-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I have just updated the validation feature in CVS (soon on the leuksman site, I hope...)
The main news is that you can now see the ratings and comments for a specific version of an article. These are shown as a table (topics in columns, users in rows).
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-)
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
In other news, it now tells you when it has stored your ratings, similar to saving your user options.
Magnus
Magnus Manske wrote:
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
At least I suggested it on wikide-l. And I am still suggesting it :)
In other news, it now tells you when it has stored your ratings, similar to saving your user options.
Thanks!
Mathias
Hi
Mathias Schindler a écrit:
Magnus Manske wrote:
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
At least I suggested it on wikide-l. And I am still suggesting it :)
Why do you think it could be beneficial to store/know this ?
Ant
In other news, it now tells you when it has stored your ratings, similar to saving your user options.
Thanks!
Mathias
Anthere (anthere9@yahoo.com) [050522 01:22]:
Mathias Schindler a écrit:
Magnus Manske wrote:
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
At least I suggested it on wikide-l. And I am still suggesting it :)
Why do you think it could be beneficial to store/know this ?
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
- d.
David Gerard a écrit:
Anthere (anthere9@yahoo.com) [050522 01:22]:
Mathias Schindler a écrit:
Magnus Manske wrote:
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
At least I suggested it on wikide-l. And I am still suggesting it :)
Why do you think it could be beneficial to store/know this ?
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
- d.
I totally see the benefit of seeing who has voted what on a specific article.
I am more dubious of a sort of reporting set to define how an editor is generally voting.
There are several things which could be reported about an editor
* if we report all the articles on which an editor has been voting : this allow you to follow how active and beneficial he is to the community as a "rater" (:-)); it is mostly a measurement of activity and quite the equivalent of the contribution list.
* if we give his average voting rate : What is the benefit of it ? What information does that bring to you ? Would you consider an editor is good in his ratings because he gives good notes ? Or would he be good because he gives bad notes ? He might be one who tries to focus on good articles to increase their visibility... or he might be one who tries to focus on bad articles to invite editors to improve them. The information about his average rate is not an indicator of how good he is in rating articles
* more tricky, if we give his average voting rate, and compare it to average community rate, per article... ie, the discrepancy between what people vote on average and how he perceives the quality of articles... does that mean that he is not good in doing the job... or does that mean he does not fit in the medium point of view shared by the community ? Would he be considered bad then ? (a bit like editors complaining some articles are not npov while most editors think it is ? does that mean the first is a bad editors while the others are good ones ?)
What happen if a policy develops which allow only those who make good average vote within community frame of vote...to vote ?
What I mean is that it goes further than editorial work....
Our list of contribution is not allowing any judgement on the quality of the work we provide. It is a good way to access our work mostly. And a measurement of our activity. There is no judgement.
Following the "grades" we give could lead to judgement by comparison of what is done by the rest of the community. This is much more inquisitive.
Hence my interest in asking what will be exactly displayed.
Ant
Anthere wrote:
I totally see the benefit of seeing who has voted what on a specific article.
I am more dubious of a sort of reporting set to define how an editor is generally voting.
There is no way to seperate one from another. In the end, one could extract the data from the sql-dump (if provided) or extract it from the web pages and include.
So we are actually talking about "How much frustration is obfuscation creating?" (apart from: how much work is it to implement [[Special:Validationcontribution/username]] ?)
- if we report all the articles on which an editor has been voting :
this allow you to follow how active and beneficial he is to the community as a "rater" (:-)); it is mostly a measurement of activity and quite the equivalent of the contribution list.
Right.
- if we give his average voting rate : What is the benefit of it ? What
information does that bring to you ? Would you consider an editor is good in his ratings because he gives good notes ? Or would he be good because he gives bad notes ? He might be one who tries to focus on good articles to increase their visibility... or he might be one who tries to focus on bad articles to invite editors to improve them. The information about his average rate is not an indicator of how good he is in rating articles
Detecting a systemic bias might indicate issues within the context. Let's assume one validation topic would be "foo" and user a would always rate 4/4 where the average rating of that article by all the other users were significant lower, this might indicate that there was some need for clearification. Or it might indicate that this user has a different level of standard. This is nothing bad, it would just be nice to know. This is actually a feature which does not have to be implemented by Magnus in the Code itself. Just give an SQL dump every week and someone might take the data and play with it. You never know.
Hence my interest in asking what will be exactly displayed.
I would be happy if just the raw data would be provided.
user | revision | timestamp | topic_1_rating | topic_1_comment topic_2_rating | topic_2_comment
Magnus: This data is stored in the database anyway, right?
Mathias
Anthere (anthere9@yahoo.com) [050522 01:49]:
David Gerard a écrit:
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
What happen if a policy develops which allow only those who make good average vote within community frame of vote...to vote ? What I mean is that it goes further than editorial work.... Our list of contribution is not allowing any judgement on the quality of the work we provide. It is a good way to access our work mostly. And a measurement of our activity. There is no judgement. Following the "grades" we give could lead to judgement by comparison of what is done by the rest of the community. This is much more inquisitive. Hence my interest in asking what will be exactly displayed.
The answer is, we don't know what we want to show yet - hence the idea to gather data in version 1.5, release it for analysis (and you can be sure people will analyse the data any way they can think of - per page, per editor, per aspect, etc.) and see what would make sense to do with the data and how to present it.
It's all hypothetical until we have actual data.
- d.
With the increasing amount of audio and media other than text and images being uploaded I think it could be worth considering the creation of an MfD (Media for Deletion), to undergo the deletion of other media in a similiar way to RfD, CfD and IfD. It could be useful for handling sound which may be copyrighted, and for other media uploaded inappropriately which lacks an obvious place to be noted.
If other forms of media than audio are unlikely to appear in the near future it may be a good idea to instead create AfD (Audio for Deletion).
The volume of audio appearing could rapidly increase with the introduction of WikiProject Spoken Wikipedia, meaning that the best time to create such a system is now.
On 5/21/05, David 'DJ' Hedley spyders@btinternet.com wrote:
With the increasing amount of audio and media other than text and images being uploaded I think it could be worth considering the creation of an MfD (Media for Deletion), to undergo the deletion of other media in a similiar way to RfD, CfD and IfD.
Are you talking about the English Wikipedia? If so, this should be on wikien-l rather than wikipedia-l since non-English Wikipedias don't generally have the same deletion procedures as English does.
"IfD" was actually renamed "Images and media for deletion" http://en.wikipedia.org/wiki/Wikipedia:Images_and_media_for_deletion last November so already covers audio and video. This is a fairly low traffic page compared to the main VfD, so I don't see an imminent need to split it at this stage.
Angela.
David Gerard a écrit:
Anthere (anthere9@yahoo.com) [050522 01:49]:
David Gerard a écrit:
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
What happen if a policy develops which allow only those who make good average vote within community frame of vote...to vote ? What I mean is that it goes further than editorial work.... Our list of contribution is not allowing any judgement on the quality of the work we provide. It is a good way to access our work mostly. And a measurement of our activity. There is no judgement. Following the "grades" we give could lead to judgement by comparison of what is done by the rest of the community. This is much more inquisitive. Hence my interest in asking what will be exactly displayed.
The answer is, we don't know what we want to show yet - hence the idea to gather data in version 1.5, release it for analysis (and you can be sure people will analyse the data any way they can think of - per page, per editor, per aspect, etc.) and see what would make sense to do with the data and how to present it.
It's all hypothetical until we have actual data.
- d.
Absolutely :-) That does not mean we can start to think about it :-)
Say... I can say a couple of things I would love to see myself. Out of the top of my imagination (even if I also know some of this is not a good idea for technical reasons)
* how many hits a page received. How many coming from inside wikipedia, how many from outside wikipedia
* evolution of access (for example, how many hits per month since the creation of a page)
* a quickly visible number of authors who participated to an article (how to count anon edits to be studied).
* Perhaps a ratio number of edits/age of the article
* links between categories between all languages and estimate on how many pages are categorized in each language for each category (to see for example how many french articles in the cooking category, versus other languages) ... this possibly to estimate where translations could be best to do ... so as to improve one category, rather than choosing articles rather randomly.
* ...
The development should be led by the highest needs we can perceive... not by what is made available.
Most of us agreed for this validation feature :-) I am very happy to be able to see it soon anyway :-)
Note : it seems some of the results which could be drawn out of the data you are mentionning... COULD REQUIRE an update of our privacy policy.
Cheers
Ant
Anthere (anthere9@yahoo.com) [050522 05:18]:
Note : it seems some of the results which could be drawn out of the data you are mentionning... COULD REQUIRE an update of our privacy policy.
Making all ratings publicly available for people to do stuff with surely wouldn't be any different than making all edits publicly available for people to do stuff with, which we do already (you can even download dumps of every edit ever made in a given Wikipedia). If it is, then the privacy policy would presumably need amending for every new function we add to the software.
- d.
David Gerard wrote:
Anthere (anthere9@yahoo.com) [050522 01:22]:
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
Indeed, since a key factor regarding how I would treat a rating or review is the reputation of the reviewer I think it's _necessary_ to keep track of these sorts of things. This isn't like scientific peer review where one at least has a reasonable guarantee that the reviewers know the subject matter; any J. Random Yahoo could be doing the rating in this case (I'm assuming it'll be pretty open or it won't scale well). People could also wind up rating their own work, which should be noted.
On the other hand, we don't make watchlists public so giving a list of all reviewed articles might not be a good idea. Maybe just make reviews show up in user contributions along with other edits, so recent work is theoretically available for checking but it slides off over the horizon of history after a while.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Bryan Derksen schrieb:
David Gerard wrote:
Anthere (anthere9@yahoo.com) [050522 01:22]:
It's editorial work on the project like any other. How good you are with ratings is as much part of your reputation as how good you are with edits. I don't see any reason *not* to make it as public as any other editorial work.
Indeed, since a key factor regarding how I would treat a rating or review is the reputation of the reviewer I think it's _necessary_ to keep track of these sorts of things. This isn't like scientific peer review where one at least has a reasonable guarantee that the reviewers know the subject matter; any J. Random Yahoo could be doing the rating in this case (I'm assuming it'll be pretty open or it won't scale well). People could also wind up rating their own work, which should be noted.
On the other hand, we don't make watchlists public so giving a list of all reviewed articles might not be a good idea. Maybe just make reviews show up in user contributions along with other edits, so recent work is theoretically available for checking but it slides off over the horizon of history after a while.
OK, the current code in CVS can show a list of the ratings of a user. All articles, all revisions, all topics, including comments. Users are, again, only refered to by ID than by name.
In the current state, it will only show these data for the user who asks :-)
Changing that so that everyone can see everyone's elses ratings would be a matter of changing two lines in the source...
Magnus
"Bryan Derksen" bryan.derksen@shaw.ca wrote in message news:428F7E94.1070309@shaw.ca... [snip]
Maybe just make reviews show up in user contributions along with other edits, so recent work is theoretically available for checking but it slides off over the horizon of history after a while.
This would be sensible, since a validation vote will be just as much a contribution to the content of Wikipedia (or whatever) as a normal edit.
I don't see why it should be any more easy for someone to "vandalise" articles by voting against them than by blanking or inserting rubbish.
Anthere wrote:
At least I suggested it on wikide-l. And I am still suggesting it :)
Why do you think it could be beneficial to store/know this ?
To my knowledge, the data is already stored, so your first part of the question does not make sense to me.
It is a good tradition to record user activities and to allow public review over it. These data could help to provide data for adjustments of the validation tool.
A pretty simple reason is that I am currently unable to remember which articles I already validated. So the best thing is similar to "my contributions", it would be "my validations".
Mathias
Mathias Schindler a écrit:
Anthere wrote:
At least I suggested it on wikide-l. And I am still suggesting it :)
Why do you think it could be beneficial to store/know this ?
To my knowledge, the data is already stored, so your first part of the question does not make sense to me.
I meant "stored" for visibility to editors. Sorry :-)
It is a good tradition to record user activities and to allow public review over it. These data could help to provide data for adjustments of the validation tool.
A pretty simple reason is that I am currently unable to remember which articles I already validated. So the best thing is similar to "my contributions", it would be "my validations".
Oh, absolutely. Maybe my imagination on what would be exactly displayed is going to far. My validations is very important of course.
Ant
Mathias
Mathias Schindler a écrit:
Anthere wrote:
Oh, absolutely. Maybe my imagination on what would be exactly displayed is going to far. My validations is very important of course.
My imagination is "display everything". I think that this is not going too far.
Mathias
True, true...
An example of "display everything" which I would love to see is
* where do editors landing on this page most frequently come from...
An extension of it which I might not agree with is
* See the pattern of navigation of an editor...(not his edits, the pages he goes to visit).
That might go a little too far... though I am quite certain those liking to manipulate things and spirits would love this ;-)
My best Mathias
Ant
Mathias Schindler wrote:
Anthere wrote:
Oh, absolutely. Maybe my imagination on what would be exactly displayed is going to far. My validations is very important of course.
My imagination is "display everything". I think that this is not going too far.
Seconded. I want to see who rated an article in which way and which other ratings the user did. Otherwise, the ratings are pretty useless. Seventy XX voted that the article on the armenian genocide is not NPOV - is that a reason to work on the article? Or did they so just because they are turkish nationalists who don't like the content of the article and nobody else bothered to rate because all other people saw no problems with the article?
Wikipedia as an open project needs transparency - no reason to hide data here.
greetings, elian
Hi!
On Sat, 21 May 2005 17:23:02 +0200, Anthere wrote:
Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
Why do you think it could be beneficial to store/know this ?
I'd want to see that, because I edit a lot of the sex/gender related articles, who regularly see "contributions" based on personal POVs, like "homosexuality is bad/sinfull/unhealthy/unnatural/unbiblical/younameit". Given that such contributors usually don't like our articles on these subjects very much, the articles would probably see lots of bad rating based on these POVs. It would simply be useful to see whether a user who gave for example [[homosexuality]] a bad rating has also given, for example, [[gay]], [[lesbian]], [[transgender]], [[Stonewall riots]] and so forth bad ratings or not. The former might somehow invalidate his vote in the mind of the reader.
Alex
Magnus Manske (magnus.manske@web.de) [050522 00:17]:
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-) Would a page of all ratings from a specific *user* be of general interest? I think someone suggested it, but...
Edits are public, votes in polls are public, votes for admins are public, comments are public, etc., etc. Rating is editorial work for the project; I see no reason it should *not* have names attached. This will also help spot antisocial or spamming raters, since rating only requires creating a login.
- d.
Magnus Manske wrote:
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-)
I find that very unusual from the perspective of how we do things, a step backwards from the open model.
-- brion vibber (brion @ pobox.com)
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Brion Vibber schrieb:
Magnus Manske wrote:
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-)
I find that very unusual from the perspective of how we do things, a step backwards from the open model.
Maybe I didn't make myself clear (happens:-)
I am "hiding" the user names just as a precaution, in case people don't want their ratings to be public.
Personally, showing the user names is just fine with me. I just wasn't sure that's true for everyone else, so I took the cautious way.
If there's consensus to show user names, so be it!
Magnus
Magnus Manske wrote:
Brion Vibber schrieb:
Magnus Manske wrote:
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-)
I find that very unusual from the perspective of how we do things, a step backwards from the open model.
Maybe I didn't make myself clear (happens:-)
I am "hiding" the user names just as a precaution, in case people don't want their ratings to be public.
But their ratings are *supposed* to be public. That's the whole point of the exercise!
-- brion vibber (brion @ pobox.com)
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Brion Vibber schrieb:
Magnus Manske wrote:
Brion Vibber schrieb:
Magnus Manske wrote:
I have treated the "user anonymity problem" by showing the user IDs rather than their names. So, in its current state, it *is* possible to find the user who did the rating, but it requires actual work :-)
I find that very unusual from the perspective of how we do things, a step backwards from the open model.
Maybe I didn't make myself clear (happens:-)
I am "hiding" the user names just as a precaution, in case people don't want their ratings to be public.
But their ratings are *supposed* to be public. That's the whole point of the exercise!
The ratings *are* public! The question was wether it should be public *who* rated *what*...
Magnus
Magnus Manske (magnus.manske@web.de) [050523 17:02]:
Brion Vibber schrieb:
But their ratings are *supposed* to be public. That's the whole point of the exercise!
The ratings *are* public! The question was wether it should be public *who* rated *what*...
I'd suggest they should be. When the data is released for analysis afterwards, even.
- d.
wikipedia-l@lists.wikimedia.org