Wikipedia talk:Recent changes patrol/Archives/2010/February

An idea for implementing collaboration in RCP
As it was tried many years ago and re-considered more recently, many members of RCP frequently work at the same time reverting the same incident. The result is a lost of energy, an over concentration of effort against the most obvious vandalism and too much discrete vandalism surviving the filtering.

Edits being numbered, they can be regrouped in blocks of say 50 changes. New blocks would be marked "Need to be reviewed".

Bots would review them first because they can found and fix the most obvious cases more efficiently than humans.

Once a bot finished a block, the block is marked as "reviewed by 1 bot". If different bots filter the same block, increase the counter.

RCP members can next "download" or "reserve" the block and do a manual review, avoiding multiple members on the same case. When done, the block is marked as "reviewed by 1 bot and 1 RCP".

Depending of the kind of anti-vandalism a member wishes to do, he / she just have to select an appropriate block. For cleaning more obvious vandalism, choose a block that has not been reviewed by others RCP. If one prefers to work against deeper vandalism, take blocks that have been cleaned of the worst by 1 or 2 revision.

Reservation of blocks would be for short term at the beginning, few minutes, and, as more and more reviews are done, would be available for longer period because the review will take more and more time.

Such a solution would work a little bit like Distributed.Net, SetiAtHome and other distributed computing.

With such an optimization, RCP would me much more effective against all type of vandalism instead of only the most obvious. Heracles31 (talk) 16:36, 23 January 2010 (UTC)


 * Sounds good. Amandajm (talk) 00:07, 27 January 2010 (UTC)


 * This solution would work if vandalism was black and white, and your could mark a block as clean. Unfortunately you can't as different people see different types of vandalism. e.g. I might see an edit as obvious vandalism as I know the subject, that you may completely miss and vise versa. When doing av work, I often like to keep my eyes on certain articles, or accounts, I don't want them "reserved" for someone else to check if I want to check them. Also with huggle I often see what myself and other people are reverting to, and can see when it needs further investigation. I think many pairs of eyes are simply better than on pair.
 * I have noticed that the times I clash with someone else investigating longer term vandalism, e.g. reverting back edits of multiple people, is rare compared with the amount of times I do it and don't clash. This to me says that there is not really a problem here. Martin 4 5 1  (talk) 09:55, 27 January 2010 (UTC)


 * Also, the blocks would not stay static (i.e., unchanged) since the nature of Wikipedia is that it is dynamic, it is always being edited. Thus, while I would be working through my reserved block of edits, some of the wikiarticles contained therein may be edited, which would put them back to being blocks of edits that should first be reviewed by the bots. My point is that, since WP is dynamic, this system may not work. At least Huggle keeps up somewhat with the dynamic nature of editing (except that it does not seem to catch edits made by RCPers using Twinkle). Although this may be moot, since when you got to a wikiarticle in your block of edits to review, you would see the latest edits anyway. Nonetheless, this proposed system seems to preclude the use of automated tools such as Huggle and Twinkle. Or, have I misunderstood it? Thanks! —  Spike Toronto  21:01, 27 January 2010 (UTC)


 * Thanks to all of you for your advices. I recognize that such a system would not be easy to build and operate. There would be a solution for each of the points you raised (like removing an article from a block when newly edited and put it in a new one), but I agree that working around all of them would be too much work for a too small benefit. Thanks again for you point of view. Heracles31 (talk) 04:11, 2 February 2010 (UTC)