it seems that people enter articles into quality assurance more often than before having the flags - which at the end leads to higher quality for these articles. but i am unsure if this feeling can be better prooved somehow.
one thing seems to be a bug: with ff3 on linux i always get the flagged revision and not the most current one, even if i unchecked "show flagged revision" in the preferences.
rupert.
On Sun, Jun 8, 2008 at 2:02 PM, THURNER rupert [email protected] wrote:
it seems that people enter articles into quality assurance more often than before having the flags - which at the end leads to higher quality for these articles. but i am unsure if this feeling can be better prooved somehow.
one thing seems to be a bug: with ff3 on linux i always get the flagged revision and not the most current one, even if i unchecked "show flagged revision" in the preferences.
rupert.
On Thu, May 8, 2008 at 11:14 PM, Andre Engels [email protected] wrote:
2008/5/7 Lars Aronsson [email protected]:
Erik Moeller wrote:
In a nutshell, FlaggedRevs makes it possible to assign quality tags to individual article revisions, and to alter default views based on the available tags.
Aka hacked up a nice script that shows how many pages have been "sighted" (basic vandalism check) on the German Wikipedia: http://tools.wikimedia.de/~aka/cgi-bin/reviewcnt.cgi?lang=english
Given that FlaggedRevs has just been live for a day or so, a review rate of 4.41% is quite impressive!
Wait now. When FlaggedRevs was first mentioned, the press started to announce that censorship was being planned for Wikipedia. This was countered with the explanation that flagging was a more open regime than page locking. We no longer have to lock pages on controversial topics, because we can allow free editing as long as the non-logged-in majority gets to see the flagged/approved version.
Is it really "impressive" to have this new "soft locking" mechanism applied to a large number of pages? Wouldn't it be better to show how few pages were in need of this protection? And at the same time, to mention how many previously locked pages have now been unlocked in the name of increased openness?
No, I don't think so. Having a flag on a page is just a way of saying "this version is ok". Would it not be much better to have a version that is 'ok' for ALL pages rather than just the controversial ones? Would it really be a good thing to say "Only these few pages have versions that are okay, we have no idea about the others, but we see no reason to think they're not okay?"
-- Andre Engels, [email protected] ICQ: 6260644 -- Skype: a_engels
foundation-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
Can we get a status update on how this experiment with FlaggedRevs went on the German Wikipedia? I remember hearing they switched it off, but not why or whether they planned to try again. Anyone have details?
Nathan
On Tue, Nov 4, 2008 at 10:23 PM, Nathan [email protected] wrote:
Can we get a status update on how this experiment with FlaggedRevs went on the German Wikipedia? I remember hearing they switched it off, but not why or whether they planned to try again. Anyone have details?
de.wp community held a straw poll and it approved Flaggedrevs in the current setting.
You can find figures about the current status of flagging at http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...
Is there a review available anywhere on the particulars of the outcome? Did it reduce vandalism, were there lots of "flagging" controversies or disputes, how many articles have been flagged, what sort of criteria did they use to identify "trusted users" capable of flagging revisions, etc.?
Why was it shut down initially? What was the (detailed) outcome of the straw poll that saw it re-enabled?
thanks anyone who has these answers,
Nathan
On Tue, Nov 4, 2008 at 4:46 PM, Nathan [email protected] wrote: [snip]
how many articles have been flagged,
[snip]
Follow the link!
http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...
Oh right, that piece of data is there. Thanks! The other pieces of information are more important in determining whether it would work elsewhere, though.
Nathan
On Tue, Nov 4, 2008 at 5:57 PM, Gregory Maxwell [email protected] wrote:
On Tue, Nov 4, 2008 at 4:46 PM, Nathan [email protected] wrote: [snip]
how many articles have been flagged,
[snip]
Follow the link!
http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...http://toolserver.org/%7Eaka/cgi-bin/reviewcnt.cgi?lang=english&action=overview
foundation-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
On Tue, Nov 4, 2008 at 5:02 PM, Nathan [email protected] wrote:
Oh right, that piece of data is there. Thanks! The other pieces of information are more important in determining whether it would work elsewhere, though.
I agreed that they would be very informative. (Though you can go try to read the discussions...) But they can't determine whether it would work elsewhere. Elsewhere is not German Wikipedia. The only way to *determine* is to try it. I think the expirence on deWP will be most informative about what sorts of challenges are likely.
On Tue, Nov 4, 2008 at 6:16 PM, Gregory Maxwell [email protected] wrote:
On Tue, Nov 4, 2008 at 5:02 PM, Nathan [email protected] wrote:
Oh right, that piece of data is there. Thanks! The other pieces of information are more important in determining whether it would work elsewhere, though.
I agreed that they would be very informative. (Though you can go try to read the discussions...) But they can't determine whether it would work elsewhere. Elsewhere is not German Wikipedia. The only way to *determine* is to try it. I think the expirence on deWP will be most informative about what sorts of challenges are likely.
Can you point me to a review discussion in English?
Nathan wrote:
Oh right, that piece of data is there. Thanks! The other pieces of information are more important in determining whether it would work elsewhere, though.
Nathan
On Tue, Nov 4, 2008 at 5:57 PM, Gregory Maxwell [email protected] wrote:
On Tue, Nov 4, 2008 at 4:46 PM, Nathan [email protected] wrote: [snip]
how many articles have been flagged,
[snip]
Follow the link!
http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...http://toolserver.org/%7Eaka/cgi-bin/reviewcnt.cgi?lang=english&action=overview
Hello Nathan,
I tried to forward your question to the german mailing-list to ask an expert for the answer. But for some reason I don't understand my forward didn't come through. So here my answer.
Philipp Birken, who is a member on the board of the German Chapter and who was the organizer of the development of teh stable versions made a very detailed and interesting talk on the Wikimania in Alexandria to this topic. The talk vastly changed my opinion on the stable versions (maybe you know during my election to the board I said I was still very reserved to the stable versions thing). Philipps conclusion was that he would recommend every big language version (that is with article counts more than 100,000) use the stable versions.
To my personal opinion I think stable version is especially useful for vandal battling. If you have say 80% of your articles checked, you can see very easily which article was changed in the last night and not checked again. Especially useful is this because you are even aware of the changes that are made on articles which no one had put it on his watchlist. With a few hundred enthusiastic voluntiers (I have no doubt that the english Wikipedia has them) this is a managable task. Sometimes there were lags on the checking. For example during the Wikimania the unchecked articles numbre on the German Wikipedia increased to a few thousand. After the Wikimania the german voluntiers started a concentrated check action to reduce that number back to a managable size.
There are some experiments and talks and votes on the German Wikipedia on how the stable version should be showed to the user: Show the checked version to all users if they request an article with a hint to not checked version, show the checked version to IP-user and new user with a hint to not checked version, ans show the actual version to users with check right. Show actual version to all users with a hint to checked version and so on. I personally would always use the actual user, but on the other hand, I am not a user who ONLY search for information and don't know the mechanisms. And the votes thus far made on German Wikipedia are also only from editors with vote right. So I cannot say how an only read user see this. Academic tests on this would be interesting but as far as I know there is no such tests planned until now. On the other hand, with one exception until now I had also not heard any negative complaints about this anywhere. That one exception is on the forum of the c't magazine and I think the user is generally unhappy with the rigid quality rules on the German Wikipedia.
As far as I know the more advanced function of the stable version, the proofed version (that is, marked as proofed by an expert) is not implemented on the de-wp until now. So from my view it is a good tool for vandal battling but it still doesn't fulfill the function it is thought of: to give a random user a proofed, qualitatively garanteed article.
Beside ther de-wp there are at least two other projects that use the checked version. One is ru-wp. I didn't see any reports of the practice there, would be interesting too.
Ting
Beside ther de-wp there are at least two other projects that use the checked version. One is ru-wp. I didn't see any reports of the practice there, would be interesting too.
We are still running it in a trial mode, we started in August and at some point later the results will be evaluated, and also the inclusion of the proofed version will be discussed, right now the proofed version is not used.
By default the last version is shown, not the checked version, but one can turn on the protection, so that the checked version is shown. I believe we have only two pages protected right now, one of them is the main page. We check articles in the main space, categories, and templates.
Sysops have a patroller flag by default (meaning they can mark any version of any article as checked), unless they explicitly refuse, in addition, we have a couple of hundred patrollers who have been voted in a typically easy vote. Together with 70+ sysops, this must give about 300 patrollers, I did not check the exact number. Also, about a 100 editors have an autopatrol flag (not voted unless self-nominated), meaning if they edit a checked version the article remains checked. Bots also have this flag.
In my opinion, everything is running more or less smootly, even though some strong opinions have been expressed in the beginning, resulting in an arbitrtation case. Unless some major flaws are found, I believe the flagged revisions system will be confirmed in the course of reevaluation (note this is my private opinion).
Cheers Yaroslav
Thanks Yaroslav for the quick report. Just what I hoped :-)
Ting
Yaroslav M. Blanter wrote:
Beside ther de-wp there are at least two other projects that use the checked version. One is ru-wp. I didn't see any reports of the practice there, would be interesting too.
We are still running it in a trial mode, we started in August and at some point later the results will be evaluated, and also the inclusion of the proofed version will be discussed, right now the proofed version is not used.
By default the last version is shown, not the checked version, but one can turn on the protection, so that the checked version is shown. I believe we have only two pages protected right now, one of them is the main page. We check articles in the main space, categories, and templates.
Sysops have a patroller flag by default (meaning they can mark any version of any article as checked), unless they explicitly refuse, in addition, we have a couple of hundred patrollers who have been voted in a typically easy vote. Together with 70+ sysops, this must give about 300 patrollers, I did not check the exact number. Also, about a 100 editors have an autopatrol flag (not voted unless self-nominated), meaning if they edit a checked version the article remains checked. Bots also have this flag.
In my opinion, everything is running more or less smootly, even though some strong opinions have been expressed in the beginning, resulting in an arbitrtation case. Unless some major flaws are found, I believe the flagged revisions system will be confirmed in the course of reevaluation (note this is my private opinion).
Cheers Yaroslav
foundation-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
Yaroslav M. Blanter wrote:
Beside ther de-wp there are at least two other projects that use the checked version. One is ru-wp. I didn't see any reports of the practice there, would be interesting too.
By default the last version is shown, not the checked version, but one can turn on the protection, so that the checked version is shown. I believe we have only two pages protected right now, one of them is the main page. We check articles in the main space, categories, and templates.
Main Page used to be stabilized, but later it was protected in sysop-only mode again, because 1) once a vandalized version of Main Page got patrolled and it was displayed to all users; 2) we had a revert war on Main Page. Right now we have only one stabilized page (see [1]).
Sysops have a patroller flag by default (meaning they can mark any version of any article as checked), unless they explicitly refuse, in addition, we have a couple of hundred patrollers who have been voted in a typically easy vote. Together with 70+ sysops, this must give about 300 patrollers, I did not check the exact number. Also, about a 100 editors have an autopatrol flag (not voted unless self-nominated), meaning if they edit a checked version the article remains checked. Bots also have this flag
See [2]. We also ran a bot which read new pages patrolling log (Special:Log/patrol) and marked all pages patrolled with an old system as sighted. That's why 1st and 3rd place in most active patrollers list [3] are bots.
Here's Russian criteria for sighted revision in main namespace: 1. It must not have blatant vandalism 2. It must not have blatant BLP violations 3. It must be readable 4. It must not contain blatant mistakes (statements, which looks false even for people unfamiliar to the topic: e.g. statement that Earth has a form of pyramid) 5. It must not contain blatant copyright violations 6. It must not contain blatant spam 7. It must not be an obvious POV-fork 8. It must not contain unmarked out-of-date information 9. It must have at least one category Also, there are 3 recommendations: 1. It should not have typos 2. It should be wikified 3. It should have interwikis
Categories have to match all articles criteria + they must have parent category. Images must have a good copyright status and FU rationale, if required. Templates: 1. Must not have links via redirects (all links should point to articles themselves) 2. Must have category 3. Should have interwiki 4. Must not have fair use images 5. If template has a long documentation, it should be split to a subpage.
All editors (or patrollers, in ruwiki terms) also have a rollback flag.
1. http://ru.wikipedia.org/wiki/Special:StablePages?uselang=en 2. http://ru.wikipedia.org/wiki/Special:ValidationStatistics?uselang=en 3. http://ru.wikipedia.org/wiki/Project:%D0%A4%D0%9F#.D0.9E.D0.BA.D1.82.D1.8F.D...
--vvv
Everything is fully correct, thanks. (I forgot to mention that we also flag images).
Cheers Yaroslav
2008/11/5 Yaroslav M. Blanter [email protected]:
Everything is fully correct, thanks. (I forgot to mention that we also flag images).
How does an unflagged image included in an article work? Does it display as an empty thumbnail box with the caption, or just not display?
Enquiring minds...
2008/11/5 Yaroslav M. Blanter [email protected]:
Everything is fully correct, thanks. (I forgot to mention that we also flag images).
How does an unflagged image included in an article work? Does it display as an empty thumbnail box with the caption, or just not display?
Enquiring minds...
--
I have just checked, it is being fully displayed, I could not find any way to figure it is unflagged other than going to the image page.
Having said that, most of our images are on Commons and can not be flagged anyway.
Cheers Yaroslav
Mathias Schindler wrote:
On Tue, Nov 4, 2008 at 10:23 PM, Nathan [email protected] wrote:
Can we get a status update on how this experiment with FlaggedRevs went on the German Wikipedia? I remember hearing they switched it off, but not why or whether they planned to try again. Anyone have details?
de.wp community held a straw poll and it approved Flaggedrevs in the current setting.
You can find figures about the current status of flagging at http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...
"5025 users have the permission to review modules. The five users that set the highest number of marks during the previous hour are Mai-Sachme (51), FlaBot (45), Ma-Lik (40), Gamma9 (38) and Zollernalb (27)."
FlaBot ? Is that a bot with review access ?
Ant
Actually, this is due to someone assigning the Editor group to a bot unnecessarily. Anything in the 'bot' groups already has review permissions.
Florence Devouard-3 wrote:
Mathias Schindler wrote:
On Tue, Nov 4, 2008 at 10:23 PM, Nathan [email protected] wrote:
Can we get a status update on how this experiment with FlaggedRevs went on the German Wikipedia? I remember hearing they switched it off, but not why or whether they planned to try again. Anyone have details?
de.wp community held a straw poll and it approved Flaggedrevs in the current setting.
You can find figures about the current status of flagging at http://toolserver.org/~aka/cgi-bin/reviewcnt.cgi?lang=english&action=ove...
"5025 users have the permission to review modules. The five users that set the highest number of marks during the previous hour are Mai-Sachme (51), FlaBot (45), Ma-Lik (40), Gamma9 (38) and Zollernalb (27)."
FlaBot ? Is that a bot with review access ?
Ant
foundation-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l