The biggest problem that wikipedia is facing right now is deterioration. I really don't find the featured article review to be effective enough to prevent featured articles from getting worse. What ideas do you guys have for how we could prevent featured article deterioration?
--Ben
On 4/3/06, Ben Greenberg bengreenb@hotmail.com wrote:
The biggest problem that wikipedia is facing right now is deterioration. I really don't find the featured article review to be effective enough to prevent featured articles from getting worse. What ideas do you guys have for how we could prevent featured article deterioration?
Huh? I thought that we had pretty much accepted that the general trend for high-profile, highly edited articles is that as time passes they tend to become higher quality. The often incredible improvement for front page featured articles is proof enough of this. Do you have a particular example in mind of this deterioration effect?
Ryan
Ben Greenberg wrote:
The biggest problem that wikipedia is facing right now is deterioration. I really don't find the featured article review to be effective enough to prevent featured articles from getting worse. What ideas do you guys have for how we could prevent featured article deterioration?
Deterioration might not be so bad. In evolutionary optimization (for example via genetic algorithms) a concept that's often mentioned is the "fitness landscape", a graph showing the relative fitness of all possible solutions to a particular problem. The landscape for most problems is usually hilly, with peaks around the solutions that are relatively good and valleys where they're particularly bad. But peaks may only be locally optimal, with higher peaks elsewhere. If an article reaches a quality peak we shouldn't be too devoted to keeping it there since somewhere else there might be an even better possible version to reach. For example, new material might be added covering some aspect of the article's topic that wasn't covered previously, but the new material has bad grammar and sparse citations and is "bad" enough to knock the article off of Featured status. The best approach in this case is not to just revert to the old version, but rather clean up the new material to result in an even better article than it was before.
http://en.wikipedia.org/wiki/Fitness_landscape#Fitness_landscapes_in_evoluti... has a nice diagram.
On 4/10/06, Bryan Derksen bryan.derksen@shaw.ca wrote:
Deterioration might not be so bad. In evolutionary optimization (for example via genetic algorithms) a concept that's often mentioned is the "fitness landscape", a graph showing the relative fitness of all possible solutions to a particular problem. The landscape for most
I'm familiar with this problem, but do you know of any detailed (if not particularly formal) analysis to see if it applies to Wikipedia, or even wiki editing in general? Not infrequently you hear people say "this article is a mess, nothing short of rewriting is required here". But I'm not convinced. Is it not possible, given an article in state T0, where state Tp is perfection, find small, realistic changes T1, T2 etc?
reach. For example, new material might be added covering some aspect of the article's topic that wasn't covered previously, but the new material has bad grammar and sparse citations and is "bad" enough to knock the article off of Featured status. The best approach in this case is not to just revert to the old version, but rather clean up the new material to result in an even better article than it was before.
In this particular instance, it seems that you could add *some* of the new material in a small enough block that the lack of citations or poor copyediting does not knock the article off its FA perch. Then, when that material has been brought up to scratch, add the next batch. Wholesale additions of poor quality material are generally quite destabilising for an article, and tend to piss off existing editors, who wonder when it will all stop...
Steve
Steve Bennett wrote:
On 4/10/06, Bryan Derksen bryan.derksen@shaw.ca wrote:
Deterioration might not be so bad. In evolutionary optimization (for example via genetic algorithms) a concept that's often mentioned is the "fitness landscape", a graph showing the relative fitness of all possible solutions to a particular problem. The landscape for most
I'm familiar with this problem, but do you know of any detailed (if not particularly formal) analysis to see if it applies to Wikipedia, or even wiki editing in general? Not infrequently you hear people say "this article is a mess, nothing short of rewriting is required here". But I'm not convinced. Is it not possible, given an article in state T0, where state Tp is perfection, find small, realistic changes T1, T2 etc?
More often than not, people will add stuff to an article, and some time later, another editor will come along and clean it up. Then some more stuff will be added, and some time later, it will get cleaned up again. Very rarely do you see an article /only/ get better (one example I *can* think of is [[Jordanhill railway station]] http://en.wikipedia.org/wiki/Jordanhill_railway_station), but that was an extraordinary case.
reach. For example, new material might be added covering some aspect of the article's topic that wasn't covered previously, but the new material has bad grammar and sparse citations and is "bad" enough to knock the article off of Featured status. The best approach in this case is not to just revert to the old version, but rather clean up the new material to result in an even better article than it was before.
In this particular instance, it seems that you could add *some* of the new material in a small enough block that the lack of citations or poor copyediting does not knock the article off its FA perch. Then, when that material has been brought up to scratch, add the next batch. Wholesale additions of poor quality material are generally quite destabilising for an article, and tend to piss off existing editors, who wonder when it will all stop...
I agree. Although the evolutionary optimization analogy works some of the time (generally up until the point the article is featured), entropy starts to take over after that point.
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
More often than not, people will add stuff to an article, and some time later, another editor will come along and clean it up. Then some more stuff will be added, and some time later, it will get cleaned up again. Very rarely do you see an article /only/ get better (one example I *can* think of is [[Jordanhill railway station]] http://en.wikipedia.org/wiki/Jordanhill_railway_station), but that was an extraordinary case.
Someone should do some analysis of Wikipedia article development cycles/processes. For some articles, I think in terms of an accordeon model. People add content, which is full of air, redundancy etc. Then more people add content in the same style. Then someone comes along and squeezes the whole lot down to its basic essence. A 500 word paragraph becomes a 200 word paragraph, almost without losing information.
Other times, I see a huge slab of prose which has become so big that people feel uncomfortable adding to it. They see that the "History of X" section of the article is so long, they shouldn't touch it any more. I simply insert subheadings by year, and suddenly they realise that the period 1950-1960 is totally blank. I don't have a name for that model.
Then other times, someone creates a stub, then someone else sees how to structure the article, and creates a whole structure full of subheadings, but with no actual text. But it actually constrains future editors, who wouldn't necessarily have used that particular structure. I came across a good example of this recently but can't find it now :/
There are probably lots of other models...a bit of analysis and we could find solutions to some of the problems, identify the models that work best etc.
Steve
Steve Bennett wrote:
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
More often than not, people will add stuff to an article, and some time later, another editor will come along and clean it up. Then some more stuff will be added, and some time later, it will get cleaned up again. Very rarely do you see an article /only/ get better (one example I *can* think of is [[Jordanhill railway station]] http://en.wikipedia.org/wiki/Jordanhill_railway_station), but that was an extraordinary case.
Someone should do some analysis of Wikipedia article development cycles/processes. For some articles, I think in terms of an accordeon model. People add content, which is full of air, redundancy etc. Then more people add content in the same style. Then someone comes along and squeezes the whole lot down to its basic essence. A 500 word paragraph becomes a 200 word paragraph, almost without losing information.
Other times, I see a huge slab of prose which has become so big that people feel uncomfortable adding to it. They see that the "History of X" section of the article is so long, they shouldn't touch it any more. I simply insert subheadings by year, and suddenly they realise that the period 1950-1960 is totally blank. I don't have a name for that model.
The wiki model? :)
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
More often than not, people will add stuff to an article, and some time later, another editor will come along and clean it up. Then some more stuff will be added, and some time later, it will get cleaned up again. Very rarely do you see an article /only/ get better (one example I *can* think of is [[Jordanhill railway station]] <http://en.wikipedia.org/wiki/Jordanhill_railway_station >), but that was an extraordinary case.
I'm really getting confused now. You just described the whole process of how articles get improved on Wikipedia, and then say that they don't only get improved. I mean, yeah; sometimes people add stuff to Wikipedia that doesn't follow the style guide / is badly written, typos etc / isn't wholly neutral / isn't referenced / etc. And then someone else comes along and makes it better. That's the whole purpose of the collaborative process. Isn't that what it means for an article to "always get better"? Isn't that what we are doing? Maybe I am missing the point here, but...
To put it in the strictest theoretical terms, an Wikipedia with zero edits is worthless. Wikipedia with infinite edits is perfect. I really believe that, mind you -- I believe that after infinite edits, Wikipedia will be perfect. Of course, we'll never get there, but each edit puts us closer. So all this talk about "deterioration" is puzzling to me.
Ryan
Ryan Delaney wrote:
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
More often than not, people will add stuff to an article, and some time later, another editor will come along and clean it up. Then some more stuff will be added, and some time later, it will get cleaned up again. Very rarely do you see an article /only/ get better (one example I *can* think of is [[Jordanhill railway station]] <http://en.wikipedia.org/wiki/Jordanhill_railway_station >), but that was an extraordinary case.
I'm really getting confused now. You just described the whole process of how articles get improved on Wikipedia, and then say that they don't only get improved. I mean, yeah; sometimes people add stuff to Wikipedia that doesn't follow the style guide / is badly written, typos etc / isn't wholly neutral / isn't referenced / etc. And then someone else comes along and makes it better. That's the whole purpose of the collaborative process. Isn't that what it means for an article to "always get better"? Isn't that what we are doing? Maybe I am missing the point here, but...
To put it in the strictest theoretical terms, an Wikipedia with zero edits is worthless. Wikipedia with infinite edits is perfect. I really believe that, mind you -- I believe that after infinite edits, Wikipedia will be perfect. Of course, we'll never get there, but each edit puts us closer. So all this talk about "deterioration" is puzzling to me.
If articles don't deteriorate, why do they get de-featured?
Deterioration happens when:
- Vandalism goes unnoticed, especially sneaky (fact-changing) or blanking-type vandalism - Random crud (aka. "cruft") gets added - Other stuff which we can't quantify yet
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
If articles don't deteriorate, why do they get de-featured?
Deterioration happens when:
- Vandalism goes unnoticed, especially sneaky (fact-changing) or
blanking-type vandalism
- Random crud (aka. "cruft") gets added
- Other stuff which we can't quantify yet
Deterioration is bad. But is it sufficiently common to be a problem? What are the relative rates of FA->non FA vs non-FA->FA transition?
Steve
----- Original Message ---- From: Alphax (Wikipedia email) alphasigmax@gmail.com
If articles don't deteriorate, why do they get de-featured?
I agree with you that articles can and do deteriorate, but another reason articles are defeatured is because the accepted standard of Featured Articles has been steadily increasing.
-- Matt
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
If articles don't deteriorate, why do they get de-featured?
Usually it's because the standard for featured articles has gone up.
Ryan
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
If articles don't deteriorate, why do they get de-featured?
Usually it's because the standard for featured articles has gone up.
Or because the article should never have been promoted in the first place. I remember [[Terrorism]] falling into this category.
-- Sam
On 4/10/06, Sam Korn smoddy@gmail.com wrote:
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:>
Usually it's because the standard for featured articles has gone up.
Or because the article should never have been promoted in the first place. I remember [[Terrorism]] falling into this category.
[[Garry Kasparov]] was a personal favorite of mine. Check out its original FAC nomination:
http://en.wikipedia.org/wiki/Wikipedia:Featured_article_candidates/Garry_Kas...
How many support votes do you see? Hmm...
Ryan
Ryan Delaney wrote:
On 4/10/06, Alphax (Wikipedia email) alphasigmax@gmail.com wrote:
If articles don't deteriorate, why do they get de-featured?
Usually it's because the standard for featured articles has gone up.
Pretty much the same is happening with the "Exzellente Artikel" from de.wp. The current standard is now much more rigid than in 2004. Last year, there was a semi-featured status "Lesenswerte Artikel" introduced.
It is the common perception that many of the first featured articles in de.wp would fail to qualify for "Lesenswerte Artikel", let alone for the current "Exzellente Artikel".
However: Some de-featured articles came to this status just because de.wp was lacking someone to point out the structural errors and problems it contained. This situation has improved as well in the last year.
The current aim of some people is to reach the (moving) target of 0,2 per cent of featured articles.
de.wp is currently 52 articles away from this moving target.
http://de.wikipedia.org/wiki/Wikipedia:Statistik_der_exzellenten_Artikel
the aim of some people is to make 0,5 per cent to get "Lesenswerte Artikel".
http://de.wikipedia.org/wiki/Wikipedia:Statistik_der_lesenswerten_Artikel
They are currently 678 articles away from that target.
Mathias
On 4/10/06, Mathias Schindler neubau@presroi.de wrote:
The current aim of some people is to reach the (moving) target of 0,2 per cent of featured articles.
I'll tell you how you can achieve that goal immediately: Lower the standard to such a degree that all articles are included.
Seriously, I don't think there is much point to such goals. Sure, they improve the encyclopedia, which is great. But people will cite statistics like "Only 0.1% of our articles are featured!" as if this means that Wikipedia sucks or something. No, it just means the standard for featuring is high, and if the gap between articles : featured articles is increasing, that just means the standard is going up. It has nothing to do with the objective quality of the encyclopedia; it's only the relative quality. That is, the relative quality between articles.
Ryan
Ryan Delaney wrote:
On 4/10/06, Mathias Schindler neubau@presroi.de wrote:
The current aim of some people is to reach the (moving) target of 0,2 per cent of featured articles.
I'll tell you how you can achieve that goal immediately: Lower the standard to such a degree that all articles are included.
There is a third option: Delete several thousand articles that are not notable for anything.
Seriously, I don't think there is much point to such goals. Sure, they improve the encyclopedia, which is great. But people will cite statistics like "Only 0.1% of our articles are featured!" as if this means that Wikipedia sucks or something. No, it just means the standard for featuring is high, and if the gap between articles : featured articles is increasing, that just means the standard is going up. It has nothing to do with the objective quality of the encyclopedia; it's only the relative quality. That is, the relative quality between articles.
Well, there are "historical" reasons why such goals can have an positive impact. In 2004, de.wp reached 100.000 articles (depending on what counts as an article). With all the media hype and more people coming in to wikipedia, it was the general perception that de.wp is now more or less "usable" for the average reader. There were at least rough information heaps on every major topic (nota bene: the more general a topic gets, the more likely it was that this was just a bunch of wikilinks, loosely linked with small sentences).
The idea was to shift the focus of attention from quantity to quality.
Yes, some people might read those statistics as "Even wikipedia says that 99,8 per cent of the articles are crap" and to some degree, one must agree: The average article is very currently poorly referenced, formatted, sourced, comprehensive and so on (this applies also to en.wp IMHO). It is no help that traditional encyclopedias have even less external references.
Alexander Bob, the CEO of Bibliographisches Institut & F.A. Brockhaus AG (publisher of the Brockhaus Enzyklopädie) said during a debate that one must not compare the 300.000 articles from the Brockhaus Enzyklopädie to the (in those days) 250.000 articles from the German Wikipedia but rather the 300k from Brockhaus to the (in those days) 400 "exzellenten Artikel". Please do :)
IMHO, if an initiative is able to draw more attention to quality related aspects, I generally support it. If an initiative no longer has this ability, it can be neglected or abolished. This is one of the reasons de.wp got rid of this {{stub}} template.
Mathias
On 4/10/06, Mathias Schindler neubau@presroi.de wrote:
Yes, some people might read those statistics as "Even wikipedia says that 99,8 per cent of the articles are crap" and to some degree, one must agree: The average article is very currently poorly referenced, formatted, sourced, comprehensive and so on (this applies also to en.wp IMHO). It is no help that traditional encyclopedias have even less external references.
You're doing it again. It's difficult to tell whether this kind of statement is pessimistic or driven toward endless improvement. But saying that the articles are crap is a subjective assertion that really says more about your attitude toward them than anything about the articles themselves. Personally, I don't agree that the average article on the English Wikipedia is crap. I think the average article on the English Wikipedia is probably decent, and likely to have more basic information about the topic in an accessible format than even a Google search would have produced. For example, if I am searching for information on something that is likely to be in Wikipedia, I'll often skip Google alltogether.
IMHO, if an initiative is able to draw more attention to quality related
aspects, I generally support it. If an initiative no longer has this ability, it can be neglected or abolished. This is one of the reasons de.wp got rid of this {{stub}} template.
Yeah, I agree with you there. But I'm sceptical about whether putting the focus on the featured article process is the best way to do this. Frequently people become obsessed with featuring the article they are currently working on and may attempt to game the political system as a way of getting a leg up on the FAC process, for example by recruiting their friends to support the article. I've also seen people object in a FAC by giving general objections and then the nominator responds that because there are no specific objections, the objection is "not actionable" and therefore invalid. When specific examples are given, the nominator corrects them but leaves many more still in the article, but expects the oppose vote to be withdrawn.
There are many other symptoms of this behavior which I think is centered around this notion that a featured article is somehow inherently better than others, or that getting an article featured is more of an achievement than making good edits. We have Wikipedians who are bragging on their userpages about their "number of featured articles" and so on. That seems very backward to me; it's editing Wikipedia for recognition, not for its own sake.
Besides, I react when people cite the number of featured articles as if it had something to do with the quality of our work. It really doesn't. Probably thousands of those articles are on subjects that just aren't interesting enough to get featured anyway. How many articles on train stations or stamps got featured? Not many. But that doesn't mean the articles are bad. Sometimes, creating a lot of short stubs is the right thing to do, in my opinion.
Ryan
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:
There are many other symptoms of this behavior which I think is centered around this notion that a featured article is somehow inherently better than others, or that getting an article featured is more of an achievement than making good edits. We have Wikipedians who are bragging on their userpages about their "number of featured articles" and so on. That seems very backward to me; it's editing Wikipedia for recognition, not for its own sake.
Is it helping the encyclopaedia get written? As long as they don't excessively game the system, draw attention away from other articles, then I would say yes. Wikipedia isn't a charity or an ideology - it's a job we have to get done, by one means or by another. If we had the funds, we would pay people to write articles. As it is, we have to motivate them however we can.
Don't you agree?
Steve
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:
Seriously, I don't think there is much point to such goals. Sure, they improve the encyclopedia, which is great. But people will cite statistics like "Only 0.1% of our articles are featured!" as if this means that
Sure, but only the ridiculously high 50% of the population with an IQ less than 100 would say that...
Steve
On 4/10/06, Steve Bennett stevage@gmail.com wrote:
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:
Seriously, I don't think there is much point to such goals. Sure, they improve the encyclopedia, which is great. But people will cite
statistics
like "Only 0.1% of our articles are featured!" as if this means that
Sure, but only the ridiculously high 50% of the population with an IQ less than 100 would say that...
You lost me.
Ryan
On 4/10/06, Ryan Delaney ryan.delaney@gmail.com wrote:
like "Only 0.1% of our articles are featured!" as if this means that
Sure, but only the ridiculously high 50% of the population with an IQ less than 100 would say that...
You lost me.
50% of the population has an IQ less than 100. An arbitrary percentage chosen much as defining an FA to be in the top 0.1% of articles would be.
Personally, I think we should stick with defining fixed criteria for FAs, and considering Wikipedia complete when 100% of articles are FAs.
Steve
On 4/10/06, Steve Bennett stevage@gmail.com wrote:
Is it helping the encyclopaedia get written? As long as they don't excessively game the system, draw attention away from other articles, then I would say yes. Wikipedia isn't a charity or an ideology - it's a job we have to get done, by one means or by another. If we had the funds, we would pay people to write articles. As it is, we have to motivate them however we can.
Don't you agree?
I don't think this is a point that is worth debating. By and large, people who make it their goal to feature a lot of articles improve Wikipedia, so they're good in my book. I'm brining up food for thought that only makes sense in the context of the discussion, not as its own meta-debate.
On 4/10/06, Steve Bennett stevage@gmail.com wrote:
50% of the population has an IQ less than 100. An arbitrary percentage chosen much as defining an FA to be in the top 0.1% of articles would be.
Personally, I think we should stick with defining fixed criteria for FAs, and considering Wikipedia complete when 100% of articles are FAs.
Well, good luck getting those criteria defined. (/not sarcasm)
Ryan
Steve Bennett wrote:
Personally, I think we should stick with defining fixed criteria for FAs, and considering Wikipedia complete when 100% of articles are FAs.
If you mean "fixed" as in "feel free to specify or improve", I agree to the first part.
If the world would stop spinning around, scientists stopped their work, wars ended and so one, Wikipedia would be complete when 100% of the relevant things and concepts in this reality would be FAs.
It could not hurt to accept that wikipedia will never be finished. Approximation to this idea by various means might serve as a goal.
Mathias
On 4/10/06, Mathias Schindler neubau@presroi.de wrote:
Personally, I think we should stick with defining fixed criteria for FAs, and considering Wikipedia complete when 100% of articles are FAs.
If you mean "fixed" as in "feel free to specify or improve", I agree to the first part.
I meant fixed as a direct contrast to constantly moving goalposts, particularly those that serve to strip former FAs of their status, when the article itself hasn't changed.
If the world would stop spinning around, scientists stopped their work, wars ended and so one, Wikipedia would be complete when 100% of the relevant things and concepts in this reality would be FAs.
There, I meant "complete" as "comprehensive". Wikipedia should of course keep evolving with the world. But necessarily there should come a point where it does not need to change as fast as it currently is. We're playing catch-up.
Steve
On 4/10/06, Steve Bennett stevage@gmail.com wrote:
Personally, I think we should stick with defining fixed criteria for FAs, and considering Wikipedia complete when 100% of articles are FAs.
Steve
It'd be a sad day if Wikipedia were complete, but even so, I don't think that FA should be a goal for any or every article. The goal should be to get every article to FA quality; whether it's been featured or not is rather irrelevant in the long run.
--keitei--
On 4/10/06, Katie (keitei) katiefromuncyc@gmail.com wrote:
It'd be a sad day if Wikipedia were complete, but even so, I don't think that FA should be a goal for any or every article. The goal should be to get every article to FA quality; whether it's been featured or not is rather irrelevant in the long run.
Maybe my terminology is confused. As I understood "featured article" means, having passed the certification, and being listed in the appropriate place. There's not much difference between that and an article that is "FA quality", is there? So, getting articles to FA quality (or getting them "featured") should certainly be the goal of every article, except possibly lists, redirects etc.
Of course, the issue of whether the articles appear on the front page is totally separate.
Steve