Talk:Community Tech/Add a user watchlist

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

IPs only?[edit]

Had you considered making it possible to track edits from IP editors only? This would maximise its counter-vandalism utility and reduce the scope for harassment - as a high proportion of IP edits are vandalism, and not many unregistered editors are regular contributors who stick with the same IP: Noyster (talk) 13:02, 14 February 2016 (UTC)

Oh, that's an interesting idea. I don't think that's come up before. I'll add it to our notes for this project, so we'll keep it in mind when we start working on this project. Thanks! -- DannyH (WMF) (talk) 19:07, 17 February 2016 (UTC)

Add a number of new user rights[edit]

To handle the risk of harrassment I propose the introduction of a number of new user rights, each of which can be assigned and revoked on a per user base:

Watcher rights

  • CanWatchEveryOne
  • CanWatchEveryOneNotOptedOut
  • CanWatchEveryOneOptedIn
  • CanWatchIndividualOptIn
  • CannotWatchAnyOne

User rights

  • CanOptOut
  • RequiresIndividualOptIn

By default new users, bots and admins would not have the CanOptOut right, but could get it on request. New users would have the CannotWatchAnyOne right and autoconfirmed users would have the CanWatchEveryOneNotOptedOut right.

Rights changes trigger an echo notification as well as opt-in requests and additions and deletions to and from watchlists.

--° (Gradzeichen) 13:54, 14 February 2016 (UTC)

That's a pretty complicated system, adding seven new user rights for one feature would probably get us yelled at by admins and stewards. :)
Unfortunately, if autoconfirmed users have the CanWatchEveryOneNotOptedOut right, then that's the harassment tool that people have expressed concern about. The vast majority of people on the site won't even know that the opt-out exists. On the other hand, we haven't really focused on figuring out how the UI for this feature will work yet, so I'm glad you're making suggestions. -- DannyH (WMF) (talk) 19:19, 17 February 2016 (UTC)
You are right they will not know about the opt-out. That is, until they get an Echo notification that a user wants to watch them or has started watching. This echo notification needs to contain the information which user wants to watch or has started watching and it could easily link to a page (probably located inside the preferences) that combines this functionality: 1) accept or decline options for all open watch requests 2) set opt-out and other options 3) list all watchers. As all rights can be set on a individual base, someone who uses the tool for harrasment could be revoked the right - and would loose his list of watched users, thereby triggering echo notifications to all watched users, that they are no longer watched. --° (Gradzeichen) 20:03, 17 February 2016 (UTC)
Yeah, a "you have been put on X's watchlist" notification could be a call to action for someone to opt out from being followed, if they want to. That could also be the mechanism for people to opt in -- essentially, you get a notification with a request to put you on their watchlist. In the case of a class project, for example, the professor sends out notifications to each person in the class. The students all accept the request, and then the professor has the students on the class watchlist.
To some degree, the opt-in/request/opt-out elements depend on how a person is likely to feel when they get that notification. I'm sort of thinking this through while I'm writing this response, so apologies if I'm going to go off-point a little.
A request like that is similar to a friend request on Facebook, or a connection request on LinkedIn. For users of Facebook and LinkedIn, getting a request like that can be a positive experience (hooray, new friend!) or an uncomfortable experience (why is this random stranger trying to connect with me?), depending on the circumstance and the personalities involved. The difference is that the FB/LinkedIn request is two-way -- if I accept, then you're friending me and I'm friending you. The idea we're talking about is one-way -- I'm interested in you.
So it's also similar to someone following you on Twitter. If I start following you on Twitter, it's one-way -- I'm going to see your posts, you're not necessarily going to see mine, unless you choose to follow me. Following you is automatic, you don't need to get someone's permission. But there is an opt-out -- you can choose to specifically block me from following you.
The difference is that Twitter's experience is entirely based around getting more followers. That's what Twitter is for. If nobody follows you, then you've failed, and there's no point in even posting anything. On a wiki, getting "followed" is incidental -- that's not why you're here, and it may not even occur to you that that could happen.
Let's say that I've decided to watch every new editor who joins the site, for their first 30 days. (English WP is so big that doing that is not very practical, but go with it for a second.) I use a bot that automatically sends a request to follow every new logged-in contributor after their first edit, or when they create the account. So a new person creates an account and makes an edit, and now they've got a follow request from me in their notifications. They don't know me, and they don't understand why I want to follow them.
Furthermore, let's say that I'm not the only person who uses a bot like that. There are 100 people who use that bot, so every new person immediately gets flooded with 100 follow requests from strangers. It's possible that the new person would say, "oh good, 100 people who are interested in helping new contributors learn how to edit." It's also possible that the new person would say, "why am I suddenly on 100 people's watchlists? Why are they all watching me?"
This feature doesn't help to facilitate communication or support; it's not an invitation to the Teahouse. It's a one-direction "follow" on a site that has very few social-network-style features. We don't require you to fill out a user profile on the user page. So the "I'm watching you" notification could come from someone who doesn't identify themselves at all. You just know that people are watching you.
I'm not quite sure where I'm going with this, so sorry for derailing. There's a strong possibility that this feature could be seen as "creepy" and unfriendly, and I'm trying to pin down the specific reasons for that, so we can avoid making those mistakes. I'd love to know other people's thoughts about this. -- DannyH (WMF) (talk) 20:02, 18 February 2016 (UTC)
Would really users add every new user to their watchlist? Most registrations are either visiting SUL-users or real new users. Most of both groups will never edit. If otoh you use a bot to add users after their first edit, the bot can decide based on the nature of this edit (name space, known honeypots, content of edit, area of interest, global user page, ...) who to request to be added. The number of edits of a user is public, will the number of watched users be public? The number of watchers of pages is public, if it is larger than 30, will the number of watchers of a user be public? --° (Gradzeichen) 07:54, 24 February 2016 (UTC)

Time limits to make mentoring more lightwight[edit]

As part of the research for Edit Review improvements, when thinking of mentoring-like relationships we considered also the possibility of limiting them in time. This alone won't solve all potential risks of "users tracking other users" scenario, but it may help by decreasing the level of commitment for both ends:

  • Editors being tracked may understand better the purpose (e.g., Cronopio helped me recently and I get a "do you want to allow Cronopio to follow your edits for the next week to help you" message) and would avoid getting a long list of people that they gave permission once in the past in a different context but now they would not.
  • For users mentoring this lower commitment may be also appealing. When interviewing users that were involved in helping new editors at some point but abandoned for some time mentioned comments like: "what started as a distraction from work, started to feel like an extension of work" or "I regard working on Wikipedia as a leisure activity, not a work activity".

-- Pginer-WMF (talk) 11:38, 1 September 2016 (UTC)

Hi, Pau! The Edit Review Improvements project is really interesting; I'm looking forward to seeing what you learn from the research.
We've decided that we're not going to build the user watchlist, in the form that it was proposed. Support & Safety expressed grave reservations. :) We posted the decision as a Status section on this project page.
But we're currently working on the Programs & Events Dashboard, which may be able to serve as a group mentoring tool. Our goal right now is to satisfy the editathon events use cases, but if there's an appetite for a version that helps specifically with on-wiki mentoring, then maybe it's something we could work together on. Let me know what you think... -- DannyH (WMF) (talk) 21:34, 1 September 2016 (UTC)

Give users the power to manage who can monitor them[edit]

This feature can be designed in a way that focus on positive usage. Every user can get a list of users that watch them and the ability to block individual users from watching them.

By default the process of getting watched could raise a notification that explicitly tells the user about the new person who follows them and about the ability to block that person from following. ChristianKl (talk) 17:03, 1 October 2016 (UTC)

Suggestions for a vandalism watch list[edit]

Vandalism is a serious threat to Wikipedia. It takes a large cadre of volunteers to combat it and some still slips through. The ranks of those volunteers is declining slowly. It is important to give them the tools they want and need where reasonable. I would ask that the decision not to develop a user watchlist capability be reconsidered, with a more restricted watchlist along the following lines:

  1. A user watchlist should be restricted privilege, either under new user right or the existing rollback bit. This will create an incentive for proper use as the privilege can be withdrawn and make monitoring its use easier.
  2. As suggested above, only IP address accounts and, possibly, non-auto-confirmed accounts should be eligible to be placed on a watchlist. This limitation is not based on a belief that auto confirmed users are somehow above engaging in vandalism, but on the common sense expectation that they are unlikely to log in to their accounts when deliberately behaving badly. In my experience actual vandalism from established accounts is quite rare and easily dealt with by warnings and blocks. By contrast IP vandalism often originates in shared accounts, such as schools. A block in such cases penalizes other users who may make constructive edits. A watch list placement would allow the watcher to easily deal with vandals on shared accounts while not discouraging constructive edits.
  3. User watch list entries should be time limited, say one week with some limited number of renewals allowed. In my experience, vandals do not persist long, especially if they see their efforts consistently thwarted.
  4. User should be placed on a watch list only after at least one prior vandalism warning on their talk page. A request to place a user on the watch list should require one or more parameters indicating a specific page, or pages that have been vandalized by the user.
  5. Adding a user to a watch list should result in a notice on that user’s talk page along the lines of: “Because of edits made by this account to <page link>, that have been identified as vandalism, this account has been place on a watchlist. Please refrain from further disruptive editing. If you believe your edits were not vandalism, please click <here (link to appropriate notice board)>.”
  6. It should be possible to provide a simple audit tool that would list all edits made by the watcher to pages that were also edited by the watched user during the time the watchlist entry was active, and perhaps a day or two more.

I believe a vandalism watchlist based on these principles would be helpful and unlikely to further stalking or other forms of harassment.--ArnoldReinhold (talk) 17:52, 16 November 2016 (UTC)

This task should be reopened[edit]

It is fairly easy to reformulate this as a follower/following system, with an additional watchlist. If the follower/following lists are publicly visible the cost would be to high for someone stalking another user. If it is possible to block users from following someone, then it would be even harder to use such a system for stalking. I wrote more about this on phab:T2470#3914850.

In short, I believe this task is prematurely closed, as this could be one of the most important tools to fight stalking and harassments on our projects. The reason for closing it is valid given the discussed system, but it is simply wrong if a system with a cost (that is a visible following page) is used. — Jeblad 16:15, 20 January 2018 (UTC)