Topic on Talk:Structured Discussions/Flow

I was looking at the archives on this subject. In a nutshell, community members felt it was intensely important that the community have control over this configuration setting, and even to my ears their arguments sounded very fuzzy, supported by minor corner cases at best. The WMF position was that being able to edit other people's posts was an alien concept for most message board users. Both sides seemed open to experimenting with it, the strong conflict was who gets to make the final decision on it. A particular comment struck me, basically saying the WMF intended to make the final decision, and that the final decision apparently was that editing would be locked off unless someone demonstrated a technical workflow blocked by that configuration.

On one side the community comments failed to give any clear reason why they felt it important for the community to have control, on the other side the WMF basically asserted dictator control and disregarded that nontechnical issues could matter.

Thinking about this, I think I have a new insight. By any reasonable standard Wikipedia should have failed. By any reasonable standard, an encyclopedia that Anyone Can Edit should have turned into a scribble board. The reason Wikipedia works has little to do with any technical details. The reason Wikipedia works is primarily sociological. If you look on any other website, people on opposite sides of significant social issues inevitably descend into endless flamewars. On Wikipedia we (mostly) avoid that. You can't achieve that by technical means, that is a purely social phenomena. We have found a way to get people to engage in collaborative work and consensus seeking.

On Wiki it is trivially easy to engage in all sorts of malicious behavior editing other people's comments. However it is extraordinarily rare for this to become a problem. There is a strong social taboo against editing anyone else's comment in any way that might be perceived as deceptive or malicious. A newbie might be shocked to discover that they have the power to edit other people's comments - but that shock is going to come with the realization that maybe it would be a bad idea to do that. It's going to make them stop and think about the power they have, and that they need to exorcize some restraint in their own behavior. It teaches people that they are expected to respect other people's comments, and maybe it helps teach them to apply more respect to the people making those comments. The fact that we can edit other people's comments sends a powerful message that we're here collaborating, that we don't engage in that sort of behavior. In the rare case that someone does abuse the power to edit other people's comments, that person gets a strong social message that that sort of behavior is Not Acceptable. Either that person gets the message and learns to collaborate (a good thing), or that person repeats the behavior and we expel them from the community. The fact that people can so easily abuse editing other's comments makes it a GREAT red flag. It's a cheap easy way to identify and expel those who are unable to collaborate productively (a good thing).

I believe the folks (don't speak for the entire community, please) have two simple use-cases -- need to sign a post accidentally submitted as an IP, and need to refactor talk pages. Admittedly an expectation of good faith from whoever edits my comment has to play some role, too.

Partly this anyone-can-edit philosophy cultivates a 'peers' and 'working all together', 'collaborating' culture. Wiki was invented as a fun way to collaborate online in the first place. Having a select group (sysops, say) who can edit others' messages (and do so as a routine) doesn't fit into such peer-to-peer directed philosophy very well.

I didn't intend to "speak for the community". When I said "In a nutshell, community members felt.." I was attempting to summarize my impression of the particular arguments I had just been looking at in that archive. I was trying to say that I could see why the WMF involved would be baffled at the intensity of the argument. I just thought my comment above might be an interesting new angle. Note that I didn't argue for anything in that post... or at least it wasn't intended to be presented as an argument. I just thought it was a striking new idea on the subject. (Or at least it was new to me)

Maybe I would have been better off presenting the thought without describing the lead up that inspired it. The archive on the subject was contentious, and I guess any description of it would tend to be provocative.

I think that "need to sign a post accidentally submitted as an IP" would impossible under Flow; what would be needed would be to (at least hide, if not delete) the initial built-in "signature" (identifier), and preferably to replace it with the user's "signature" (identifier).

I've been wondering about this recently, because at Wikimania someone mentioned to me, "often my second action at a talkpage, will be to log-in, and sign my accidentally-anonymous post."

But, surely this would be possible, as the software could just match IPs. -- If I understand correctly: the database knows what IP a logged-in user is using, but that info is only revealed to human checkusers on an as-needed basis. But is it available to the software anytime? We could (I believe) have a button on Flow topics/posts, for anyone whose current log-in matches a recently posted chunk of content from an IP.

I.e. If I accidentally make a logged-out post, and then log-in, that post would now have an extra line saying:

"Was this your post? Claim this post (transfer the contribution to your account, a logged-action), or request an Oversight action."

Maybe it could be restricted to a limited time (eg. only 2 minutes after posting).

However, I'm possibly not thinking of some obvious problem with this idea, such as how it could be misused by shared-IPs ? Not sure.

Excellent thought. If feasible, the also question remains as to who could see the old "signature". Perhaps Oversighters?

Although I tend to agree with Fram that this system is never going to work properly, it's nice to know that you (WMF) are considering some of the serious problems.

I think that a more general solution would be helpful: if your account has been logged out (at least during the last day or two–this is something that checking cookies ought to be able to identify), then pop up a large, blinking warning saying "Guess what? You got logged out when you weren't paying attention!" and give people a chance to login before saving their contributions. The oversight teams would doubtless be grateful to stop the flow of requests due to people's login sessions expiring while they're typing their messages.

Gryllida, thanks for the second paragraph edited into your post. I find it rather frustrating that the three other people replying here failed to even acknowledge the subject of this thread, and bittersweet consolation that even your reply required an edit to do so.

Edit: This reply system is unusably ambiguous. My reply to Gryllida looks like a reply to WhatamIdoing or maybe to Arthur Rubin. A whopping 8 comments in this thread and already it's gone into total failure mode with no way to tell what is replying to what.

Do I get this right? Your answer was to this post by Grydilla?

It definitely looks anything but. Could be WhatamIdoing, Arthur Rubin or Quiditty, or even the former post by Arthur Rubin or your own one, but never the one by Grydilla.

And why was I asked "Do you want to leave this page?" once I hit the "Save" button? I don't want to leave, I just want to save, for fuck sake.

Test to edit this one again. This time everything went smooth, strange.

Alsee: There will be good news coming soon, regarding Editing other people's comments.

Sänger S.G and Alsee: There will be changes at some point, for this currently-limited-to-3 number of indents. Danny wants to change it, but they can only change/fix/add/update so many elements at a time. :-/

Sänger S.G: I've been trying to reproduce that "do you want to leave" bug (bugzilla:70586), but cannot. Please could you confirm, if you're only seeing it after you follow a link from a notification (i.e. with "&fromnotif=1" in the URL)? Also, what Operating system and browser and skin, are you using? That info should help the developers narrow down the problem. Much thanks, and sorry that it's frustrating to encounter.

This is a test for coming here from notifications.

There was no such bug here, and also not after I edited this post.

For info: I'm on a MintMate 17 Live-CD with FF31

Replying to: Do I get this right? Your answer was to this post by Grydilla?

Correct, I clicked reply on Grydilla's comment, and the reply system craps out. It simply dumps new posts on the end with max indent.

P.S. It's utterly gross that I needed to type-out what post I am currently replying to.

I'm testing the indentation. This reply is to Gryllida's message (beginning with the words, "I believe the folks (don't speak for the entire community, please) have two simple use-cases").

Update: Well, that worked correctly for me. I clicked the 'Reply' immediately under Gryllida's message.

Testing. This reply seems to be getting zero indent. I'm not sure if there has been a bugfix or I somehow botched my earlier reply to Gryllida. This is marginally better, but most places to reply in this thread still leave it ambiguous who the reply is to. Any reply to Arthur Rubin, or any reply to a post in the third column, all look like they land in the same spot.

well, implementation

i think one option is to make flow + VE opt-out on mobile & app, so it will be a parallel experience. the community can stay on desktop with their old talk pages

you could also migrate flow functionality onto the notification channel. add a reply button, or discuss button for thank button. (and a pop up for opt-in VE) this would have parallel channels for discussion. this would encourage "flow" late adopters to opt-in without needing a vote. you can add thread functionality to notifications allowing sorted threads, work flows

this implementation plan would minimize the chance of AfC drama, since existing experience is unchanged, newcomers are sent to forked mobile site; they get an opt-out better interaction experience. also new devices get interaction suited to their touch screen, tablet.

so you would have editable talk on talk pages, and non-editable on notification/flow threads technical overhead is larger, since duplication of functionality.

Having parallel, disjointed, article discussion on Talk pages and Flow pages would be a problem. You'll get nothing but edit wars if two people (or two groups of people) don't see each other's comments. Trying to automatically copy Flow comments into Talk pages would be crude and fragile to breakage, and trying to automatically copy Talk comments into Flow is even worse.

well, you would be noticed in notifications / flow, and could respond on talk page. talk would feed flow. the parallel experience would be people talking on notification / flow, which talk page only users would not see. similar to the IRC experience. less disjointed than current experience. and i don't know about others, but i really don't read my talk anymore, there's nothing collaborative there, merely warning template spam.