# Wikipedia talk:Flow

The following WMF employee is active on this page:

## Vanilla idea

I know that this is extremely vanilla, but should we activate it on the Signpost or WikiProject newsletters for commenting? It would be similar to existing systems of social participation/engagement. Discuss-Dubious (t/c) 16:44, 17 May 2016 (UTC)

No. Wikipedia is an encyclopedia where the purpose is to edit articles. BTW, those watching this page may like to know that the rumors of Flow's death have apparently been exaggerated as a page at nowiki that I watched has recently been converted to Flow, and I get excited notifications here, whenever a change happens there. Johnuniq (talk) 23:44, 17 May 2016 (UTC)
I don't really understand your objection. I don't understand how a comment system affecting a few pages like this would go against the purpose of the encyclopedia anymore than they already are by simply existing. Are you saying it would eat up more time than they already do? Can you draw the line connecting these points for me? Discuss-Dubious (t/c) 01:08, 18 May 2016 (UTC)
Why would anyone able to edit articles want to use a different system to edit a comment? How do you use a link/template/whatever? How do you see an old revision or revert spam? The current system makes it exactly the same as every other edit! If people can't chat using wikitext they are welcome to use Facebook or whatever interests them. Some improvements to how MediaWiki operates on talk pages might be good, but throwing it all out and creating a new system is pointless. Johnuniq (talk) 03:54, 18 May 2016 (UTC)
I had no idea what your leanings were. I agree with the reasons myself. I'm not saying that we should use it on talk pages. I'm saying that if they want to use it, then VG newsletters and the Signpost are a reasonable place, but not if it is a slippery slope to all articles. Frankly, I prefer the current method of editing talk pages and agree with your concerns. Discuss-Dubious (t/c) 16:26, 18 May 2016 (UTC)
I don't really have skin in the game (to be brutally honest I don't particularly like flow though I've certainly used it many places and find it usable) there are definetly some good reasons people could want a different system. The reality is that MediaWiki itself is designed to write Wikipedia articles and long form pages in general. It's not designed for conversation or chat. We've hacked together systems such as indention that many of us are used to but it's still not designed for that. Flow (for both it's good and bad qualities) IS designed for that and I can completely understand someone wanting to use 1 system (wikitext) for what it's designed for (articles) and 1 system (flow) for what it's designed for (conversation). James of UR (talk) 06:35, 18 May 2016 (UTC)
I honestly understand and agree with his reasoning, but I figured, if they really want this, they can put it where it can do least harm and feel most natural. Discuss-Dubious (t/c) 17:43, 18 May 2016 (UTC)
Putting Flow on a few pages where "it can do the least harm" is a rather poor case for actually doing it, chuckle. There's an inherent downside to having one system for 99.999% of all pages, and a second system for 0.0001% of pages. It's a very poor "compromise" just for the sake of offering some token compromise. The WMF has zero interest in deploying Flow on 0.0001% of pages.
The big problem is that Flow is intended to eliminate all our talk pages, Flow is designed to be a chat board, and it's utterly utterly unsuitable to wiki work. Flow is a preview of their plans for article pages as well. Flow doesn't actually use wikitext. It lets you put in wikitext, it literally throws your wikitext away and translates it into something else, and whenever you ask to see "the wikitext" it invents new, fictional, mangled, wikitext to show you. The entire underlying system is a god-awful kludge and it's broken as hell. If you save and re-edit, it translates back to invents new (often mangled) fictional wikitext. Merely clicking back and forth with the preview button can mangle what you typed in, and clicking preview and back a second time can re-mangle it even further. It can delete tags from your wikitext and mangle the rendered page. That's how broken the system is. The translator is called Parasoid, and WMF literally plans to do the same thing for article pages. The vision is move everything off wikitext, to make Visual Editor the editor, to make Flow-chat-boards the talk pages, and only to maintain fictional wikitext support via the Parasoid translator. If you edit an article in wikitext it will throw the wikitext away, and when you try to look at it again it invents new, mangled, fictional wikitext for the page. I've run into so many Flow bugs that I just plain stopped reporting them. There's literally negative value in diverting valuable programmer-hours continue working on it. Alsee (talk) 04:00, 19 May 2016 (UTC)
I see. I'll take your words for it. You make a good case for throwing the idea out entirely. Discuss-Dubious (t/c) 02:04, 22 May 2016 (UTC)

## Archiving flow page

https://www.mediawiki.org/w/index.php?title=Talk%3AUniversal_Language_Selector%2FDesign%2FInterlanguage_links%2FArchive_1&type=revision&diff=2146526&oldid=2140172 These archiving actions triggered the mediaiwiki notification as if there're something new happened on the page and made it feel like users have edited an archived page if you look at contribution history of users who have leave their comments in the page before. Can that be improved?C933103 (talk) 11:10, 24 May 2016 (UTC)

## Flow satisfaction survey

I'm just letting interested parties know that the WMF is planning to run a satisfaction survey on Flow soon. I'll probably post the link here when it's released, but here is a preview. Alsee (talk) 15:57, 22 July 2016 (UTC)

User:Alsee Can you ping me when it is released? Thanks Doc James (talk · contribs · email) 14:07, 22 August 2016 (UTC)
I'll be interested in knowing whether there will be a difference between the large wikis such as ours and the small ones. Also, Doc James, I think you got your sig wrong? Jo-Jo Eumerus (talk, contributions) 08:09, 22 August 2016 (UTC)
Thanks one tidel to many. Doc James (talk · contribs · email) 14:07, 22 August 2016 (UTC)
Doc James, and everyone, I'll definitely let you know when the survey goes live. I'm keeping an eye on MWF's task for it. It was supposed to go out months ago but progress has been extremely slow. I'll probably post it at Village Pump and even add it to Central Notice. I'm severely concerned that their current plans for selective survey invitations seem inadvertently designed to bias the results as hard as possible. They plan to post invitations on Flow board only, and to send survey invitations to people's talk pages... but only to people who have replaced their talk-page with Flow.
Hi! I noticed you installed a wine cellar in your house! Would you mind answering our survey? Do you prefer beer or wine?

Alsee (talk) 04:31, 24 August 2016 (UTC)
You totally called it.  — Scott talk 20:34, 7 September 2016 (UTC)
More like
Hi! I noticed you've used steam automobiles in the past, and are now trying out electric automobiles. Would you mind answering our survey on how you'd like electric automobiles to evolve in the coming years?

Quiddity (WMF) (talk) 02:06, 10 September 2016 (UTC)
That perfectly represents how the WMF and probably a dozen others see it. For most people, it is
Hi! Bicycles have too many wheels, so we invented the unicycle. Aren't we brilliant?

And then send that survey only to circuses... Fram (talk) 05:16, 11 September 2016 (UTC)
It looks like the survey should be released in the next few days. It was bumped up to High priority, they have compiled a list of pages to send the invitation to, and there was a comment targeting yesterday as the release date. Alsee (talk) 09:19, 7 September 2016 (UTC)
Note: survey announcement moved into a new section. This will make it easier to link to it as a fresh discussion, without the text in this section being a distracting lead-in. And after I did it, it suddenly struck me as ironic how easily and thoughtlessly I did it in Wikitext. All I did was insert a new section heading above it. Flow's "Structured Discussions" are an iron cage. It gives you zero power or freedom to make unusual edits like that. Alsee (talk) 14:14, 8 September 2016 (UTC)

## Flow survey released

Survey now closed. Thanks for all the feedback. Will update once the results are available. Quiddity (WMF) (talk) 22:16, 21 September 2016 (UTC)

The survey has been released. @Doc James, Jo-Jo Eumerus, and Alsee:. The full announcement is:

Hello!
Like some other community members, you are using Flow.
An increasing number of communities now use Flow or are considering it. Although Flow itself is not scheduled for major development during 2016 fiscal year, the Collaboration Team remains interested in the project and in providing an improved system for structured discussions.
You can help us make decisions about the way forward in this area by sharing your thoughts about Flow — what works, doesn't work or should be improved?
Please fill out this survey, which is administered by a third-party service. It will not require an email or your username. See our privacy statement.
Thanks for your ideas and opinions about Flow!

HTH, Quiddity (WMF) (talk) 17:59, 7 September 2016 (UTC)

• For the record, I've stated that the pointlessly different parser/markup and the difficulty of performing structured discussions are the key problems. Incidentally, the latter is one of the two reasons why I am wondering about small wiki/large wiki differences - the smaller projects are less likely to have issues with these as they tend not to have as much structure. The other reason being that they generally are younger and thus don't have as much entrenchment. Jo-Jo Eumerus (talk, contributions) 21:50, 7 September 2016 (UTC)
• Done the survey, despite some useless or leading questions (I don't want Flow, so I skipped the "order by importance" question: does that mean that the default order will be counted as my preference? Probably, but in reality I have no preference, which I couldn't indicate). Fram (talk) 06:40, 8 September 2016 (UTC)
@Fram, answers are recorded only if you click on the "next" button. If you haven't finished the survey, your results will be incomplete, but your previous answers will count for the final results. Trizek (WMF) (talk) 15:48, 9 September 2016 (UTC)
Thanks for the explanation, but that's not a solution of course. I can't skip a question which has no importance for me, and I don't think I had any indication of how many questions were still coming (nor of the fact that my answers until then would be counted), so I could hardly stop the survey then and there. Basically, you will have no way of knowing how many people really answered a question, and how many just skipped it, making the survey results rather meaningless (never mind the heavily biased way the survey has been posted to the remaining Flow users). Fram (talk) 15:55, 9 September 2016 (UTC)
• It's beyond belief that the WMF built Flow without genuine wikitext support. The simulated wikitext is a bottomless bug-hole, and it's aggravating how often it mangles wikitext that I type in. Flow's "Structured Discussions" turn into unreadable spaghetti when more than about three people are involved. It doesn't have the power or flexibility of talk pages. If the WMF puts up chatboards the WP:NOTFORUM/WP:NOTHERE disruption would be catastrophic. I was working on a bio where a partisan-political website with significant traffic linked directly to a BLP's talk page. The site sent probably hundreds of angry POV-Warriors on an explicit mission to attack the bio and attack Wikipedia. They would have spread everywhere if we had chatboards. And that was just ONE inbound link. I don't see any prospect that Flow could be upgraded into a viable replacement for talk pages. Alsee (talk) 20:08, 8 September 2016 (UTC)
Huh - why do you think that Flow would trigger WP:NOTFORUM/WP:NOTHERE disruption, other than the one that already happens on regular talk pages anyway? Jo-Jo Eumerus (talk, contributions) 20:14, 8 September 2016 (UTC)
For starters, that's exactly what happened at Wikinews when LiquidThreads was installed along side Talk pages. They refer to it as "trollspace". Neo-nazis and other random internet commenters rant about the stories. If it looks like a typical internet chatboard then that's how people will use it. Anyone who edits an article has no problem editing Talk pages, but a generic internet-commenter with no interest in editing is likely to see Talk, shrug, and move on to conventional chatboards. Alsee (talk) 20:58, 8 September 2016 (UTC)
I'll need to check that out. Not sure about "no problem", though. JoJo Eumerus mobile (talk) 21:44, 8 September 2016 (UTC)
, Alsee's referring to Wikinews, which has used LiquidThreads for reader comments (not for discussing article improvements) for years. I'm sure that nobody means to imply that Wikipedia editors would behave like random commenters on news websites. Whatamidoing (WMF) (talk) 12:41, 10 September 2016 (UTC)
Alsee Just for future reference, my username has a hyphen between the two "Jo"s. One wonders why that doesn't happen with current talkpages - maybe because the WMF is actually right and the current talk page system is a piece of junk that discourages posting in it? It would fit the pattern I've seen on Commons, with changes that facilitate the use of upload tools (e.g mobile upload or the VE upload interface) caused a flood of bad uploads. That said, the difference in average upload quality before and after the changes didn't seem all that large, certainly not as much as between current talk pages and the scenario you are describing. I am also not convinced that Flow is the same thing as a chatboard except for having a similar thread structure. Jo-Jo Eumerus (talk, contributions) 19:29, 12 September 2016 (UTC)
• FWIW, if I were involved in closing a discussion about Flow on the English Wikipedia, I wouldn't be relying on results from a third-party survey. That's not how we do things. - Dank (push to talk) 20:35, 8 September 2016 (UTC)
The WMF doesn't have much interest in RFC's for the community to make decisions on stuff. They want to start up work on Flow again, and this is a survey to help them decide what parts of Flow to prioritize. It does contain one question on whether people prefer Flow or wikitext, and there is a free-text area we can comment in. In theory a sufficiently negative response might persuade them to reconsider the Flow plan. However the results are almost certainly going to be inflated. They explicitly WP:CANVASSED survey invitations onto the talk pages of people who opted-in to Flow on their User_Talk. The intent was not an unbiased survey on Flow itself, the intent was to find out what Flow-users want upgraded. Alsee (talk) 20:58, 8 September 2016 (UTC)
Also, I am fairly sure a survey is not meant to make binding decisions. JoJo Eumerus mobile (talk) 21:44, 8 September 2016 (UTC)
• I just filled out the survey, and I found it pretty easy to communicate the things I feel negatively about (which is most of it). I agree that the survey comes across as biased in its design, but that did not prevent me from marking the negative things as negative, and using a comment field to point out the trollspace issues. I encourage other editors to respond to the survey, and to do it on one's own terms, without accepting any implied assumptions. --Tryptofish (talk) 21:47, 8 September 2016 (UTC)
• Done, Told them I couldn't find any useful aspect of Flow... which is true, You'd think the WMF would give up on this lost cause and simply create something that we'd all like, Sure you can't please anyone and everyone but I wouldn't be surprised if Flow only pleases about 5-10% of editors on this entire project. –Davey2010Talk 23:18, 8 September 2016 (UTC)
• Done. But one comment: I believe that Flow is an attempt to solve a problem that doesn't exist. If newcomers are confused with wikitext-based talk pages, then build something over it, not nuke the whole thing out of existence. Flow provides no useful tool: it just separates Wikipedia into a content edition and a bulletin board system. In fact, the WMF has already done what I've suggested: we have VisualEditor for the newbies. The WMF software department, or even a volunteer programmer, can just add some features into VisualEditor that simplify talk page editing. And at the reciprocal of umpteen the cost! 12 years have been wasted on trying to replace talk pages: see LiquidThreads, another doomed project. What happened to learning from history? 00:06, 9 September 2016 (UTC)
<squeeze> I've asked aboiut this on the talkpage Flow forum impersonation of the VE at Mediawiki, and got as an answer, that it's impossible to use VE on talk pages, and I should shut the fuck up about this issue. The thread was closed several times by WMFers, who didn't want to listen. Feel free, to start your own thread there, or post something in my thread as well. I wonder, how such a squeeze would work in this Flow thingy ;) Grüße vom Sänger ♫ (talk) 11:58, 9 September 2016 (UTC)
If you want to experiment, I've copied this talk page to my sandbox, where Visual Editor is enabled. ;-) Diego (talk) 12:33, 9 September 2016 (UTC)
It is easy to port VE to talk pages: add an indent feature, remove the features that are only for "content". If VE was programmed with even the smallest whit of focus on abstraction and modularity, it would take at most a month to port it to talk pages. 20:36, 9 September 2016 (UTC)
If you could please enlighten those, who vehemently deny this possibility in this thread, and tried to close it with flimsy denials about the usage of VE anywhere on talk pages, I'd be grateful. It looks more and more that it's not implemented, because it would work against the pet project Flow. Grüße vom Sänger ♫ (talk) 10:06, 11 September 2016 (UTC)
I think your position would held more weight if we actually built something over wikitext-based talk pages. I've drafted a proposal to activate the "Join this discussion" widget used by at Wikipedia:Teahouse/Questions to make using talk pages easier. I think I will start that RfC when the Flow survey results are published, depending on how they go. Diego (talk) 08:46, 9 September 2016 (UTC)
Off-topic for here, but I would probably support that RfC with the option that it needs to be disabled or disable-able (is that a word?) in preferences for logged-in editors (it should be default on for all newbies and IPs of course). The WMF sometimes seems to forget that there are other (and probably better) ways of making our talk environment more accessible than just building something new from scratch, which they have now tried twice (at least). Fram (talk) 08:55, 9 September 2016 (UTC)
As WP:VPR#Regularise spacing between paragraphs on talk pages (and below) notes, a historical quirk means that we're abusing the HTML "description list" feature by using ::::'s to indent. This is opaque to screenreaders and hard to understand for many sighted people. The "rules" of how it works have to be intuited via observation (or waiting for someone to correct any mistakes). Yes I'm completely used to it, as are most editors, but it's not ideal, and I don't think it will be enshrined as such by baking it into VE. Look at this discussion for example, and the random mix of * and : that have been used above, and my own comment below which is ambiguous as to whether I'm indenting (I am) or other people are interleaving replies (they might, but I request they don't).
There are a number of things that it is utterly unfeasible to do with wikitext talkpages at a large scale, such as
• watchlisting individual topics
(a few Wikipedias maintain individual subpages for every single discussion topic (e.g. da:Wikipedia:Landsbybrønden), or even every day and every topic (e.g. it:Wikipedia:Bar) on their village pumps, for this kind of functionality. But AFAIK no wiki has done so throughout all namespaces, because it makes it hard to keep track of new discussions.)
• or connecting discussion topics to maintenance tags so that in-article problems can be discussed in-situ with direct context, at will
ibid
• or moving individual topics or comments from one location to another, whilst keeping attribution intact
(no, it hasn't been implemented in Flow yet, but in wikitext it can only be accomplished by cut&paste moves).
Some sort of a "structured discussion & workflow system" where each topic and comment and piece of metadata is independent, but logically connected, is necessary for any kind of powerful enhancement (that is done with any kind of reliability/automation). (It could be argued that we "just" need to create a few (hundred) bots that keep track of all subpages, and section-title changes, and page-moves, and etc. But bots are hard to develop/maintain across 280+ languages that have individual wikitext-talkpage informal rulesets.) It doesn't have to look or work like Flow currently does (which is part of the point of the survey), but it would have to be "structured".
Just like we don't use plain wikitext pages for the contents of Wikidata, and can't reliably use them for MassMessage or other uses of JSON, it similarly isn't ideal to use them for the complexity of discussions and !voting and tagging and bot-manipulation and more that occur on our discussion pages. We're used to it, but it's not ideal. Quiddity (WMF) (talk) 02:02, 10 September 2016 (UTC)
If MediaWiki has to use bodged indents, that is a problem with MediaWiki itself, and shows that the code for indenting needs a rewrite. The visually impaired population have waited 12 years for a solution. Perhaps the MediaWiki wikitext-to-HTML parser and translator also needs one. But again, build something over wikitext:
• watchlisting individual topics: the section is the basic unit of discussion. It is easy: make the watchlist seek individual sections that users have watched. It will not cripple the WMF's servers, as long a table is maintained storing the sections watched by users, sorted by the title name and the section name combined in order. If the software detects a section change is changed, search through the wiki's database and change the name using the same algorithm. SQL maintains a B-tree, so searches should be fast, worst-case ${\displaystyle O(log_{\frac {m}{2}}n\log m)}$, where ${\displaystyle m}$ is the order of the tree. (TAOCP §6.2.4).
• Section linking. Easier to update every wiki's maintenance tags to support section linking than to spend 12 years trying to fix the problem.
• A note in the edit summary is sufficient for attribution, I believe.
Flow actually makes it harder to keep track of large discussions. Go on any bulletin board system and see how large informal conferences on one thread are handled using even a mature forum system. Wikitext is very space efficient and long individual reply chains can be sustained with occasional {{Od}} to glue things together, although not optimally (building something over wikitext would be helpful, again). Not to mention that unless Flow is optimized correctly, loading busy talk pages would lag readers with older computers. Not to mention a nightmare to read. 03:21, 10 September 2016 (UTC)
If memory serves, #1 would require an unacceptable amount of work to implement and has thus been declined in Phabricator a couple of times. And Wikitext talk pages lead to TL;DR much faster than a bulletin board thread, at least for me. Jo-Jo Eumerus (talk, contributions) 09:29, 10 September 2016 (UTC)
It is amusing for the WMF to suggest that watchlisting sections would take "too much work". For 1% of the cost of Flow, they could have implemented section watchlisting and a pile of other enhancements. If I have a page watchlisted, how about highlighting new text? That way I don't have to read it from the DIFF portion of of watchlist's changes-since-last-visit link? They have a perverse interest in not working on that sort of stuff because it undermines the primary goal of replacing Talk pages with Flow. (I'm not saying they are malicious about it, but they just don't perceive value in upgrading something they expect to be gone "soon".) Regarding TL;DR, Flow's spaghetti message threading rapidly becomes unreadable, and the WMF have been dead-stubborn against endless complaints about Flow's whitespace problem. I did a test copying a reasonably typical Talk discussion into Flow. Flow has ~40% of the density of a Talk page. That means a discussion that is ten screens tall in WikitextTalk will be around twenty-five screens tall in Flow. Alsee (talk) 13:25, 10 September 2016 (UTC)
The higher density of wikitext is the reason why TL;DR happens more quickly there, more whitespace is not always a bad thing. The Phab task for section watchlisting is phab:T2738, as an aside. Jo-Jo Eumerus (talk, contributions) 14:41, 10 September 2016 (UTC)
If section watchlisting works only 98-99% of the time, then it is a net positive. Of course edge cases will have to be handled:
• If the section is moved (except to an archive), then the editor moving the section has the responsibility to ping all participants.
• If the section becomes "dead" and archived, then the space efficiency of section watchlisting may decrease. The solution is easy: if there are no edits to a section in n days, then strike it off the watchlist.
• If the section is renamed trivially, then rename the instances in the database.
• If the section is split for some reason, then the editor doing the split should have a very good reason, making this rare. In these rare cases, the editor doing the split should also ping all participants.
And with little impact to WMF servers. Let's assume that editors are watching 10 million sections across all wikis. We will assume that an average entry in the database costs 200 characters. Let's assume even further that the Unicode characters take the full four bytes to store in the database. Then 4 times 200 times 10 million equals eight gigabytes. You can literally buy a flash drive for about the price of a few cups of coffee that can store all our sections. And not to mention that we can compress the string, to 30-50 percent, such that it only takes four to six gigabytes. The Internet Archive can store historical copies of almost every major website, not to mention historical books, with about $5 million in assets, so the WMF can store four gigabytes with$82.7 million in assets. 01:17, 11 September 2016 (UTC)
There is only one valid response to comments like this, and that is, that this is opensource software that you can contribute to. We definitely need more hands to do this kind of stuff and you clearly qualify and should be able to dive straight in. —TheDJ (talkcontribs) 12:03, 11 September 2016 (UTC)
• I most strongly concur with all Alsee's comments. I don't think I've ever seen (except work by my 1st year students on communication science or communication psychology courses) a less professionally compiled survey. It's just full of leading questions designed to solicit responses that are positive towards Flow, and there are no options to skip totally irrelevant questions. In fact the list of glaring errors in the concept and design of this survey is too long to mention here. There is also the absolute proven fact (a 2011 discussion is on meta) that the results of surveys of this kind are presented to the community in a manner that they reflect what the WMF wanted to hear. It is possible that the largely new Foundation staff may have a higher level of moral and financial responsibility towards the stakeholders but this remains to be seen. There are far more serious issues to be resolved than even thinking of creating some other form of talk page communications. I mean big issues that affect the very reputation and quality of Wikipedia - and whether or not Quiddity (WMF) is personally responsible for prioritising the various development projects, he certainly is in no doubt whatsoever as to what I am alluding too here. Perhaps in fact, these are issues for which the community should now probably put direct pressure on the new CEO and the Board of Trustees to look into. Instead of Flow, the WMF staff should be looking into means to allow the volunteers to be more actively engaged in the overall concept and development of te encyclopedia. Kudpung กุดผึ้ง (talk) 03:38, 10 September 2016 (UTC)
• I had a long and futile discussion about this with Trizek here as I started to translate it. I first tried to correct all this faults an errors in the German translation, and tried to convince him to get them more truthful and less biased. But he insisted on his bulls*** and I simply sett all my translations back, as I was not ready to deceive the German readers with such utter nonsense translated. Grüße vom Sänger ♫ (talk) 18:46, 10 September 2016 (UTC)
• WTF WMF? Why are we even considering bringing back something that (a) wasn't that bright of an idea to begin with (b) didn't even work properly and (c) was resoundingly rejected by the community? In what stygian depth of utter madness does this even come close to resembling a good idea? Andrew Lenahan - Starblind 01:58, 11 September 2016 (UTC)
• You've obviously never worked at a big software company. Standard operating procedure is the suits decide what to do, and then order their underlings to find justifications for said decisions. Everyone needs to "ship" "deliverables" regularly so they can add bullet points to their performance reviews and resumes. What the victims users feel about said "deliverables" aren't listed on such things, so no one cares. The WMF has been pretty transparent about wanting to be a Silicon Valley tech company, so none of this should be surprising. Coming soon: "We've removed the headphone jack Talk pages. They're outdated and you didn't want them anyway." The WMF (or at least the people at the WMF calling the shots), in true management form, views all editors as interchangeable widgets, so even if people leave en masse due to Flow or some other WMF fiasco, they figure they can just spend money on recruitment drives, hence the lack of concern over the dismal relationship between the WMF and community and the constant focus on growing the number of active editors. --47.138.165.200 (talk) 10:02, 11 September 2016 (UTC)
• I'm very interested in the comments above, about watchlisting sections while still using wikitext. I would think that such a system would only be needed for WP: and WT: pages. And I also would suggest that it be implemented only at the level of "Level 2" headers (the section headers with two = signs on each side of the title), and not for subsections. There is little need to watchlist only subsections of a discussion. Would that simplify some of the issues involved in writing the software? --Tryptofish (talk) 01:00, 12 September 2016 (UTC)
That would be consistent with Flow, as in Flow is as well only the whole topic is watchable. In Flow it doesn't matter to differentiate further, as there are no substructures possible. Grüße vom Sänger ♫ (talk) 04:27, 12 September 2016 (UTC)
No, that would not simplify and of the issues involved in writing such software. —TheDJ (talkcontribs) 07:51, 12 September 2016 (UTC)
Thanks, both of you, for the replies. I realize that it would be consistent with Flow, but my point is that most of us don't want the very thing that Flow offers in that regard: that absence of sections, replaced by what looks like a chatroom. I see, above, that editors disagree about whether it is practical to watchlist sections in the existing wikitext, and I think that the community would welcome that one feature. It probably would make more sense to put a lot of effort into making it work, instead of putting the effort into Flow. --Tryptofish (talk) 20:51, 12 September 2016 (UTC)
I've been a bit sarcastic, imho it's one of the severe failings of this forum impersonation gimmick Flow that it's not better structurable. But this way none of the WMF-fanboys could argue, that it should be any deeper then their stuff goes at all. You can’t even watch the whole talk page at once, you'll only get a message for new started threads, and if you want to watch longer, you have to watch any single thread by itself. OK, You won't be bothered by stuff, you don't want to, but there's currently no option to watch the complete talk page with one click under Flow. Grüße vom Sänger ♫ (talk) 14:24, 13 September 2016 (UTC)
That's interesting, thanks. So really, what editors would find most useful is the ability to choose either to watch the entire page, or to watch individual sections. And Flow is no better at offering that choice than the existing wikitext is. In fact, that's all the more reason to seriously consider working this feature into wikitext. --Tryptofish (talk) 21:45, 13 September 2016 (UTC)
The related feature-request for Flow is tracked at phab:T121138 and further discussion in the 3 linked tasks and more in the links therein. (Flow is still very simple, with many as-yet-uncoded features - that's one of the many difficulties with agile/live development). The only way to do it 100% reliably with wikitext, would be to have every single section as a separate wikipage, as done by the limited examples mentioned above. Quiddity (WMF) (talk) 20:31, 14 September 2016 (UTC)
Thank you very much for that information, Quiddity. Based on the discussion above within this section, I get the impression that, even if it cannot be done 100% reliably within wikitext, it may (perhaps?) be possible for it to be >90% reliable. I think we all agree that making every section a separate page is not what we are looking for. But I can also tell you that, even with a lot more development, Flow is also not likely to be what I and many other editors are looking for. I'm not asking here about features for Flow. do you see this issue the same way that Quiddity just described? --Tryptofish (talk) 20:56, 14 September 2016 (UTC)
The ancient feature request for this as part of wikitext is phab:T2738 (applying to both content pages and discussion pages and everything inbetween), which contains some discussion. I'll try to track down related onwiki discussions and link them there. Quiddity (WMF) (talk) 21:19, 14 September 2016 (UTC)
Again, thanks. Speaking as an editor who has a very poor understanding of software writing, I wonder whether software could recognize Level 2 headers (or Level 2 headers could be made recognizable). --Tryptofish (talk) 21:34, 14 September 2016 (UTC)
The software can easily recognize headers (see how TOCs work). I'd imagine the difficult thing would be to tell the watchlist that, in a way that works gracefully if a header changes, doesn't require one to write the whole software anew and doesn't have unacceptable performance issues. Jo-Jo Eumerus (talk, contributions) 21:44, 14 September 2016 (UTC)
Yes, that makes good sense. Wouldn't a change of header, at the section level, be similar to page move, at the page level, in terms of what a watchlist could detect? At a minimum, I would think that an edit made to the section header would be recognized by the watchlist as an edit to the original section. --Tryptofish (talk) 21:48, 14 September 2016 (UTC)
Pretty sure that from the software's point of view it's a completely different thing. But I am not a developer and only comment here because some of the Flow and WMF debates remind me of past disputes on TV Tropes, a website I have been active on for a long time. Jo-Jo Eumerus (talk, contributions) 22:00, 14 September 2016 (UTC)
• Dear WMF, please do not spend donation money on Flow. In fact, please spend less money on software. Please divert some of donation money to spend directly on content, such as funding access to paywalled sources for trusted editors, creating a fund which offers incentives to photographers who take decent pictures of hard-to-photograph subjects to release their work on Commons-compatible licences, or to people who create useful sound files in the same way. Please divert other donation money to increased legal resources to deal with claims and allegations against content editors. Please set the software staff to developing scripts and other automated solutions that will help with the colossal backlogs which are everywhere you look. Please do not hand them yet another trainwreck-in-the-making. All the best—S Marshall T/C 20:01, 12 September 2016 (UTC)
Until the WMF is accountable somehow to the donors hoping that Wikipedia will improve based on their efforts, it probably won't happen. 20:17, 12 September 2016 (UTC)
May want to reconsider that. Jo-Jo Eumerus (talk, contributions) 20:21, 12 September 2016 (UTC)
• Somebody break out the wooden stake an kill this vampire. Liquid Threads was a mess, this is a mess. Utterly unwanted software that only project engineers seem to love. Carrite (talk) 16:17, 14 September 2016 (UTC)
• NOTE: The WMF is currently discussing whether they have enough responses to end the survey. As of Sept 14 they had 358 recorded responses, 193 responses in progress. They don't want too many responses because it means more non-English translation work. If you haven't responded yet, or if you have only partially completed it, I advise doing so as soon as possible. Alsee (talk) 05:54, 16 September 2016 (UTC)
Done for the survey but I'm neutral with Flow talk page format proposal. KGirlTrucker81 huh? what I'm been doing 12:23, 17 September 2016 (UTC)
• Note: With 583 responses, the survey has been closed. Trizek will take some time to work on and publish the results (phab:T144730). Thanks for all the input. Note that there was/is no intent to use the survey results in antagonistic ways, as some expressed concerns about above - it was initiated by the liaisons as a way to get a better understanding of how the editors who use the current version of the software feel about it; that's all. HTH. Quiddity (WMF) (talk) 22:16, 21 September 2016 (UTC)

## My 2 cents

Put some type of forum thing on the admin and 'crat noticeboards and their (non-archive) subpages, Arbitration/Requests/Case subpages, and "X for (deletion OR discussion)" pages, but NOT on talk pages. KATMAKROFAN (talk) 00:27, 22 September 2016 (UTC)