Talk:Article validation feature
Where is the code?
What to validate
I would find this feature more useful if it was used to identify specific problems instead of just passing "school notes". This could be used as a replacement for what is now realized by using all sorts of tag messages (stub notice, needs cleaning up, POV warning etc.). I would thus prefer checkoxes, something like the following:
This article is: *too short: **( ) lacking definition **( ) lacking basic information **( ) lacking specific information (say which) **( ) lacking depth **( ) ...other *too long: **( ) excessive detail **( ) trailing off-topic **( ) verbose language **( ) ...other *has other problems: **( ) incoherent **( ) lacking neutrality **( ) factually incorrect **( ) unstructured **( ) ...other
I would also prefer to have a separate page (and name space) for this, i.e. beside article and talk namespace. This should combine a display of current statistics (possibly a table listing each user's validation, user name, and date/revision link) and allow submission of a new validation in one page. Users should be able to retract from or update earlier validations (only 1 validation per user and article, retracted validations can be deleted from DB). (Of course only registered users should be able to use the feature at all). It could even be used for deletion warnings/polls, i.e.:
*should be deleted: **( ) patent nonsense/vandalism **( ) not encyclopedically relevant **( ) breaks naming/other conventions **( ) ...other
--126.96.36.199 12:36, 24 Jun 2005 (UTC)
"data for a user reputation system"
I view this as tautological. What would the user reputation system be used for, other than to help validate pages? If we're not going to restrict editing (to those of higher reputation), it really does not matter in my view whether we have a reputation system or not.
That issue, in turn, poses its own questions, but if we do, may I make two suggestions:
- Count me among those who do not wish to make reputation public.
- Use reputation-based restriction only when 'necessary' -- ie when a page has already been deemed by the community to be controversial or NPOV.
--stancollins 29 June 2005
- To help validate pages - and I think a paradigm shift might be in order: we shouldn't be thinking in terms of "good" or "bad" users. We're talking statistics here. Maybe something like "normal" or "abnormal users", that is, how close is their rating to the mean rating (not including their rating), in the aggregate? (of all articles) This can then be used as a weight on votes to help predict the mean rating. Kevin Baastalk 22:40, 2 April 2006 (UTC)
- Hmm... no, that wouldn't work. how about a vote weight that is some formula of the differences in rating from your article revisions to the revision just before it? Then people could have negative vote weights, but if there is decaying votes such that final drafts don't get "stuck", then vote weights will be biased positive, and older revisions will, in the statistical aggregate, have lower ratings. Kevin Baastalk 22:40, 2 April 2006 (UTC)
With some articles changing daily, and possibly changing from correct to completely incorrect info from one day to another, I don't feel that "rating an article" is the correct approach: We might end up with a highly rated article that is nonetheless just plain wrong. If the rating would be implemented so that one "rates a version", then the problem exists that ratings are very quickly outdated.
We can't expect people to search for the "highest rated version of an article that is still quite up-to-date" manually: A casual reader shouldn't be bothered with the concept of "article versions".
The goal has to be that every article is as valid and up-to-date as possible. I think the only solution to solve this challenge is by having a "agreed to be valid and good" view and a "most up-to-date" view. Users can vote which version is the best, and this votes can be constantly changing. To prevent vote-stuffing, we might also introduce karma and metamoderation (like with Slashdot).
Now, to prevent stiffling cooperation, the "most up-to-date view" should be the normal view. This leaves the only question of "how to present the valid and good" view. Nobody will click a "valid+good" tab on the top of each page to see this view, it's too tiresome. I see 2 solutions:
- 1. publish the "valid+good" infos on a different domain and advertise it as a "more valid and better wikipedia".
- 2. have a "valid+good" tab on top of each page, but, using cookies, keep such a selection active for as long as a user don't switch to the other option.
As an added bonus, mirror sites that only update once a month will have the option of downloading the "valid+good" version, making those mirrors more reliable, too. What does everybody think about this idea? 19:42, 1 December 2005 (UTC)
- Some questions:
- "What will stop anonymous ballot stuffing?"
- Some system of mod points and karma, to be defined more clearly
- "Ratings would need to be based on a large set of opinion, which time between versions might not allow"
- Yeah, well the idea is that an article/version has to be called "very good" by a few people to make it into the "good version tree". Whenever a different version is rated better, this new version gets promoted as the "good version". Therefore, yes, the most recent version might not be the "good" version, but this is actually ok, as with some articles, you have 2 vandals a day (see Eiffel tower for an example). The idea is to intentionally go for a "not-so-up-to-date" version that has a very high likelyhood of being correct, on the other hand. It would fix Seigenthaler incidents in the future because if an article is not yet rated, the article simply doesn't yet exist in the "good version tree" space. Alons 03:35, 6 December 2005 (UTC)
Also, instead of re-rating the complete "most up-to-date" view, it might be useful to highlight the changes between this view and the normal view and offer the reader to confirm/validate these changes. Anonymous, 08:07, 6 December 2005
- That's a cool idea! Rating the users instead of rating each individual article version could speed things up considerably. Thanks for that! :-) Alons 23:04, 6 December 2005 (UTC)
(Matthew Wedgwood, via anon edit, 6 Dec 2005) Multiple-version/view systems have serious drawbacks. Edits are the lifeblood of the Wikipedia. If two active versions are maintained, they will both also need to be edited. Who performs the edits on the "best" version? If the best version is simply a blessed snapshot of the "current" version, then a viewer who reads the "best" version and wishes to edit it will be thrown out of context into a different version. This dual-personality approach suffers the same drawbacks and overhead of maintaining versions of a website for different browsers (or different languages - perhaps a more relatable analogy). At the very least, it completely discourages the casual reader from making improvements to articles (grammar, perhaps, or even important updates).
- Yes, people will have to make a choice: "Do I look at the valid version or the up-to-date version". This allows us to have our cake and eat it too. Yes it requires the reader of the "valid" version to understand that he can't just edit this valid version, but that's something he will accept without problems - after all, he gets a better encyclopedia than before. And yes it requires that when a reader of the "valid" version wants to edit, he has to understand the concept of "there are 2 versions, valid and bleeding-edge". And to understand "we edit the bleeding edge, and when we like it, we vote it 'valid'". This is some sort of "editorial workflow" that you can't remove if you want to both have a validated and quick-to-update wikipedia, I think. But if you have an idea how to make it different and better, go ahead! :-) Alons 22:56, 6 December 2005 (UTC)
- (Just to reiterate: The idea was to have only one "trunk" - to keep wikipedia as is, and just pinpoint to certain article versions (pretty often outdated versions) that we know are good. This are the versions that can be found for every article in the "history" tab. We vote which one of all those versions is good + valid, and the version with the most votes will get copied over to the "valid+good" section, where nobody can edit them anymore. If you want to edit, edit the live version, which will show up immediately at wikipedia.org, but your edits will only show up in the new "valid+good" section if people vote for your new edits and make your edit the most "valid+good" version. Alons 22:56, 6 December 2005 (UTC))
There doesn't seem to be much consensus to use this(at least on en:). I didn't even know it had advanced beyond the theoretical. I'm sure some others are as badly informed. Can we slow this down and have some discussion first? As is, we already have templates to note serious problems in the listed areas. The kind of rating system proposed seems purposeless. It would be better people fixed the article rather than moving it down one or two points. This system seems like a way of discounting parts of Wikipedia rather than improving them. I'm not sure yet, but this may not be a good idea. Superm401 | Talk 23:39, 2 December 2005 (UTC)
I am definetly badly informed. AFAIK there is no article describing this content at en-wiki, and if not for the en:Wikipedia:Signpost reference I would not even be here. Besides, why this is called 'validating' instead of something simple like 'rating'? Is the rating going to be rested after every edit? All things considered this feature seems interesting but not that useful. --Piotrus 01:29, 4 December 2005 (UTC)
- Good points. I'm all for some sort of discussion first. Alons 15:41, 4 December 2005 (UTC)
- Agreed. This should be discussed much more rather than rushing it in. I think the current approach shown isn't solving the right problem. 188.8.131.52 09:54, 6 December 2005 (UTC)
- Good. Where do we go from here? Where do we discuss? Would somebody create a new topic and post a link here? What's the standard procedure here? Alons 23:07, 6 December 2005 (UTC)
Wiki Sort project for WP 1.0 on en
I noticed that you didn't have a link to the Wiki Sort project, the main validation/rating project on en. I have added the link, but I wanted to make sure that people working on this at Meta were fully aware of this (I know some of you are). The project does have some of its own ideas particular to the participants, but it also takes into account a distillation of many hours of discussion on the en WP 1.0 pages, so please take a look! Also, can people at Meta working on this keep us informed over at the en:Wikipedia:Version 1.0 Editorial Team informed of any new developments? This work is very exciting for us, as it should make it much easier us to put together a print/CD version of Wikipedia. Thanks, Walkerma 04:36, 5 December 2005 (UTC)
As always, a good idea is to look at how others have tried to solve similar problems in the past. As others have already suggested, slashdot is a good place to start. compare any decently commented article with a view of -1 to a view of +2. Slashdot even deals with anonymous posts and users elegantly, which is obviously something that Wikipedians need to be conscious of. Not that the slashdot form of moderation should be lifted wholesale and dropped on wikipedia, only that it's an awfully good place to start.
Accepting or Rejecting Changes
A similar idea to the "Stable" and "Testing / Bleading Edge" versions of a linux distribution like Debian. Perhaps new information is included in the article with a note describing it as such - a red border or the like. Each paragraph gets a red border when it is edited to show it's "new / unvalidated / possibly-bad information".
Two buttons to provide a simple "agree" and "disagree" functionality would allow users (and visitors) to vote on the changes, with rules such at 75% agree to accept the changes, or 75% disagree to reject the changes. These rules would take place only after enough votes had been places (eg. 100 votes).
A third button "hide" would simply replace the updated information with the previous version, while a fourth "split" would enable the two versions of the information to be viewed side-by-side. If the user is viewing the "stable" version and unaccepted changes exist, a "show" button would enable them to view the updated information.
Users could have an option in their preferences whether or not to display this "new / unvalidated / possibly-bad information". Site visitors would have this preference toggled by using the "hide" / "show" buttons embedded in the pages.
Apologies for a question surely asked before, but: will this acknowledge (admin) reverts? So that a stable, rated/validated article doesn't lose its rating by a simple attack+revert? Or is the rating kept across revisions or something? -Splash 23:16, 14 December 2005 (UTC)
I don't get it
I tried the demo but I couldn't figure out what it did. Great -- I rated a page. Now what? Can a rate an old page, or only the current one? I tried to validate old versions, but seemed unable to do so.
Also -- how do I view the "most highly rated page"? ?? Are these votes averaged somwhere? I looked at the validation history -- they're all over the map. What should I conclude? I really don't get it. Linas 22:09, 15 December 2005 (UTC)
Here is WebDSign, a (IMHO) pertinent proposal.
The existing user accounts and articles history offers a way, through some automagic analysis, to detect existing 'Wikipedia experts'.
The analysis will calculate, for each existing user, an 'efficiency score' on each category based on the volume, age, audience and stability of his writings. On each category the one-per-thousand best writers (who produce good-and-stable articles) will be immediately promoted into some 'Wikipedia expert' status. Those experts will form the category's "council", able to 'promote' other users into the 'Wikipedia expert' status.
A confirmed feature?
Has there been confirmation that some sort of article validation feature will be implemented in Wikipedia? Obviously changes will still be made in its structure, but will it still be implemented? Or is it just demos for now with just a possibility of being featured? Thanks. 184.108.40.206 00:46, 10 March 2006 (UTC)
I don't know if this is the same thing, or under the same umbrella. But some vandalism, and many careless editors, may state facts, or even quote apparent sources and citations, and do so incorrectly or even fabricating them.
Example: I'm a POV or subtle vandal. I add a citation to some paper and a plausible sounding quote, or a footnote "As identified by J.Doe in Book p.200-204". Nobody checks this. When we consider article validation, I'd like it if there was some way to say that facts have been validated.
This is a very difficult problem, because how do you tag a "fact"? I can't think of a good way. But I think its an important problem.
Maybe a simple way would be that users can vote "for" or "against" an edit? or have a way to add a count to an edit "# of users who have checked this fact" vs. "# of users who dispute this fact"? I don't know an answer, but that doesn't invalidate the question. FT2 15:16, 21 April 2006 (UTC)
A voting plugin for MediaWiki is available here : http://wiki.electorama.com/wiki/Electowidget
Does anybody have tried it ?