS workflow: everyone can make changes, but changes gets voted in. Not sure if that is scalable either -- but perhaps what is need is a three-tier system: default to open (like any wiki, just make the change, done); somewhat moderated (anyone can suggest an edit, votes (with comment field) need to pass/accept the vote [essentially making the talk page a vote to pass/fail], and locked (protecting popular articles against spam, as today).
Or maybe what is needed is to accept that WP is "everone's wiki" (eg: it's a great source for tv series episode dates and synopsis -- not really something that should go in a traditional encyclopedia) -- in other words, classify some pages/sections as "not encyclopedia grade". It's hard to see how one can have one process that both accommodates NBA transfers, tv series episode synopsis, modern history of Palestine and Ukraine, and articles that document the current state of photovoltaics and quantum physics.
I notice you use --. Discussion around dashes is extensive. Here's just one example: https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(policy...
That single discussion is around 15,000 words. Here's a search for some more: https://en.wikipedia.org/wiki/Special:Search?search=dash&pre...
Note the search results helpfully give a word count - there are several hundred thousand extra words there.
Dash-discussion was not limited to the style guide and its talk page, or to the village pump. Disputes were escalated to ANI (a notice board where incidents for administrator attention can be raised); finally ending up at arbitration. (This is pretty severe level of dispute.) https://en.wikipedia.org/w/index.php?title=Wikipedia:Arbitra...
PEDIT: I mean, just look at the article talk for "Mexican-American War" https://en.wikipedia.org/wiki/Talk:Mexican%E2%80%93American_... - roughly 21,000 words.}
And that's about –,— and -. Now imagine how editors behave over contentious issues. (What are those islands called and who do they belong to? is a good example.)
Wikipedia stopped being an encyclopedia long ago: now it's just Bureaucracy Simulator 2015 with the AI cranked up to nightmare.
Anything that resembles good faith effort and isn't flagrantly wrong gets ok'd (though others have complained about me doing that... I'm supposed to at least verify it or something).
Not everyone can down vote, flag, etc.
Irony isn't lost on me.
By policy anybody should be able to submit draft edits, and these edits should hang around in a review queue for a large-ish group of reviewers to look them over and thumb-up, thumb-down them in a "pink" review. Any acceptance or rejection should require 3 or 4 thumbs up or down. If a thumbdown is given, the reviewer must state what's wrong with the edit (bad spelling/grammar/poor citation, whatever).
If the edit is ultimately rejected, the person who submitted it now has 3 or 4 pieces of feedback, hopefully one of them useful. Now they can fix the edit, resubmit and now it goes to the red review.
If the edit passes it goes to the red review.
Red review is the same, but if it passes it becomes "published" and the change is made. Red reviewers should be the top editors of WP. Because the pink review should filter out most of the crap, they'll hopefully be dealing with higher level stuff and their rejection rates should be lower. Only in cases where the edit somehow was really bad, should they reject it.
Now here's the other thing. If a red review results in a rejection, the submitter has to start over, but the editors that passed it in the pink review are penalized in some way. Perhaps a Karma system (e.g. to become a red reviewer, you must have pink reviewed correctly 500 edits or something, to become a pink reviewer you must have 200 edits succeed). Red reviewers with too high of a rejection/pass ratio may lose their editing powers as well in case some power-mad editor starts bulk deletionist policies.
This way people can both submit edits and be engaged as editors (so they can both see the crap that editors have to deal with and participate in improving WP at the meta-level).
Problem editors can also be flagged by contributors, an editor with some number of flags (say 10) by different submitters should be reviewed. After each review, the number of flags that will cause the next review should increase (maybe double or something). This will keep high quality editors in place and not waste time reviewing editor's work.
(1) https://en.wikipedia.org/wiki/Wikipedia:Flagged_revisions
An important feature of my proposal is also to hold reviewers accountable...a notable feature that's missing in the current WP world. (I'm sure there's some purposefully byzantine process to do so today, but hell if a casual contributor can figure it out).
I wanted to know more about my representative to the state legislature. First I went to Wikipedia. It had a stub page, with only basic biographic information. Checking elsewhere, I found that the FBI and IRS raided his office, home, and business some time ago. His chief of staff, his business partner, and and a former staffer that became a judge were all convicted of fraud and/or tax evasion. Other than this, his political career was a whole lot of nothing.
So, having done this research, I put it into the relevant Wikipedia article, with appropriate references to news articles and legal documents, carefully noting that he himself was never charged with anything. Reading between the lines, I realize that the authorities tried to get his subordinates to roll over on him, but failed - but that kind of subjectivity doesn't belong in an encyclopedia article.
Anyway, less than a minute later, the change was reverted because of "vandalism". Whoever reverted it had either not even glanced at the change, or was a supporter of this politician, protecting his page from negative information. Probably the former, but who knows?
This was the last, and worst, example of me wasting my time trying to contribute to Wikipedia. I have had every single change I have ever made to Wikipedia reverted (going back years). Sometimes I have done it anonymously, sometimes I have logged in; it doesn't make any difference in my experience. What seems to matter is if you're willing to put in the time to master Wikipedia's bureaucratic procedure, which I am not. I just don't bother anymore. So to me, this article makes perfect sense.
But I have to admit that I still read Wikipedia all the time. It is without question, the most useful web site there is. It's not as comprehensive as I would like, and for subjective information it's actually quite bad, but I think that in general it seems to be getting better over time, on both counts. So I don't know if you can say it's declining.
By the way, despite him either being a crook, or merely employing crooks on his staff, I voted for him anyway, just last week. I really don't have a vendetta against this guy.
I am not interested in participating in a system where my hard work is discarded regardless of excuses.
So you end up with notability being rather arbitrarily decided by admins who see nothing wrong with long articles on obscure pieces of geek fandom while being unconvinced about the notability of people like influential executives at major companies, journalists, and the like.
I think the it's a good idea, but it's not a good fit for wikipedia as it exist today. I think it's grown to the point that for good or bad, it needs to be managed more like the Google corpus, than a traditional encyclopaedia. I'm not advocating abandoning human editing all together, but I think a new ideas are needed: wp's success has made it into a fundamentally new thing. It exists in the conceptual space between the sum of all newspapers and scientific journals, traditional encyclopaedia and the whole World Wide Web.
The problem arises on Wikipeida because its editors act as fact checkers. But your idea of what a fact is can be quite different from mine. On Newslines we create the archive and let the reader decide what is true, by marking claims as claims.
Very few bloggers break original news. If they do we can add it as a source.
And getting into Wikipedia has some obscure benefits that make it worthwhile.
All of which were basically from the same small, inexpensive PR campaign.
Wikipedia runs a serious risk of becoming a commercial echo chamber (if it hasn't already).
What're you guys doing at Newslines to prevent it from turning into a vehicle for self-promotion?
So basically, if you publish a blog with some information about a topic, then have "your friend" make the change and cite your blog, it's good. But try to do it yourself and it's no good.
I don't buy the argument that there isn't enough space on the internet. Wikis are at their worst on high-profile, contested topics, and at their best in minutiæ Wikipedia's entrenched cabal passionately hates.
Curious, as a social experiment. It's interesting it became as good as it did, but it's firmly in maintenance mode now and that won't keep the cobwebs off forever. I wonder how much it's being gamed by PR companies and governments now.
I am a Wikipedian. My experience as a Wikipedian suggests that Wikipedia's administrators need to be much more alert than they have been to the possibility that the Wikipedians motivated by money (or by ideological bias) will stay with the project and persist in making edits contrary to Wikipedia policy. They edit more articles, and edit in greater numbers, than most admins guess or notice. And the point-of-view-pushing editors often inject so much wikidrama into discussions about how to improve articles that they drive away the participation of conscientious editors who know reliable sources about the article topics. A new effort to build a free, online encyclopedia with lots of user input (which I would enthusiastically support) would have to figure out how to deal with this problem.
In respectful disagreement with some comments posted earlier in this thread, I think that most Wikipedia articles on many subjects are terrible, actively misleading and very poorly sourced. I am trying my best to fix articles on the topics for which I have the most reliable sources. (I have an office full of reference books for writing projects I do for my paid work.) It has been a long, slow slog since 2010 for me to improve the sourcing and make more neutral in point of view some of the most-read articles about the psychology of human intelligence, for example. Many Wikipedians don't read books, but only look up information on Google University, and think that anything that isn't found in an online source is not a verifiable fact. That makes the editing process very slow.
Without the biases of human beings, articles might start to say very strange and unwelcome things.
2) As a former WMF staffer - let me tell you they are well aware of the issues mentioned by Halfaker. They've been commissioning and releasing studies like this since 2009. They all had the same conclusion, that 2007 was the point where the community shifted, from collaboration to make cool stuff, to defense against vandalism.
It's not clear what the WMF can do to change this, without engaging in a "takeover" of the community or completely changing the model of how the wiki works. The former would alienate everybody, even the editors who want change. The latter is technically infeasible; even if they wrote new tools from scratch, Wikipedia interacts with everyone else's tools and plugins. Even if you change the id of a div you'll probably have a thousand people screaming at you the next day.
The WMF can propose that people be nicer, or write tools that are more welcoming to new users, but all that means is that you get to have your edit rejected with a prettier icon. The main problem is in the edit cycle itself, not the tool that's used.
Many editors think that Wikipedia is basically done, and repelling new users is actually a good thing. They aren't bad people -- they just think that encyclopedia quality is far more important than growing the community. So when they see something that seems wrong to them, they revert. Reversion, for Wikipedians, is really how they open the discussion. You're expected to keep going if you think your edit is good. But because we have a linear edit history, with only one live copy of the article, it feels like a slap in the face to the new user. Things that would pass muster for weeks or months, in the past, are reverted instantly. Often times the criticisms of the new edit are justified, but the standards are ludicrously high now. You'd better be ready to cite everything rigorously and be familiar with a zillion policies.
Not all of them are like this, but enough to make a gauntlet for new users. And those new users just go somewhere else.
As an aside, the first and only time I came across a bug in Mediawiki, back when I was pulling the nightly trunk, it turned out to be induced by a change of semantics in the minor version of PHP I had (which was well ahead of what Wikipedia was using at the time). Tim Starling not only weighed in, but took PHP to task. That semantic change in PHP was promptly reverted!
If you see him, say hi. He'll never remember me.
I found this page where you can listen to the 'sound of wikipedia', basically a different guitar string is plucked at a different pitch when certain changes are made to wikipedia articles. It is pretty cool in the sense that it shows a lot of changes are being made, however you are able to see and hear the amount of changes made by bots, which is surprisingly few
It is also especially fascinating to watch during a time with realtime news you care about. I've seen edits to Rosetta and Philae while just watching briefly!
The software causes almost every problem that people complain about, mainly because it lets real or imagined experts have too much power over the page. It's simply easier for an editor to reject new information, than to go through the pain of adding it. In one case, I struggled for weeks to try to get a simple sentence into the site, only to be harassed by the site's administrators, when I challenged a particularly ambiguous rule. Entrenched editors such as William Connolley [2] on Global Warming, act as guardians for whole sections of the site, and do not allow any dissenting opinion, censoring and minimizing views that do not suit their agenda. So much of this is simply unnecessary in a better designed system.
As a result of the open nature of wiki editing, Wikipedia has built up an increasingly restrictive, massive, rule set of how to reject information, which makes it increasingly difficult to add new information. Many of those rules, such as those for notability, conflict of interest, and reliable sources are ambiguous and arbitrarily applied, giving even more power to editors and admins, instead of content creators. This means that the site has become rigid and nobody wants to maintain it. The quality of pages is declining rapidly. It is also very easy to harass other people on the system, so much so that women and minorities are virtually excluded.
We have taken a different approach with great results. By avoiding the page-based metaphor and giving better tools to our editors have created a system that makes it easy to add information without being harassed. We also pay our writers and editors, and are implementing a revenue share system so they can actually get some benefit from their work. So far our writers have added 25,000 posts without a single edit war, and zero harassment. 80% of them are women.
[1] http://newslines.org [2] http://en.wikipedia.org/wiki/William_Connolley
>The software causes almost every problem that people complain about
The real problem is:
>As a result of the open nature of wiki editing, Wikipedia has built up an increasingly restrictive, massive, rule set of how to reject information
Wikipedia is not longer a place setup so people contribute information, it's a place set up to reject the addition of information (as you point out).
It's been a goal of my wiki to be the precise opposite, and it's been very interesting to see people go from "I hate wikis b/c of Wikipedia's stupid rules" to "huh, my edit was accepted" to "wow, this is extremely useful and collaborative".
I believe a wiki is not inherently bad, it's all the rejection rules they've built on top that are the problem.
I'd like to hear your reasons why you think this will be needed?
As long as the edit is something useful to traveling overland, it's allowed. That's the only rule, and I think it's the only rule that will ever be needed.
It's useful to note that this problem relates more to wiki-encyclopedias, and perhaps not as much to non-encyclopedic wikis. If you take a look at the C2 wiki or MeatballWiki, rejecting information make less of a problem even though they share the principle of anyone being able to edit anything on any page. MeatballWiki states that it is not an encyclopedia, but rather a set of pages which feature discussion. Because of this, there is less grounds to refuse or revert edits with the reason of being "off-topic" (because discussion usually means merging several related topics) or for being "biased" (again, a feature of discussion is to get one's personal viewpoint across).
In short, I don't think this is a problem due to wiki software or the open nature of wikis, but rather wikis-as-encyclopediae or even Wikipedia specifically. "Encyclopedia" places a huge set of rules as to what can be added and what not.
So, everyone -- including Mr. Halftaker -- when you publish anything, put a date on it!
I think the drop in new-contributor retention is an intended effect, an example of the system succeeding.
Paper encyclopedias followed the same pattern. Creating Encyclopedia Britannica was a huge job. Annual maintenance was done by a small staff.
Maintenance doesn't quite capture that.
That sounds exactly like maintenance to me, as distinct from documenting new things from scratch
An unnecessarily complicated and difficult to read article isn't really done.
Agreed. But I stopped contributing ages ago, because all of my contributions were almost immediately reversed.
It's pretty obvious that most pages are monitored by people with a stake in the content of those pages, and they are much more energetic in pitching their story than I am in trying to make some small improvements.
Maybe Wikipedia is ok for current stuff and the 20th century, but it barely scratches the surface for most of medieval Europe, for example. There's absolutely no need to be constrained to the terse articles and limited scope of a traditional encyclopedia in this respect. In fact, Britannica has substantially more information on certain topics.
I don't think Wikipedia should base its policies to be suitable for pop culture and sports fact checking if this conflicts in the slightest with good coverage on science, research, and well established philosophy and culture.
I can imagine how hard was it to get Wikipedia where it is today, therefore I will trust their judgement moving forwards.
It's not the best at everything but that's okay.
Or perhaps another example: most politicians have some sort of twitter or facebook account listed on their promotional material. It doesn't make those twitter accounts particularly popular, despite being frequently referenced.
It would be nice to see a graph where new article creation was declining exponentially now, with article turnover (editing) increasing exponentially.
And hey -- if I'm not perfect by the standards of "No Original Research" or even "Neutral Point of View", I don't have a role that allows me to be punished. I just move things in the right direction (by Wikipedia's standards as I understand them) as far as I can be bothered to, then go on with my life.
If these restrictions are really affecting the quality of the articles in Wikipedia, we'll see the quality of the articles on the fork grow, at which point they can be merged back by Wikipedia.
Wikipedia _was_ forked, several times. The reason you dont seem to know about them are:
1. The added utility was not sufficient to make an impact strong enough to pull people away from wikipedia.
2. The added no utility to the fork at all or made things bad at Wikipedia even worse.
Wikipedia is like democracy, it is the worst free encyclopedia we have, except all the others.
Who has the time to compete against something that is free and has mindshare?
Yet, for all of this, I have never been able to understand why, whenever a discussion about Wikipedia comes up on HN, so many people complain about getting all their changes reverted. I have edited Wikipedia for years and I think at most 3 cases did I see something reverted for no good reason. Maybe this is because I edit in less controversial areas (computer science), or because most of my edits are just typographic fixes, or because I also edit on different Wikipedias (especially the French one), or because I'm vaguely familiar with some of the customs or policies?
It would be really nice if people complaining about bad reverts could provide links to the diff of their edits, the reverts, and the subsequent discussion if any. One nice thing about Wikipedia is that the history is always public, so if something valuable has been reverted for bad reasons, it can be seen, and added back; if someone seems to be discouraging new users from contributing, anyone can see it and let them know; and if there was a misunderstanding, maybe we can see that the two sides were of good faith.
It's a bit frustrating to see everyone complain but never have links to objective facts where you can see what went wrong and try to understand why.
Kind of reminds me of Twitter banning all my games that let users tweet scores using their APIs as being too much like a full Twitter client. It's pretty obvious Twitter just aimed a bot at everyone and didn't actually look at who they were banning.
Edit: the first sentence of the abstract indicates that they don't mean a literal decline in pure numbers :)
* I knew personally that a regionally important corporate lobbying group was funding a ballot measure. (For the moment assume I am correct: this was not a 'secret': everyone in local politics knew this. It just wasn't 'official')
* I edited the organizations wikipedia page to state that the group was promoting the ballot measure.
The edit was reverted.
Was I going to spend time trying to track down obscure articles on local newspaper websites to prove this to wikipedia? No.
It just wasn't worth it.
I have come to the conclusion that the really interesting information is still in peoples heads.
I rarely find wikipedia turning up in my searches any more - in particular anything "controversial" or anything where there is moneyed interests involved who are interested in suppressing the other side.
I actually don't see a solution for Wikipedia. It is trying for some sort of 'objective' encyclopedia. This is well and good for questions like what is an atom in the 21st century. However, lets pretend that it is a 1901. "What is an atom?" was controversial and the back and forth that would have been interesting to know as a 1901 reader would be edited out of the 1901 version of Wikipedia.
Quite simply for me:
* something is interesting if it is "controversial" (global warming, health care, etc.)
* I am not interested in an "objective" sanitized view. I have a bias and I am not interested in hearing about the "controversy about global warming". Therefore I am biased toward sites predisposed to my perspective.
* Others have different preferred bias.
* Trying to come to a group decision where both sides are convinced that their side is correct is impossible ( in particular if one side is paid to not change their mind )
* If something is really important (i.e. job critical) I want to go to a source expert that I trust irregardless of what the 'crowd' thinks. I know what the standards I am applying to determine who I regard as the expert.
* rarely search for important things and find the wikipedia article interesting
Therefore:
* Wikipedia as a common community encyclopedia only works for the uninteresting 'factoids' that are not important
* I filter out wikipedia when looking at the google results when I am looking for something I really care about getting correct.