Hi
look at ksh wiki
http://s23.org/wikistats/wikipedias_html.php?sort=good_desc
good: 10032 total: 508246 edits: 1135239
depth: 5509 :)
best regards
przykuta
2009/3/21 Przykuta przykuta@o2.pl:
Hi
look at ksh wiki
http://s23.org/wikistats/wikipedias_html.php?sort=good_desc
good: 10032 total: 508246 edits: 1135239
depth: 5509 :)
best regards
przykuta
zh-classical also has an unusually high depth (for a Wikipedia with a significant number of articles). Unfortunately, I don't speak either language so I'm not entirely sure what is going on, but it seems that the ksh figure may be to do with articles like this one:
http://ksh.wikipedia.org/wiki/L%C4%97%C3%9F%C3%9F_met_Affk%C3%B6%C3%B6zonge_...
It seems to be some kind of index system, there are enormous numbers of pages and redirects related to it, none of which qualify as "good articles" under the definition used on that stats page, so skew the depth figure.
Przykuta hett schreven:
Hi
look at ksh wiki
http://s23.org/wikistats/wikipedias_html.php?sort=good_desc
good: 10032 total: 508246 edits: 1135239
depth: 5509 :)
Recipes: - don't delete bad stuff, instead tag it as being bad (in the category for deletion requests, there are still open requests from May 2007 and the main active admin knows about it, cause I told him, but he doesn't care) - create masses of redirects for different spellings of this non-standardized language (up to 17,000 for a single page -> http://toolserver.org/~vvv/sulutil.php?user=ReDirBot)
Marcus Buck User:Slomox
17,000 for a single page!? Wow.
I don't think redirects should impact depth, though. If they do, we're calculating wrong I think.
2009/3/21 Marcus Buck me@marcusbuck.org:
Przykuta hett schreven:
Hi
look at ksh wiki
http://s23.org/wikistats/wikipedias_html.php?sort=good_desc
good: 10032 total: 508246 edits: 1135239
depth: 5509 :)
Recipes:
- don't delete bad stuff, instead tag it as being bad (in the category
for deletion requests, there are still open requests from May 2007 and the main active admin knows about it, cause I told him, but he doesn't care)
- create masses of redirects for different spellings of this
non-standardized language (up to 17,000 for a single page -> http://toolserver.org/~vvv/sulutil.php?user=ReDirBot)
Marcus Buck User:Slomox
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
Woa! You managed to force my email client into right-to-left mode there and I could get it back without deleting all the quoted text... Could you please set your email client to left-to-right when writing in English, please?
Redirects are counted in depth because they aren't articles. It is an extremely simplistic measure and one I completely ignore. I think the idea is that non-article pages mean there is lots of collaboration going on but actually it is edits to non-article pages that means that. Non-article pages created once and ignored increase depth while doing very little to improve the project.
My question: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=prev&am...
His answer: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=next&am...
Przykuta escribió:
Hi
look at ksh wiki
http://s23.org/wikistats/wikipedias_html.php?sort=good_desc
good: 10032 total: 508246 edits: 1135239
depth: 5509 :)
best regards
przykuta
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/23 emijrp emijrp@gmail.com:
My question: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=prev&am...
His answer: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=next&am...
Seems sensible. The only real disadvantage is that it messes up the depth statistic and I, for one, can live with that.
I think we should find a way to exclude redirs from depth stats.
Mark
2009/3/23 Thomas Dalton thomas.dalton@gmail.com:
2009/3/23 emijrp emijrp@gmail.com:
My question: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=prev&am...
His answer: http://en.wikipedia.org/w/index.php?title=User_talk:Purodha&diff=next&am...
Seems sensible. The only real disadvantage is that it messes up the depth statistic and I, for one, can live with that.
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/23 Mark Williamson node.ue@gmail.com:
I think we should find a way to exclude redirs from depth stats.
Just define "non-articles" to mean pages outside the main namespace. Lots of redirects, disambig pages and stubs does not indicate greater collaboration, so I'm not sure why there were counted to start with.
Mark Williamson hett schreven:
I think we should find a way to exclude redirs from depth stats.
Redirects _are_ a sign of depth. Well, _meaningful_ redirects of course. But there's no automatic way to distinguish meaningful and less meaningful redirects.
And that's the main problem of the whole "depth" metrics: It wants to be a measure for collaborativeness. But its counting methods are so rough and simplicistic, that inefficiency, messiness and mindlessness are pushing the depth too. Creating a 100 KB article in one edit lowers the depth, while creating a 1 KB article in 30 edits most likely will increase the depth. Creating ten useless templates or creating ten discussion pages with ditsy comments on the articles is good for the depth while ten new elaborate articles is bad for the depth. An edit war is very good for the depth while adding 100 KB text to the 100 KB article of another user adds few to the depth.
Well, in the end it's not the fault of the metrics. It's the fault of the people interpreting it as a measure of quality. It's not a measure of quality. The results can easily be skewed by individuals who have much power in a single project (Volapük, Ripuarian for example), it's always skewed for very small projects (Kanuri, Greenlandic), and it is often skewed due to the specific methods of a wiki (English Wikipedia's wikiproject ratings on almost every single discussion page for example put the depth higher). Comparing depths for different projects is almost futile, if you don't know about the specifics of the project that influence the depth.
Marcus Buck User:Slomox
To me, this indicates a problem with the metrics used to calculate depth.
skype: node.ue
2009/3/23 Marcus Buck me@marcusbuck.org:
Mark Williamson hett schreven:
I think we should find a way to exclude redirs from depth stats.
Redirects _are_ a sign of depth. Well, _meaningful_ redirects of course. But there's no automatic way to distinguish meaningful and less meaningful redirects.
And that's the main problem of the whole "depth" metrics: It wants to be a measure for collaborativeness. But its counting methods are so rough and simplicistic, that inefficiency, messiness and mindlessness are pushing the depth too. Creating a 100 KB article in one edit lowers the depth, while creating a 1 KB article in 30 edits most likely will increase the depth. Creating ten useless templates or creating ten discussion pages with ditsy comments on the articles is good for the depth while ten new elaborate articles is bad for the depth. An edit war is very good for the depth while adding 100 KB text to the 100 KB article of another user adds few to the depth.
Well, in the end it's not the fault of the metrics. It's the fault of the people interpreting it as a measure of quality. It's not a measure of quality. The results can easily be skewed by individuals who have much power in a single project (Volapük, Ripuarian for example), it's always skewed for very small projects (Kanuri, Greenlandic), and it is often skewed due to the specific methods of a wiki (English Wikipedia's wikiproject ratings on almost every single discussion page for example put the depth higher). Comparing depths for different projects is almost futile, if you don't know about the specifics of the project that influence the depth.
Marcus Buck User:Slomox
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
On Mon, Mar 23, 2009 at 12:36 PM, Mark Williamson node.ue@gmail.com wrote:
To me, this indicates a problem with the metrics used to calculate depth.
I'd say it indicates not that the depth calculations need to be tweaked, but that they are intrinsically inaccurate and not meaningful and should be disregarded altogether. They foster inter-project competition more than they accomplish any other task; I personally don't think making projects competitive (and encouraging work that does little besides increase the depth mark) is the way to go.
Nathan
There are many situations in which it could be useful to have a way to quantify the quality, rather than just number of articles, of a Wikipedia edition. If the whole formula is flawed, we should find a better one.
Mark
2009/3/23 Nathan nawrich@gmail.com:
On Mon, Mar 23, 2009 at 12:36 PM, Mark Williamson node.ue@gmail.com wrote:
To me, this indicates a problem with the metrics used to calculate depth.
I'd say it indicates not that the depth calculations need to be tweaked, but that they are intrinsically inaccurate and not meaningful and should be disregarded altogether. They foster inter-project competition more than they accomplish any other task; I personally don't think making projects competitive (and encouraging work that does little besides increase the depth mark) is the way to go.
Nathan _______________________________________________ foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/23 Mark Williamson node.ue@gmail.com:
There are many situations in which it could be useful to have a way to quantify the quality, rather than just number of articles, of a Wikipedia edition. If the whole formula is flawed, we should find a better one.
Step one: Define "quality".
If you give me an unambiguous, uncontroversial definition of quality, I'll find you a formula for it.
Дана Monday 23 March 2009 20:00:06 Thomas Dalton написа:
2009/3/23 Mark Williamson node.ue@gmail.com:
There are many situations in which it could be useful to have a way to quantify the quality, rather than just number of articles, of a Wikipedia edition. If the whole formula is flawed, we should find a better one.
Step one: Define "quality".
If you give me an unambiguous, uncontroversial definition of quality, I'll find you a formula for it.
It doesn't have to be unambiguous or uncontroversial, it only has to be useful.
Perhaps a better thing to quantify is the usefulness, rather than the quality? That is, ask the people reading and using articles how useful the article has been to them?
Or, more generally, ask them to rate articles on a scale of 1 to N, where N is e.g. 5.
By doing that, you can learn about the distribution of ratings (== quality/usefulness/???) within a wikipedia, or within a subsample of the wikipedia (e.g. "featured" or "good" content). It provides a complementary statistic to article ratings, which are generally done by editors. It also highlights articles where we as editors think we've done a good job, but perhaps readers don't. Add in the evolution of the rating with time (possibly with a half-life for an individual rating) and you get to see the direction that the article's heading in. It's a simple, unobtrusive, commonly used tool that's much more likely to be used than any type of survey, yet is direct from the users rather than being an inferred quantity.
(This isn't my idea; if I remember correctly, it's [[en:User:Majorly]]'s. I hope he doesn't mind me passing it on. I've just added my slant, and hopefully inserted it at a useful point in this discussion.)
Mike
On 23 Mar 2009, at 20:26, Nikola Smolenski wrote:
Дана Monday 23 March 2009 20:00:06 Thomas Dalton написа:
2009/3/23 Mark Williamson node.ue@gmail.com:
There are many situations in which it could be useful to have a way to quantify the quality, rather than just number of articles, of a Wikipedia edition. If the whole formula is flawed, we should find a better one.
Step one: Define "quality".
If you give me an unambiguous, uncontroversial definition of quality, I'll find you a formula for it.
It doesn't have to be unambiguous or uncontroversial, it only has to be useful.
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/23 Michael Peel email@mikepeel.net:
Perhaps a better thing to quantify is the usefulness, rather than the quality? That is, ask the people reading and using articles how useful the article has been to them?
Or, more generally, ask them to rate articles on a scale of 1 to N, where N is e.g. 5.
There are several extensions available for this:
http://www.mediawiki.org/wiki/Extension:AjaxRatingScript http://www.mediawiki.org/wiki/Extension:JSKitRating http://www.mediawiki.org/wiki/Extension:Rating http://www.mediawiki.org/wiki/Extension:Rating_Bar
If there is a consensus for it, it shouldn't be difficult to get it turn on. I'm not sure you'll get such a consensus, though... (not because it's a bad idea, but because we don't seem to be able to get a consensus on anything these days)
wikimedia-l@lists.wikimedia.org