Wikipedia is really decaying, from my point of view. I keep coming across pages with broken/false/outdated links, completely outdated information, blatant advertising, sneaky vandalism, and outright false information. I was looking at a famous American gunsmith's son, for example, that claimed as an American soldier he fought at the Battle of Verdun. No Americans fought at Verdun and Verdun was fought one year before the U.S. declared war on the Central Powers. The most maddening thing is trying to look at reference notes and find that no link works at all. I know Wikipedia has suffered from these problems since its inception and my observations do not make a credible sample of the project's success or failure, but I keep stumbling on terribly-done articles more now than I ever did.

Has anyone else observed this? Do we have any statistics to back this up?

I find that too, if I dare to look anything up, but statistics no. I find articles about subjects nobody has an agenda about badly vandalized - but of course nobody is watching those articles.

(Yeh, I look up obscure stuff...)

I have not observed this directly in that I haven't checked on my old watchlist for ages and rather rarely look at any articles unless they are linked to from here. However, it would not surprise me given all that I am reading about burnout and generally lower participation levels. That would also go a long way in explaining the darker mood and shorter tempers we are seeing over there.

Not aware of any newer studies either re stats.

Hard to tell ... of course, like some manic PC reseller, all they are interested in how big the numbers are, how much storage it has and how fast it is all meant to goes ... we all suffer from selective thinking.

It could well be true ... and it could well be a symptom of your own maturation. The 'Honeymoon Phase' has worn off and you are starting to notice the 'warts and all'. Proportionately though, there will always be more garbage producers than experienced writers.

Are there are notably good authors involved any more? Is there any evidence to suggest their numbers are falling off?

Next will come the phase when you become irritated that, there it is again, the Wikipedia staring at your from the top of every Google search like some jilted ex-lover turned stalker.

And why is there a batch of handguns at the top of the topic on Confirmation bias?

The articles seem as mediocre as ever.

The fact is, they've never really had the manpower to maintain articles against subtle, gradual degradation. It isn't to the point where most of it would be easily apparent to a non-expert in any given topic, but it'll get there eventually.

They'll probably have enough "RC Patrollers" for the immediate future at least, because the "OMG Eric is a Fag"-type vandalism doesn't require actual fact checking or research talent. They can just hand that off to the kids like they always have, gives them something to do... But the subtle stuff, they just don't have enough people, they never have, and the numbers are dwindling.

For example, there are 565 results for the term "later this year" (with quotes), and I'd guess that at least 60 percent of those were posted prior to this year and are now incorrect non-specific date references, just from nobody having done anything whatsoever. True, that's only about 340 articles, but hey, it's only one phrase.

Dead external links are also a huge maintenance issue - I'd guess that at least 80 percent of WP articles have external links and that at any time, about 10 percent of them are dead. Who really wants to go through a million articles (i.e., I'm not counting stubs, disambiguations, etc.; the number is probably lower than 1 million) and click on all those links? Some of them are probably malware-spreaders by now.

Maintenance is a management and retention problem, whereas new-article development is more of a recruitment tool. They're good at recruitment, but they've always sucked at management and retention - and that's assuming you can say they even have any management, in any real sense of the term.

One of the problems they will have is link degradation, where other sites move stuff around due to standard maintenance.

We can then look forward to the problems of controversial edits supported by missing links that when the new link is found doesn't support the assertion. Unless someone took a shapshot, they can't prove their assertion. Given the aversion to dead trees, over time, that is a lot of maintenance.

Of course anyone running a web site that is used by Wikipedia as source for plagiarising (sorry, Mr. O, sourcing) should be using some dynamic delivery scheme just to mess with them.

The complement of the problem of crappy product is the problem of crappy learning.

The product of WMF-sponsored sites is not limited to goal of educating the public. The process of participating in WMF-sponsored sites is also supposed to be a constructive educational process for the participants.

But I have copious evidence to demonstrate that precious little learning is taking place among the most active players of the game. And what little gains in insight can be documented, they appear to come grudgingly and at a snail's pace. I reckon that Socrates, Plato, and Aristotle would be appalled at the sorry state of the Academy.

This is why having Good and Featured Articles is good. You have a reference point, so a year later you can easily compare versions and cull dross and keep any improvements.

Sure, we have academic evidence that the problem was getting worse and worse between 2002 and 2006 -- no reason to believe that curve ever reversed itself. See Figure 7 in the original paper, and you may also be interested in reading my interpretation of this study and how it clearly underestimated the magnitude of the problem.

Any more recent evidence of degradation would be useful. As I mentioned in another thread, this is exactly the sort of thing to put into a letter to donors. Donors aren't interested in spats and arguments and drama or administrative evil, they are interested in whether their money will help bring the sum of all human knowledge to every person on the planet - or not.

We need conclusive proof that

* The quality of genuinely encyclopedic content was never that high in the first place.
* It is actually getting worse

I think the reasons are quite obvious - the administrative system favours "vandal fighters" and is entirely hostile to subject-matter experts. But this is a cause which is secondary to its effect, for which evidence is needed.

The problem is that this sort of research can't be done by a "bot", so the only way to carry it out would be to find someone with far too much time on their hands and/or a grad student with funding to perform the research.

Not that the researcher would find out anything most of us here don't know already, of course.

Of course, this statement has a major flaw. It assumes that Good and Featured Articles are particularly fine ones when these honours are bestowed. Thus they are benchmarks for assessing future versions. Anyone who's ever been involved in the processes may well doubt that.



I'll second that. The GA and FA systems aren't too conducive to experts either.

Ottava and Durova were both big on the FA stuff. I imagine their replacements probably come from the same mold.