On a default MediaWiki installation, this is enabled by default--to render the basic number of views a given page has received. Has this ever been enabled on Wikipedia for public view? Do the database dumps even track this on, say, en.wikipedia?
It might not be completely encyclopediac, but having this enabled and tracked could be of tremendous use to see what people are actually *looking* for. The technical overhead could be extreme from something like this, but I'm not sure. Perhaps a section of the encyclopedia with the Top 10 Pages By View for each given name space--main, main talk, Wikipedia, Image, Portal, etc., and then perhaps also broken down by certain designated categories? I.e., the Top 10 BLP articles, the Top 10 Geography Articles, and so on. Perhaps sorted by Month, with the results clearing each month, if a raw ongoing count would be too intensive?
Regards, Joe http://www.joeszilagyi.com
On 21/05/07, Joe Szilagyi szilagyi@gmail.com wrote:
It might not be completely encyclopediac, but having this enabled and tracked could be of tremendous use to see what people are actually *looking* for. The technical overhead could be extreme from something like this, but I'm not sure.
It is, unfortunately. And once you get into the caching required to deal with MediaWiki's hugely bloated arse, the numbers become meaningless.
Perhaps a section of the encyclopedia with the Top 10 Pages By View for each given name space--main, main talk, Wikipedia, Image, Portal, etc., and then perhaps also broken down by certain designated categories? I.e., the Top 10 BLP articles, the Top 10 Geography Articles, and so on. Perhaps sorted by Month, with the results clearing each month, if a raw ongoing count would be too intensive?
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
This works by taking a sample of the article hits (1 in 1000, I think) and providing a raw chart.
How closely it reflects popularity is questionable. That is, "of articles people went looking for in Wikipedia, the top 100 are ..." We don't know that this chart reflects that. So it may be of interest with quite a grain of salt.
- d.
On 5/21/07, David Gerard dgerard@gmail.com wrote:
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
This works by taking a sample of the article hits (1 in 1000, I think) and providing a raw chart.
Looking at [[MediaWiki:Common.js]], it is 1 in 6000. We don't want to overload the tools-server :)
How closely it reflects popularity is questionable. That is, "of articles people went looking for in Wikipedia, the top 100 are ..." We don't know that this chart reflects that. So it may be of interest with quite a grain of salt.
I'm sure that the script uses good statistical methods to calculate error. The accuracy is displayed right there in the list, with the margin of error. It rises fairly quickly, but the top ten articles all stay under 10%, and I think we can be fairly certain that those do, in fact, represent the most viewed pages of wikipedia.
--Oskar
On 21/05/07, Oskar Sigvardsson oskarsigvardsson@gmail.com wrote:
On 5/21/07, David Gerard dgerard@gmail.com wrote:
How closely it reflects popularity is questionable. That is, "of articles people went looking for in Wikipedia, the top 100 are ..." We don't know that this chart reflects that. So it may be of interest with quite a grain of salt.
I'm sure that the script uses good statistical methods to calculate error. The accuracy is displayed right there in the list, with the margin of error. It rises fairly quickly, but the top ten articles all stay under 10%, and I think we can be fairly certain that those do, in fact, represent the most viewed pages of wikipedia.
I'm not questioning the stats - I'm questioning that those accurately represent what people want to look up on Wikipedia. Are wikis really such a popular subject? Or do people type "wiki [subject]" into Google? How many of the porn hits are from people looking for porn and disappointed to end up at a text page?
- d.
On 5/22/07, David Gerard dgerard@gmail.com wrote:
On 21/05/07, Joe Szilagyi szilagyi@gmail.com wrote:
The technical overhead could be extreme from something like this, but I'm not sure.
It is, unfortunately. And once you get into the caching required to deal with MediaWiki's hugely bloated arse, the numbers become meaningless.
From memory, it got to the point where the servers keeping the logs
were filling their buffers faster than the hard disks could write the logs down, and so page view logging was switched off.
On 5/21/07, Joe Szilagyi szilagyi@gmail.com wrote:
It might not be completely encyclopediac, but having this enabled and tracked could be of tremendous use to see what people are actually *looking* for. The technical overhead could be extreme from something like this, but I'm not sure. Perhaps a section of the encyclopedia with the Top 10 Pages By View for each given name space--main, main talk, Wikipedia, Image, Portal, etc., and then perhaps also broken down by certain designated categories? I.e., the Top 10 BLP articles, the Top 10 Geography Articles, and so on. Perhaps sorted by Month, with the results clearing each month, if a raw ongoing count would be too intensive?
http://en.wikipedia.org/wiki/Special:Mostrevisions might be a good place to start. Except a lot of these most-edited pages are usually semi-protected, so the percentage of non-editing readers will be much higher for those, but hopefully much closer to equal for all non-protected pages.
—C.W.
The stats of the userspace are particularly interesting:
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
Jimbo's page is #63....I'd expect it to be much higher...
And there are other weird things too--a banned user is at #1, empty IP pages show up in the top 25, and my signature page ended up in there too (?)
On 5/21/07, Charlotte Webb charlottethewebb@gmail.com wrote:
On 5/21/07, Joe Szilagyi szilagyi@gmail.com wrote:
It might not be completely encyclopediac, but having this enabled and tracked could be of tremendous use to see what people are actually
*looking*
for. The technical overhead could be extreme from something like this,
but
I'm not sure. Perhaps a section of the encyclopedia with the Top 10
Pages By
View for each given name space--main, main talk, Wikipedia, Image,
Portal,
etc., and then perhaps also broken down by certain designated
categories?
I.e., the Top 10 BLP articles, the Top 10 Geography Articles, and so on. Perhaps sorted by Month, with the results clearing each month, if a raw ongoing count would be too intensive?
http://en.wikipedia.org/wiki/Special:Mostrevisions might be a good place to start. Except a lot of these most-edited pages are usually semi-protected, so the percentage of non-editing readers will be much higher for those, but hopefully much closer to equal for all non-protected pages.
—C.W.
WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l
On 22/05/07, ikiroid ikiroid@gmail.com wrote:
The stats of the userspace are particularly interesting:
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
Jimbo's page is #63....I'd expect it to be much higher...
And there are other weird things too--a banned user is at #1, empty IP pages show up in the top 25, and my signature page ended up in there too (?)
Look at the margins of error. Only the top 13 have less than +/- 100%, and everything from 13 down has the same number of hits. I would guess that number is what you get when a single hit goes through to the counter. In other words, user pages aren't viewed often enough for the sample size to be sufficient to draw meaningful statistics.
Look at the margins of error. Only the top 13 have less than +/- 100%, and everything from 14 down has the same number of hits. I would guess that number is what you get when a single hit goes through to the counter. In other words, user pages aren't viewed often enough for the sample size to be sufficient to draw meaningful statistics.
PS I've just checked the maths and can stop guessing. From 14 down each page has received one hit. 1 in 6000 hits are counted, so each hit is multiplied by 6000, and then divided by the number of days they've been counting, which is 21, and you get 6000/21=286.
Thomas Dalton wrote:
On 22/05/07, ikiroid ikiroid@gmail.com wrote:
The stats of the userspace are particularly interesting:
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
Jimbo's page is #63....I'd expect it to be much higher...
And there are other weird things too--a banned user is at #1, empty IP pages show up in the top 25, and my signature page ended up in there too (?)
Look at the margins of error. Only the top 13 have less than +/- 100%, and everything from 13 down has the same number of hits. I would guess that number is what you get when a single hit goes through to the counter. In other words, user pages aren't viewed often enough for the sample size to be sufficient to draw meaningful statistics.
Even [[User:Jesus Christ]] has a higher rating than Jimbo. Who knows what the result would be if his supporters started visiting that page more often? ;-)
Ec
On 5/22/07, Ray Saintonge saintonge@telus.net wrote:
Thomas Dalton wrote:
On 22/05/07, ikiroid ikiroid@gmail.com wrote:
The stats of the userspace are particularly interesting:
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?lang=en&wiki=...
Jimbo's page is #63....I'd expect it to be much higher...
And there are other weird things too--a banned user is at #1, empty IP pages show up in the top 25, and my signature page ended up in there too (?)
Look at the margins of error. Only the top 13 have less than +/- 100%, and everything from 13 down has the same number of hits. I would guess that number is what you get when a single hit goes through to the counter. In other words, user pages aren't viewed often enough for the sample size to be sufficient to draw meaningful statistics.
Even [[User:Jesus Christ]] has a higher rating than Jimbo. Who knows what the result would be if his supporters started visiting that page more often? ;-)
Ec
Nah, Jimbo Wales will always be more important. ~~~~