Jury system
Appearance
A jury system is proposed for whatever Wikimedia projects have contentious issues to settle. Editors could request a "jury of their peers" for the sort of "trials" of conduct issues that have become all too popular on Wikipedia. Serious conflicts over the direction of Wikimedia projects, such as the Chechen Wikipedia, could be addressed by setting up juries a chance to decide on new admins or removing old ones to deal with a crisis.
Procedure
[edit]To select a jury:
- Begin with a random number source. This should be one or more state lotteries with public drawings and results that are archived on the Web long after the drawing. These are used to select a random post that has been made during a specified time interval prior to the drawing.
- Example: Two of my posts are [1] and [2], made 4.5 days apart, separated by 13645 revisions. Therefore, Meta oldid values increased by close to 3000 edits a day during that period. Knowing this, one can designate that the person who made a post 5370000+NNNN, a random four digit number, was active during a period of around three days around April 8. We can extend this back to 5360000+NNNN, 5350000+NNNN, etc., until the desired number of jurors is found. To find NNNN, we could agree in advance on http://nylottery.ny.gov/ , which provides archived results for a Numbers drawing held twice a day. For 4/11/13 evening this is 7-4-0, for midday it is 7-6-2. Assuming the need for a jury and the method of selection was agreed before 4/11, these numbers could be used to say that oldid 5377407, 5367407 ... etc. are the ones to be asked.
- Check eligibility. Some exclusions would be:
- Bots
- In some wikis, edits to certain namespaces, such as User: or perhaps Talk:
- Edits made shortly after other edits. To set a limit on how much increase in odds of being selected that an editor gets for editing frequently, some time is required between edits. You go to his history, find an edit made after that period of time, designate it as "live". Block out subsequent edits within that period from consideration. Then designate the next edit as "live" and so forth. This period can be five minutes (to represent editors according to volume of contributions, but recognizing that very frequent edits don't really represent more time spent) or it could be longer (to get a very representative sample of anyone interested in the project). However, a long period is troublesome to interpret when active editors are encountered
- Example: [3] is made by FuzzyBot and so is ineligible. The same is true of a large number of edits, excluding 5377414 and 18 which don't exist, until we reach [4] by User:Xuacu. Looking at his/her contributions [5] we see that with a five-minute exclusion, it is eligible. If the jury system is adopted, tools could be written to automate this tedious process, but it should always be accessible to direct checking by any interested editor.
- Solicit participation. The editor is "cold-called" with a template explaining the issue, the jury system, and asking if he/she would like to participate. A deadline, which should be one or more weeks after the notice is set for a positive response, after which an alternate will be contacted by going to another block of edits 10,000 before the last. Willing jurors then participate in a special RfC. This RfC works just like any normal RfC - anyone can participate - but only the jurors make the final close. Close can be by majority for many issues. Some decisions, such as banning an editor, might require a higher majority under future community policy.
- Role of ArbComs or other admin institutions. These should generally have some ability to act like the judge in a jury proceeding - for example, dismissing cases where basic proof is clearly lacking, or weighing in on questions of what the existing policy actually is. The degree of power they have will be determined in part by the role of the jury proceeding - for example, if it is to replace or assign admins from scratch when a project is found to be completely disrupted, they may have no say at all.