Proposal talk:Filtering out vandalism in edit history

From Strategic Planning

If you filter out the vandalism you might as well also filter out the vandal reversion. This should potentially be automatable as you would be filtering out any edit that used Rollback or had an edit summary of rvv, and of course the edit they were reverting. WereSpielChequers 08:36, 9 August 2009 (UTC)[reply]

Fundamental objection

A basic problem of course is "what is vandalism?" In practice the term "vandalism" is used whenever something happens that is not liked by the user making the remark, especially when some pet practice or pet belief is involved. I have seen plenty of exemplary edits being marked as vandalism, just because they did not violate NOR, NPOV, V, etc.

        So, this proposal would further open the way to a 1984-type thought-police, eliminating 'wrong' ideas and facts. - Brya 05:41, 10 August 2009 (UTC)[reply]

This is where I believe a variation on Minor Edits is the solution, the only difference is that it needs some way of being edited. Users can filter edit histories of edits marked as vandalism or not, but the default will be to filter them out. Markhurd 01:17, 12 August 2009 (UTC)[reply]
"I cannot define vandalism, but I know it when I see it." - Learned Hand
If an admin, or other, decides a change is vandalism, the rollback or undo function could provide a vandalism clickbox for the undoer to click. This way the undoer, not the proposer of this function, nor the software defines what is Vandalism. A vandalism filter might be relevant for certain vandal targets, let's say en:Adolf Hitler or similar. Normally it is irrelevant, since vandalism is comparatively rare. Having a vandalism filter as an option would be a fine thing, and if an article is already very often vandalized, the editors maintaining it could explicitly by hand tag up changes afterwards as being vandalism, so that the article becomes easier to maintain. I think it would not be to hard to achieve. Rursus 07:42, 18 August 2009 (UTC)[reply]
IMO such tool would be useful even for much less visible articles than Hitler. Once you have hundreds of edits navigating through history becomes painful. The mechanism to identify vandalism can be made foolproof (several randomly selected editors would need to agree, the system will keep out those malicious). 85.70.83.93 21:46, 20 October 2009 (UTC)[reply]
As rollback is only for reverting vandalism and is taken away from those who misunderstand this, I think we could define some vandalism as that which is reverted by Rollback. However that would still leave a lot of vandalism including that which is reverted by vandal fighting bots, though the antivandal bots could also be hidden along with the vandalism that they revert. But clearly ignoring these edits would have to an option, useful for understanding the development of an article, not so useful when assessing a candidate for adminship. WereSpielChequers 22:26, 20 October 2009 (UTC).[reply]

Some simple changes to avoid the complaints and concerns

Simply make the hiding status a soft flag that any user can add or remove, and instead of making the edits completely hidden make them hidden by default and provide a button to unhide them. Rather than calling them vandalism, simply call them "hidden edits". Display their presence in the history with a "22 hidden edits (show)" button. Allow the Wikimedia communities to develop and adopt their own standards on how the soft-hiding feature can be used. I believe this alternative would achieve 99% of this proposal while avoiding most of the costs and risks. --Gmaxwell 22:09, 13 August 2009 (UTC)[reply]

I concur. Seems like a reasonable and easily-implemented solution. I hope it will be adopted. Hertz1888 07:57, 14 August 2009 (UTC)[reply]
Seems reasonable. -- MHM55 12:20, 14 August 2009 (UTC)[reply]
Absolutely. Great idea – page histories can become unusable following a series of IP vandalism and reversion. This should be set up as an opt-in though, so that the default remains 'show everything'. A simple way could be to hide edits that were completely reverted, rather than requiring edits to be flagged as 'hideable'. Rjwilmsi 21:31, 14 August 2009 (UTC)[reply]
Hear, hear! 99.56.136.133 23:59, 23 August 2009 (UTC)[reply]

'Yes, I like Gmaxwell's proposal a lot more. The original idea seems flawed as vandalism is so hard to define. The best way would be to only hide Blatant Vandalism (things that are clearly and obviously vandalism, such as blanking the page). Also, only "trusted" editors (or editors with a certain flag) should be able to mark edits as vandalism-that-can-be-hidden.--Sampi 13:20, 15 August 2009 (UTC)[reply]

Yes, and should be tecnically feasible, as a sort of sof revision deletion... Nemo 22:24, 19 August 2009 (UTC)[reply]

I support Gmaxwell's suggestion; by allowing interested users to reveal "hidden edits", it goes a long way towards avoiding the problem of editors simply hiding edits that they don't like from anyone's sight. There are still problems with the flags being placed wrongfully, but I reckon it would only be on the level of arbitrary flagging of 'vandalism' in editing generally. Tgr's suggestion of hiding bot edits would also provide a relatively easy and quick way to initially test its practicality (for instance, bot creators could flag their bot's edits to automatically be hidden by default, or to hide after a certain time period). — Sasuke Sarutobi 18:57, 29 August 2009 (UTC)[reply]

Hear, hear. I also think "soft hiding" is a solid idea with little downside.--ragesoss 23:23, 29 August 2009 (UTC)[reply]

Bot rollback

You could hide bot edits in the history (mostly not vandalism, but not interesting to most readers either) and add bot rollback rights to a lot of people. That is trivial to implement, and would filter vandalism from RC and watchlists too. --Tgr 21:56, 28 August 2009 (UTC)[reply]

Impact?

Some proposals will have massive impact on end-users, including non-editors. Some will have minimal impact. What will be the impact of this proposal on our end-users? -- Philippe 00:09, 3 September 2009 (UTC)[reply]

Proposal:Filtering_out_vandalism_in_edit_history#MOTIVATION
  • For the readers nothing will change. The editors will be able to use article history (this is almost impossible now). For people validating edits it will add a lot of boring work (although ways to reduce and spread the workload are suggested in my proposal). IMHO even fully automatic and thus very limited validation (a bot searching for all those JOHN IS GAY) would make a difference in usability of article history. 85.70.83.93 21:37, 20 October 2009 (UTC)[reply]

Other ways to address the underlying concern

Seems to me that the main concern here is the difficulty of discerning what's happened in the edit history of highly active pages. This is definitely something we've all encountered, and worth thinking about how to address.

Vandalism edits are only one component of the problem. Are there other, more general changes that could be made that address the basic concern?

This comes to mind: how about a way of browsing history that shows, for instance, every 10th revision, or every 50th revision? There could be a drop-down menu: "every revision, every 10th revision, ..."

Google Docs revision history has something like this: you can see an overview of all revisions summarized by about 8 of them, and then you can get a more detailed view within any two of those if you choose.

Just an off-the-cuff thought. Are there any other good approaches to this issue? -Peteforsyth 21:15, 3 October 2009 (UTC)[reply]

It's easy enough to hit the [500] button, the pick a span (or adjust the URL to see up to 5000 revisions). I don't see the strided history as accomplishing the same thing as is intended here. What do you do when you want to know where a particular addition was made or what a sentence originally said? In those cases doing a span-diff fails, but so would browsing by n-th revisions.--Gmaxwell 16:13, 15 October 2009 (UTC)[reply]

Watchlist

I propose to add similar filtering to watchlists.--Kozuch 15:24, 9 September 2009 (UTC)[reply]

Suggestion on how to identify vandalism

At the bottom of every edit box you have the option of checking one of two boxes, "This is a minor edit" and "Watch this page". To help identify vandalism better, include a third box that will allow you to indicate if you are reverting vandalism. The third box would only be included as an option, if you selected "undo" of a previous edit. MissionInn.Jim 15:18, 26 November 2009 (UTC)[reply]