Magnus Manske (magnus.manske(a)web.de) [050523 01:05]: Presumably the ratings data will be released for people to analyse any way they like. There is that minor point ;-) It just occurred to me that leaving it completely open would produce interesting data we wouldn't get otherwise. - d.

On 5/22/05, Magnus Manske <magnus.manske(a)web.de> wrote: We can always collect anon votes but then exclude them from any decision making process.

On 5/22/05, Ray Saintonge <saintonge(a)telus.net> wrote: Ratings are per version not per article, so it's somewhat less likely that some one will want to change their vote. Usually editors will just vote again on a newer version.

Ray Saintonge wrote: Remember: we are doing absolutely nothing with the votes, other than logging them. So you can vote now, and vote again 10 minutes from now, and both will be recorded, and both will have no impact on anything. It is desirable that this be equally true for anons and logged in users. Notice the idea that one should be able to "change" one's vote (rather than having the date/timestamp/version recorded) is one which ought to be an _outcome_ of our study of the raw ratings data, rather than something we assume a priori about how it should work. --Jimbo

Erik Moeller (erik_moeller(a)gmx.de) [050523 06:51]: Mmmm. Possibly the results should not even be viewable for 1.5. Yes. However, articles with low ratings will be an obvious place to go for people to fix things up. I think it'd be horrible if "low ratings" suddenly became a VFD reason. - d.

Gregory Maxwell (gmaxwell(a)gmail.com) [050523 08:15]: No, no, only for 1.5 - so that the data gathering can proceed without any visible output as a running score. I anticipate the results of the data-gathering test to be available in a downloadable form for people to pick over in close detail. I'm sure someone said that about "notability", which isn't in the deletion policy either. You watch. I bet someone will try just that and someone else will think it's a dandy idea for culling Wikipedia. It provides some framework for article metadata, yes. Or at least for ideas on metadata. - d.

Gregory Maxwell (gmaxwell(a)gmail.com) [050523 09:55]: The thing is we don't know what the real usage will be until we have some data. An "About the ratings trial" link may be applicable. Then they can't claim they weren't warned, just because they didn't see the bit on the twelfth page in <font face="Flyspeck" size="-3">. I dunno. I was worried about observer effect - particularly with the live updated ratings tallies - but then, that may also be a test of just how spammy their little hearts are. Also, that way we don't have to change the current code and can bug Magnus about other aspects of it instead ;-) True. People out to do something obnoxious will find an excuse. Mmm. Those, like interwiki links, are pseudo-metadata - they work like metadata, but they're actually in the wikitext, which is not ideal in many ways. - d.

Brion Vibber (brion(a)pobox.com) [050523 11:26]: I really really would like it in for 1.5, as the Wikipedia 1.0 or stable version plans pretty much require a distributed rating system. So whatever that takes. The name is the least important thing. Article rating? "Rate this article" on the tab instead of "Validate"? - d.

On 5/22/05, Erik Moeller &lt;erik_moeller(a)gmx.de&gt; wrote: eh. They are not useful at all if not ever used as part of a decision process. I didn't suggest that we'd ever have policy that is directly dependant on them. When I decide I'm going to burn a CD of wikipedia, I'll take the data and massage it before using it to decide what I put on my CD. If you don't like my decisions, feel free to not download the ISOs I make :). Right, well we have that today. A tally can be one way of formalizing the results of a discussion. I am a card carrying member of the 'voting is not the answer' crowd. You're jumping to a conclusion no one has suggested, relax. I agree. People execute polls on wikipedia (at least the english one that I read) all the time. It hasn't killed us yet. :)

Hi, Le Sunday 22 May 2005 17:05, Magnus Manske a écrit : I think it better if only logged in users can validate articles. But well it depend what we want to do with this feature: selecting articles for an offline publication or studying psychology and sociology of Wikipedia readers ? We have a precedent: only logged in users can upload images. Yann -- http://www.non-violence.org/ | Site collaboratif sur la non-violence http://www.forget-me.net/ | Alternatives sur le Net http://fr.wikipedia.org/ | Encyclopédie libre http://www.forget-me.net/pro/ | Formations et services Linux

Yann Forget a écrit: least marks for articles Wikipedia This is not the same. Uploading images is a way to improve the quality of the project. We had to put restrictions to limit vandalism. However, I think we need feedback from readers as well. Possibly a way to go around of vandalism would be to allow voting with a valid email address on top ? Ant

Le Monday 23 May 2005 13:24, Anthere a écrit : For me this is the same. Validating articles is a way to improve the quality of the project. Which is stricter than creating an account and logging in, as we don't require a valid email for that. Yann -- http://www.non-violence.org/ | Site collaboratif sur la non-violence http://www.forget-me.net/ | Alternatives sur le Net http://fr.wikipedia.org/ | Encyclopédie libre http://www.forget-me.net/pro/ | Formations et services Linux

Anthere (anthere9(a)yahoo.com) [050524 03:58]: There's no reason not to gather the data from anons. We don't have to do a particular thing with it. - d.

Magnus Manske (magnus.manske(a)web.de) [050524 06:20]: Definitely! I can see a lot of other wikis using this feature, however it turns out ... This may not actually be much of a problem. If we have casual Wikipedia readers clicking on the "Rate this article" tab, most will be hit-and-run anyway. Possibly an explanatory note for anons when a rating is already present, if you can be bothered coding one. (People like anon NTL users - with tens or hundreds of thousands of users showing the same proxy IP - will definitely see this effect.) Suggest to them they get a login ;-) Can anyone else think of an elegant solution to this one? - d.

Magnus Manske (magnus.manske(a)web.de) [050524 18:32]: Mmm. Although anon ratings (from the vast Wikipedia readership, which is much larger than the editor base) are clearly of some interest - just what *do* the end users think of the articles? - we needn't extend to them the full service, with all bells and whistles, that logged-in users get. (Which reminds me: I need to add a "How useful was this article to you?" option to the validation suggestions.) - d.

Some ideas: A validation by everybody using points and calculating mean values presents finally the vote of the average reader. Another average reader simply learns how himself would probably vote after reading the article. Ok that's some time saving – better than nothing ;-). I think it's more interesting to focus on presenting comments of potential experts. The main problem for the reader seems to be how to find experts comments easily among others. To check the user page and perhaps some additional home page in case of a few tens of comments is cumbersome. But candidates for experts often can be recognized by their comments. At the moment the space for comments is limited, what seems to make sense. But not anybody likes to comment for any validation topic, therefore much space is lost. I like to suggest to offer only one comment window but with the possibility for several sentences, perhaps at least 4-6 comments of different users for one screen. The featured article candidacy page discussions show that often one single well founded slating review contains more valuable information than 10 supporting comments. But this requires usually more than 2 sentences. In the planned investigation phase for studying how people use a validation feature it is perhaps an idea to offer such a larger global comment window additionally. By the way I would prefer a general and more detailed point range from –3 to +3 (-3=very bad, -2=bad, -1=somewhat bad, 0=neutral, 1= somewhat good, 2=good, 3=very good). Using singed numbers instead of 1 to 7 makes any number self-explanatory. Wolfgang Beyer (Wolfgangbeyer)

Till Westermayer (till(a)tillwe.de) [050529 21:37]: Yep. Having gathering the data and processing the data as two completely separate things will be a good way to go forward, because we can build tools on top to analyse the data any way we happen to feel like. Yes. Though it'll need to be noted on the ratings page - anon users who rate things on other sites wouldn't normally expect the source of a rating to be world-visible. Something like: "Note: <b>All ratings are readable, just like article edits, with the person and time.</b> Like edits, they are part of the process of improving Wikipedia and so responsibility is attached to them. See the <a href="[whatever]">privacy policy</a>." This will discourage some anon raters, but if we don't then many people will get a nasty surprise and be quite upset. That second sentence, giving the rationale, needs work. - d.