Talk:Community Wishlist Survey/Editions and projects

From Meta, a Wikimedia project coordination wiki

Is it possible to correct facts?[edit]

I see some projects marked as "Done" which are far from being done:

  • Real Time Preview for Wikitext is only available using the old wikitext editor, not the "New wikitext mode", use by 111,000 users at English Wikipedia. The result should be marked as in progress. If no new edition is going to be made to make it available wit the "New wikitext mode" it should be marked somehow as "partially done" or "abandoned".
  • Gadget: Who is active is an experimental gadget only available at MediaWiki. Not only that: it is twice in the list.
  • Enable live preview by default is not done.
  • Autosuggest linking Wikidata item after creating an article doesn't happen either.

Theklan (talk) 17:02, 24 January 2023 (UTC)[reply]

Plus, if there is an aim to prioritize the tasks that have been done, there should be an extra column corresponding to the voting ranking of that year. So that people can judge that a proposal ranked 20th may have been completed while the top 5 were ommitted. Otherwise, the list is utterly biased towards the preferences of the WMF and unrelated with the real results from the Wishlist (which is in fact a popular voting). Xavier Dengra (MESSAGES) 17:07, 24 January 2023 (UTC)[reply]
Hey @Theklan,
Sounds like you are hoping we will change the status in our Projects from Done to Partially Done or Abandoned. You’re right that in a sense, software development is never done. The way software is built is iterative in nature, and even after it is built, a team must maintain it for bugs.
The intent of labeling a project as “Done” is not to mislead anyone to think we do more than what we’ve completed— it is to communicate what is currently under active development with the “In Progress” label because those are the wishes that we are actively seeking feedback on to collaborate and build a better solution with volunteers’s input. “Done” signals we are no longer taking input, but we are still actively maintaining and supporting the tool we built.
For Realtime, we did not include new wikitext as part of the project as the original wish asked for a functionality similar to Live Preview which was only available for Wikitext 2010.Enabling live preview by default is considered Done since Realtime preview in 2010 editor appears in everyone’s toolbar — and that uses the same engine that renders Live preview.
As for @Xavier Dengra's concerns with completing wishes that were lower than the top 10 wishes, here’s an update we recently published on the Wishathon explaining why we completed lower scoring wishes. Our goal was to get other engineers at the Foundation to look at smaller problems in the wishlist that could reasonably be completed within a week. Most wishes in the top 10 take months to complete, as they should-- the hard stuff is not for a wishathon, it's for a dedicated team who can do consultations with Legal, Security, Performance Teams, and lead the user research necessary to build things the right way. We care deeply about what people ask for in the Wishlist and as a Community Tech team, we tried to ask ourselves, “how can we fulfill more wishes without taking too many resources away from the top 10?”We figured that a week’s worth of time compared to the multiple months it usually takes to grant wishes in the top 10 was a net positive. Do you think using one week to complete smaller wishes was a bad strategy on our part? Those lower in popularity wishes are still improvements people need— and often there is an inherent bias in our survey since there are different sizes of populations that use different tools and that manifests itself in the voting pools.
We know our process is far from perfect. We care and it upsets us as a team that every year 100s of proposals go unmet but this is why we put our thinking brains together and try to figure out what to improve about the process every year. We’ve added all of your comments to the retrospective meeting we will have after this Wishlist is done. We’ve been having internal conversations about how short staffed Community Tech is to try to make this better. The Wishathon was one idea that came from those conversations. We are the first ones to admit that things can always be better, that we could have more resources.
We welcome your critical feedback— and I hope your feedback can be constructive. I think “abandoned” sounds misleading, as you know from our Maintenance documentation, we are currently very much still supporting all the tools you want to label as abandoned. “Partially done” sounds like we did not complete what the original wish asked once you read more details and not just the title. Do you mind suggesting a third way we could change the word Done to better reflect what is meant given what I explained above?
Right now, I'm monitoring and reading the many other proposals as the Wishlist is ongoing so my bandwidth is limited, note this may mean delayed responses.
With my sincere best intentions, N NRodriguez (WMF) (talk) 18:12, 25 January 2023 (UTC)[reply]
Well, there are many things here to comment. Lets' try to make it ordered and civilized.
  • No, I'm not proposing that projects should be labelled as partially done or abandoned because projects are, by definition, never ending, but because they are partially done (i.e. they don't meet the needs it was intented). Something that only works in very specific circunstances, or doesn't work in all that should be expected can't be, by definition done. Marking it as done reflects that the current situation, excluding possible bugs or maintenance, is what we will get. Indeed, something is in progress if it is being researched, developed or tested. So a project that is not done and not in progress can't be neither. What is the correct word? It can be abandoned if no new effort will be done, or partially done if it meets some of the expectations the project had, but not all it should. You chose which is the best one for each project.
  • You say that Realtime preview asked in the origial wish a functionality similar to Live Preview which was only available for Wikitext 2010. This is false. Plainly false. The original wish is here. Wikitext editors often skip the step of previewing their edits, missing simple typos and formatting errors that could be easily avoided if previewed live. It doesn't say wikitext editors with the old editor, it says wikitext editors which includes all people using wikitext. Obviously, the solution developed by TheDJ in 2015 couldn't be available for people with the new wikitext editor because this was launched in 2017. TheDJ may be good, but not a futurist. Your job was to develop a tool that could be available for all people editing wikitext, and you have chosen a subset. Furthermore, you have chosen a subset that should be deprecated, as the new wikitext editor should go out of Beta some time in the future (or not, after 5 years in Beta, it also seems done but not finished). The second thing (available by default) is tied to this. Is not available by default (till it is deployed, it isn't even for the old wikitext editors). So, this is not done, nor in progress. You need to chose partially done or abandoned, is up to you.
  • We figured that a week’s worth of time compared to the multiple months it usually takes to grant wishes in the top 10 was a net positive. Yes, actually doing something is better than nothing. Having some extra wishes would be great, and even fair for other projects that don't have so many votes available. But extra means extra. NONE of the top 10 wishes have been done. That week was great. The other 51 were missed. I repeat: none of the top 10 wishes have been done. Zero. I would expect something not done because of time issues, but not doing anything, and then claiming that you did some other stuff during a week is a bit insulting for the community.
  • You claim that you know that the process is far from perfect. This problems were cited last year, and YOUR TEAM decided to forget the second most voted proposal (even after YOUR TEAM decided to hide it in a place called trashbin larger proposals ), which proposed a partial economical solution for this issue. It was said that the problem would be discussed with other teams, but this didn't happen. And that's why the process is exactly the same we have in the previous year. You can say that this year you will make changes, but is difficult to believe it, if there are no changes from the previous year. In fact, there is a change from 2 years ago: voting is not relevant anymore. It used to be, but now it doesn't matter.
  • Lastly: this list is also misleading because you are adding only the wishes you have been working on. Why aren't those chosen and not done added also to the list? That would be fair for the volunteers who made the proposals, discussed them, opened phabricator tickets, gave solutions and voted here to prioritize your [undone] work.
Theklan (talk) 21:21, 25 January 2023 (UTC)[reply]
I'm responding in my capacity as a fellow volunteer, so for the avoidance of doubt I am not speaking on behalf of my team — I appreciate you're frustrated by something to do with the wishlist, and this frustration likely comes from a place of positive passion for the community we all care deeply about, but the above comment is starting to stray from "ordered and civilized" (i.e. needless shouting, antagonistic wording etc.), which only serves to detract from your otherwise valid points.
I have always considered myself a community member first (with my staff hat a close, but distinct, second), and have publicly criticized the WMF for previous poor decisions — I'd like to ask you, Theklan, to trust me when I say that you are being listened to here.
We both want the same things: more community wishes completed per wishlist cycle. Well reasoned, civil and respectful criticism helps build a case for more time/resources/support — please can you help me here? — TheresNoTime (talk • they/them) 22:04, 25 January 2023 (UTC)[reply]
Sure I can! The first thing is sincerity. The team is not being sincere with the table of results, that is the discussion topic here. Once I see sincerity there, we can build a trusted discussion. Next, I took my time last year to make a simple proposal: Community_Wishlist_Survey_2022/Larger_suggestions#50_wishes. There was one with even more votes: Community_Wishlist_Survey_2022/Larger_suggestions#1%. What was the result of the consultation with the leadership? Because it waas promised that this discussion should happen... I don't know if it did, but there was no answer or solution to the problem.
This process is broken. Sorry, but it doesn't work. Only 20% of the proposals chosen by the community were done in the last 4 years. I trust you when you say that you want more community wishes completed per wishist cycle. I think that no one would say the opposite. The problem is that the team responsible (and I'm saying this to you as a volunteer, not as a member of that team) for doing this is failing to deliver. Who is accountable for that? What can we do if every year the resolution of the wishes is worse than the previous year? This year we had the impressive result of zero wishes done. There's no way this can be presented as a process where volunteers should be using their time, because it doesn't matter what we do: the result is 0. Proposing is a waste of volunteer time. Discussing is not effective. Voting is not even relevant. Is like asking for things in Phabricator and waiting for someone to solve them, but with extra steps. Theklan (talk) 23:00, 25 January 2023 (UTC)[reply]
Thank you Theklan, I really appreciate your reply — I realized after sending this that Dismantling of the annual Wishlist Survey system is probably the most appropriate and "centralized" location for actionable changes (bar the table of results issues you mention, which I suppose do belong here). I'm sorry for pinging you in multiple places!
Some of your answer above touches upon the points I raised at the proposal, and if you have the time I'd love to continue figuring things out there so that everyone can discuss it together transparently.
As for the table of results, I do see your point; as a fellow Wikipedia editor, I think we can both agree that differing opinions on the definitions of words/statuses can get fairly tense. To me, this seems a little less like insincerity from the team, and more like a dispute over the difference between done, partially done, not done etc.
I'm not overly familiar with eu.wikipedia.org (though Txikipedia:Azala looks very interesting, what is it!?), but I assume you have a process similar to requests for comment? I am a big fan of doing things "the wiki way", so perhaps a joint WMF-Community RfC on what counts as done etc. for community wishes would work here? If that sounds like an interesting idea to you, I can try to convince the team to do it TheresNoTime (talk • they/them) 23:22, 25 January 2023 (UTC)[reply]
I think we can both agree that differing opinions on the definitions of words/statuses can get fairly tense.. Indeed. But done is not usually a word where a tense debate should be made. Done means everything finished. If is not finished, or is only partially finished, or the final result is not what it was asked, then is not done by definition. We are not going to make an RfC for discussing what done means. I kindly ask, again, to stop discussing what words that everyone knows mean and just correct the table to aknowledge that:
  1. The projects marked here as done are not those that were chosen by the community (which are not done, most of them not even started)
  2. The projects that are not 'completely finished are not marked as done, because they are not.
Thanks. Theklan (talk) 08:16, 26 January 2023 (UTC)[reply]
I agree that "done" is a very unambiguous word, and I'm certainly not suggesting that it isn't — the differing opinion seems to, at least from my point of view, be coming from two different contexts:
(1) Is the wish considered "done" — i.e., is it complete to the specifications defined by the proposer?
(2) Is the work that Community Tech will do on the wish considered "done" — i.e., is the team going to explicitly work on developing the wish features any further, or are they switching to a state of maintenance?
Again, I cannot speak on behalf of the team, but my own understanding of what "Project status" means on these tables is closer to what I describe in (2) — hopefully with this context you can agree that "done" might mean something different to different people?
That being said, would you prefer that some tables instead track the outcome of a given wish, closer to what I describe in (1)? I don't see why both tables can't coexist on either this page, or another which it links to. — TheresNoTime (talk • they/them) 09:00, 26 January 2023 (UTC)[reply]
Done means (1). The other thing is abandoned before being done. The current table doesn't reflect the status of the wishes (0 done from the most voted 10) and even those marked as done (1) are not done. We can be discussing here about epistemology and metaphysics, but the table will still be wrong. Thanks. Theklan (talk) 09:52, 26 January 2023 (UTC)[reply]
I appreciate that is your opinion, and to some degree I share in the points you're trying to raise. I'm not sure adding a {{disputed}} tag (which appears to be tailored towards Wikipedia articles anyway..) is nessesary given the ongoing conversation here is reaching an impasse, but such is life.
I do really appreciate you taking the time to discuss this with me, and I will come back to this TheresNoTime (talk • they/them) 10:26, 26 January 2023 (UTC)[reply]
I have changed the template, so it's not about articles but about pages. Then... will the inaccurate content of this page change or should I start editing it on my own? Theklan (talk) 11:26, 26 January 2023 (UTC)[reply]
Hey there, there were a lot of points covered and perhaps the one I want to address the most is the idea that we've accomplished zero work on the top 10 wishes this year.
We deployed Phonos, the extension to support translation of IPA into audio to Afrikaans Wiktionary and Arabic Wikipedia. This was the #9 wish from last year. The process of product development is always to partner with pilot wikis to get feedback before we roll out to more wikis. This ensures the quality of the tool. The work is ready for feedback, and in that sense, it is in the final stages. Feel free to check out a demo of Phonos in beta.
For diffs, the number one wish on the 2022 wishlist, we've completed research and architectural planning. We've also completed clearing the proposed changes with Security, Legal, and the Performance Teams. We've worked with volunteers to test the designs of our proposed changes and even published the outcomes of our usability research as a video. We have a public Phabricator board with all the tasks associated with this work.
Votes mattered heavily when we accepted taking on this wish. Diffs was the top wish, so even though that wish is one of the most complex asks on the list of 2022, (no one at the WMF has built for diffs in the past 5 years since it is incredibly complex to change the C++ algorithm that compares what changed) we still decided to tackle it and partnered with the Performance Team because we care about votes and what people need.
I know there were more points you raised and I hope to address them as time allows, but I wanted to clarify this idea that we've gotten zero work done. Best, NRodriguez (WMF) (talk) 21:03, 26 January 2023 (UTC)[reply]
Ok, so from 10 wishes 8 are not done and 2 are in progress. Let's calculate: 8+2 = 10 not done from 10. 0/10 done, 2/10 in progress. Can this be reflected in the table, or should I proceed myself? Theklan (talk) 21:06, 26 January 2023 (UTC)[reply]
Today @STei (WMF) made a new edit, but still the issue is not solved. I'm assuming that there won't be accountability on the real situation here. I will proceed to change it by Monday if there's no answer on how to handle this. Theklan (talk) 19:01, 3 February 2023 (UTC)[reply]
Pinging also @Superpes15, as he suggested to discuss this before making any change. Theklan (talk) 19:01, 3 February 2023 (UTC)[reply]
@Theklan let me know what the problem is. –– STei (WMF) (talk) 19:29, 3 February 2023 (UTC)[reply]
Is a problem about facts. You have a full discussion here. Theklan (talk) 19:34, 3 February 2023 (UTC)[reply]
The page lists projects we have committed to. It doesn't make sense to list other wishes under "Current projects". Perhaps some simple language like "Only projects Community Tech has committed to are listed here" is fine enough?
I know it's disappointing to see so many unfinished wishes, which we wholeheartedly agree with, but it's important to read our prioritization framework to understand why we chose to work on what we did. It may be that in the future, this process changes, or even goes back to the top N, but I don't think that warrants pretending we committed to the top N on the years we did not. MusikAnimal (WMF) (talk) 20:38, 3 February 2023 (UTC)[reply]
I know @Theklan is blocked and can't reply, but I wanted to share my thoughts now that I've had time to go through all of this discussion (if you haven't noticed, we've been a bit busy :). I'm going to use my volunteer account as my comments have not been vetted by the team, though I suspect they largely agree with what I'm saying. We are across many time zones so it's pretty difficult to sync up quickly. Also it's the weekend. And we're running a survey at the same time. It's been pretty stressful :( So surely it's understandable that given an ultimatum to address all issues by Monday was not appreciated, as well as the insults to individual members of our team, including but not limited to our very hard-working PM :)
Re: Realtime Preview – doing this also for the 2017 editor I think is an unrealistic expectation. The 2017 editor is part of VisualEditor, which is an entirely different codebase. We consulted the voters of the proposal and there was definitely some confusion on 2017 vs 2010 editor, but if anyone opposed our focus on 2010, they didn't tell us, even after we pinged users again asking for feedback once we had visual prototypes to share. I think we would have loved to work on the 2017 editor too, but that would been at the cost of not working on more wishes. Balancing it all out, it seemed focusing on the 2010 editor was the right move. To clear this up on the Editions and projects page, we could rename the project page to state it's only for the 2010 editor (the page itself already states this clearly), which I attempted months ago but unfortunately it cannot be moved anymore because there are too many translations. This is a limitation with Extension:Translate. But I guess we could change the display title, at least, and/or the link label on the page. Also, we are not aware of any plans by the WMF to retire the the 2010 editor, so I don't think it's fair to say we worked on something that should or would be deprecated. We've done a lot of work on the 2010 editor, and we definitely wouldn't have if we knew it was going to be deprecated!
As for Live Preview, we tried very hard to get that properly on by default. It's actually been attempted in the past, but for various technical reasons it doesn't have the same parity with the normal preview and thus couldn't be enabled. As our PM has explained, Realtime Preview gives you both, so in that sense we have deployed live preview to everyone. I personally think this is ideal functionality, since you can use the normal preview to get the guaranteed final appearance of your changes, while still getting a live preview via Realtime Preview. Reception for our work has been very good, so it seemed to safe to call the Realtime Preview wish as "Done". The live preview wish though could probably be "Partially done" for clarity. It's not just Realtime Preview; we did significant work to live preview as well, bringing it closer to parity with normal preview. What's left may be out of scope for us. I admit we fell short here, as I was pretty confident myself we could truly make live preview on by default. Being more clear about the end result on our project page is fair. I am certain no one on my team is intentionally trying to lie to the community, we just perhaps did't convey things best, and it's great to see issues like this raised so we can fix them. We are only mere humans.
Balancing out our time to fully satisfy wishes versus getting more wishes done is extremely difficult. It isn't perfect, but I certainly hope most would agree that progress has been made. It's not like Community Tech hasn't been doing anything this past year. Speaking for myself, 2022 was probably the hardest year of work in my ~6.5 years at the Foundation. I feel that many of the requested changes to our project Editions and projects page and commentary over at the dismantle wish are going a bit too far and falsely give the impression no progress has been made. This is upsetting as we have poured our hearts and souls into our work. What's ironic is we all actually want the same thing – for more wishes to be done.
The "50 wishes" and "1%" proposals from last year were symbolic. We did raise them with leadership. There was nothing our team was able to do beyond that. We unfortunately don't have any updates, but I suspect things will eventually change with regards to resourcing in some way or another. These larger suggestions explicitly are not addressed by our team, so we aren't going to list them on our Editions and projects page.
I'd also like to address again the claim that voting isn't relevant. It absolutely is. I think the equations we use could be explained better in our docs, which we will fix, but it's also worth noting the prioritization framework is always subject to change (with the community's interests in mind), and we are clear about that. We were hoping folks would read the prioritization list with all the scores (even the spreadsheets used and their formulas are publicly viewable). Doing so, you would get a better sense as to why we chose what we chose. Understanding how product teams actually work in the real world is pretty hard to convey. This was our attempt at it. Maybe it's not good enough. This is only the second time we've used it, and as Selena Deckelmann pointed out, we weren't originally even going to have a survey this year, and instead work on more wishes from 2022, which would have surely included wishes from the top 10 as despite belief otherwise, those are weighted higher.
I hope this helps clear things up. So far, I've been straying away from the dismantle wish for mental health reasons. However myself and my team, as well as relevant leadership, will absolutely be reading every word and I am certain things will improve. MusikAnimal (WMF) (talk) 05:57, 5 February 2023 (UTC)[reply]