They’re not discoverable, you say… are they not indexed by search engines? Do they rank low?
I didn’t see voting yet. I’d love to have an experimental category in our Discourse where we could play with it. Can I get permissions to set that up if no one else wants to do it?
I’m not sure which social issue you’re referring to. The benefit that I see from a Q&A platform is the accumulation of discoverable (hopefully clearly articulated) questions and answers in a convenient platform with appropriate features.
How do I know that this repo has discussions if I do not know in advance that it has discussions?
Edit: Okay, I have to revert this. They do not have discussions enabled, that has been on another repo, which I can’t find currently. I tried enabling discussions in my personal repo, and they appear in the bar then.
Still, my opinion about the “its hard to follow discussions there” remains.
We do not have categories, they are not included in notifications when watching a repository (which you generally do not want for nixpkgs anway).
They are simply badly integrated and not well though through in my opinion.
@mightyiam the social issue is to have people hang around on the platform, use its tools effectively, and actually answer those questions and keep the information organized. I think it doesn’t matter much which tool this is done with above some threshold of functionality.
Any response can be chosen as “the answer” and any response can receive likes. In Stack Overflow (and I suppose any reasonable clone), answers and responses are distinct. In SO format, a question has a linear thread of comments and each answer has a linear thread of comments. Also in the SO format, answers can be both upvoted and downvoted, which adds value. Also in the SO format, both questions and answers (and probably also comments on either) can be flagged, encouraging involvement of moderators.
The SO format is designed for Q&A, whereas Discourse’s feature is bolted-on to an existing model that is not as nice for strictly Q&A.
If there is an open source SO clone that is sufficiently feature rich, I’d consider deploying one and playing around with it. Does that sound reasonable?
Like for any decision that is hard or costly to revert, while it sounds heavy-headed, I think the best strategy is to create an thorough overview of different choices and their consequences for the people who will eventually use and run the thing, as well as project leadership, to decide on. We have Discourse, Stack Overflow, GitHub, and possibly self-hosted solutions - one can make a table with feature comparison, and briefly discuss pro and contra.
But before going into that, I’d first ask if this is something worthwhile to spend time on in the first place, whether the payoff is worth the effort. @mightyiam I remember you suggesting in the last documentation team meeting to work on the deep issues first.
Right now, in my opinion, supporting more users with working around their problems is secondary to some deeper issues we have with enabling contributions which would hopefully solve many of the user-facing problems to begin with.
Yes, this is a reasonable strategy where the decision doesn’t feel clear.
So, I have a suggestion. I just found this lovely platform, Codidact.
I looked around in it and without really trying it, it seems to offer at least the base feature set of a SO clone. Also, it is open source, so we could probably get to keep our data. Also, it is a non-profit that seems to have goals that are aligned with ours:
Ever had a question? Ever had an answer? Ever had knowledge to share? We believe that coming together to inquire and learn, to share knowledge and to teach each other, makes the world a better place. We also believe that Q&A communities should be free from the politics and shenanigans of private, profit-focused companies.
Regarding prioritization and available resources, I am volunteering to set this thing up. To contact them on behalf of the Nix community and ask whether the Nix stack could have a “site” in their hosted platform.
I would then cooperate with them to set up that site. And I would then communicate the new site to the Nix community. I would then file an issue on the nixos.org website repo asking to refer to this site as the official Q&A platform.
I think this is a question that must be directed to the moderation team @ryantm@zimbatm, because they are supposed to be running that show, and possibly escalated to the foundation board since this is a strategic decision that concerns the whole community.
The current official Q&A platform is the Discourse Help section. If that’s not obvious, then let’s fix the various places where it’s mentioned to make it clearer.
If you look at how many posts are getting replied to, it seems to be working pretty well. I don’t mind if there are duplicates; asking questions is part of the learning process. And who knows, the newest answer might be different as the project keeps evolving. Or if it becomes too repetitive, it might trigger somebody to improve the learning material.
One thing we could advertise a bit more is to ask the post authors to mark when their question has been solved. There might be an option to remind them. If you have some suggestions, I’m happy to try them out.
Fundamentally, StackOverflow’s incentive is for users to spend as much time on their website for as little cost as possible to them. That’s why they try to delegate moderation using a points system. Or encourage de-duplication because stable URLs are better for Google juice. Some of the tensions mentioned in this thread are basically a result of their incentive model. Part of their incentive is also to be useful to users so I’m not saying it’s all bad. But I think that Discourse is OK, and it’s something we can modify to our own needs.
@zimbatm you’re discarding the idea that setting up an official built-for-purpose Q&A platform and recommending it for that purpose instead of Discourse is not overall a good idea or not worth the effort or what?
I’m of the perspective that a nicer, more pleasant and feature-rich Q&A platform that resembles the familiar Stack Overflow format would attract more and higher quality participation.
Discourse is a linear forum where posts are on a continuous timeline.
SO is a public Wiki where comments and answers are strictly separated and answers are sorted using a metric of usefulness.
The only thing they have in common is that people can post questions and answers.
Claiming that one can be good replacement for the other would be plain wrong.
The main features of SO that are desirable over a linear forum are the fact that anyone can edit questions and answers, that the “good” answers are the most visible rather than the ones that came first and that unwanted noise like comments or further troubleshooting/questions can easily be skipped over when you’re looking for solutions.
The ability to mark threads as duplicates or accepting solutions are rather minor features that SO could totally do without and lose none of its usefulness as a learning resource/source of unofficial documentation.
I don’t like handing control over such a vital part of the UX to a proprietary service run by a for-profit organisation either. I know that there’s at least one self-hostable SO clone that the Vim community used to use (Askbot) but it’s an eyesore and isn’t maintained anymore I believe.
Maybe there are other similar solutions. Discourse however isn’t one of them. Linear Forums are terrible for finding answers.
To its credit though, it could be worse: Chat rooms are information black holes, especially proprietary ones (looking at you, Discord).
SO-esque Q/A wiki > Reddit > Linear Forum > Chat room
These points are valid, SO is more featureful than Discourse. But it also splits the community into yet another data silo. Whether you’re for or against SO is mainly a question of how much weight you put on each argument.
For me, the SO wiki is an anti-feature. It’s good in the short term but doesn’t push us to update the official documentation. Slowly, SO is becoming the source of truth instead (which is good for SO because they get more traffic). I would rather have us reply to questions here, even if it’s repetitive, and fix the docs.
It’s not that it has more features, it’s that it works differently.
Both have up- and downsides for different purposes.
I think SO absolutely sucks for having a discussion for example.
I’m not sure I get what you mean by this.
If we had capacity to “fix the docs”, that would be nice but I don’t think we do. Short or long term.
Even if we did, I don’t see how a QA site would be detrimental. Adding your QA answer to the official docs would require the same amount of work as adding your forum post or GH reply.
A QA site answer can still have links to the official docs. In a sense, that’d make the official docs more accessible because they’d become googlable (currently, they are certainly not).
The QA site could become a glorified “How do I …” → Relevant place in the documentation map, that’d be a totally valid use.
That’d require capacities to “fix the docs” of course but those would be required in any case and, until we have good official docs, we’d at least have decent discoverable unofficial docs.
Also note that I’m not advocating for SO in particular, I’m advocating for their style of QA-focused forum. Whether that’s Stack Exchange Inc.'s software or anyone else’s, I don’t care. See @mightyiam’s suggestion for example.
Sorry to throw something else, somewhat tangential in here, but I think there’s something to be said for not continuing to adopt platforms that we don’t control.
I am very, very upset every time I run into a nixpkgs issue or PR that I simply can’t see, read, anything because of how GitHub handles [some] user bans. I don’t know that SO has ever committed such fouls, but I also know that I’ve never once had this sort of issue on a self-hosted platform.
So, separately from the concern of having “another place to check” (which, is real, I already can’t keep up with Matrix rooms + Discourse + actually get anything done, but obviously one person doesn’t have to be everywhere), I think maybe there should be further consideration given to self-hosting something, I guess like Codidact as was mentioned earlier.
(I do also understand that users are on StackOverflow and that to some extent we’ve got to meet them there as well. But, I suspect a “Community” or “Help” nixos.org page could easily link to “Official Q&A Platform” and address most of it.)
