Do we want a wiki (or similar) alongside Q&A?

Please no. I had a bad experience of SO's failed and cancelled "Documentation" project. I raised the same concerns on the Electronics site here. I'll quote that post:

The worst that can happen is something similar to SO's failed "Documentation project" a couple of years ago, where users were to write documentation and examples of misc programming-related topics. These ended up in a separate place at the site - a category if you will. There was no clear structure for how to organize these articles - anyone could write one, edit one etc. You could provide feedback to specific articles. What happened was this:

It was rather unclear and subjective what problem "Documentation" was supposed to solve in the first place. (Wikipedia 2?) Documentation pages ended up all over the place, broad and arbitrary. It was impossible to search through it looking for a specific topic. Everyone and their mother felt the urge to write these posts, meaning that the overall quality ended up very low. The domain experts who wrote good articles had them drown in floods of crap. Duplicates and overlapping topics happened frequently, with no easy way to resolve them.

And then there was actually a peer review system in place + you could leave comments etc, but it wasn't nearly enough. The whole project collapsed under the weight of crappy, chaotic articles.

Something similar to SO Documentation must be avoided. Then the question is - how do you do that, on an open Q&A site where everyone is (and should be) allowed to post, regardless of knowledge level. And how do you provide structure and guidelines for what the papers must look like, how they are categorized and so on.

The above concerns are valid for a large, active and healthy community, which Software Development is not. It is currently struggling with lack of content, lack of scope, lack of users and lack of domain experts. There are many better and far more important things we could focus on, like coming up with a community consensus for on-topic and off-topic.

Basically, we need to learn our ABCs before running off to write encyclopaedias.

posted about 4 years ago

CC BY-SA 4.0

Lundin‭

8822 reputation 36 119 1014 163

I think the framing of this question (and the prior discussion) is wrong, and I think that conditions have evolved since it was originally asked - in particular, we can now see how articles have turned out for other communities. Rather than try to define "wiki" or consider ways to implement that strategy, I want to consider:

How could a category using the "article" page type be used here?

My own answer seems quite clear to me. The model that makes the most sense is already demonstrated by the "Papers" section for the EE community. The examples here generally look, to me, like one of:

• Guided "tutorials" that build to an existing concept step by step (e.g. the actually-existing design of basic op-amps)

• Closed-ended, but non-trivial, lists of well-agreed-upon reasons for a practice, or methods for some objective. These are susceptible to being presented in a way that looks more like Q&A, because there is more of a need to specify the underlying problem in detail.

• Alternatives for ways to visualize or understand a difficult or subtle concept

All of these make sense to me in a software development context, although I would modify the last one somewhat - we might instead have articles that focus on a single, commonly misunderstood concept (say, recursion), and address multiple misconceptions with that same concept. I'm a little less positive about the idea of doing tutorials, but I can imagine offering e.g. a collection of ways to do individual tasks that sum up to "managing a language-X environment" or "making use of branches in a repository" or "choosing between absolute and relative imports in Python". You get the idea.

Where I think this model could really shine is for the medium-length-list type questions, where there are a few well-defined ways to do something simple, and a fair bit that is useful to say about each of them. For example, the nearly 3000 more than 3200 words (if we count the dozens of short code examples as text normally) I recently wrote about string formatting in Python. When there are fundamentally about 6 (depending on how you count) sensible approaches to a problem, that isn't too widely scoped for a question (the question itself is quite coherent, and would look the same if there weren't the historical baggage of %-style formatting etc.). It offers an executive summary that fits within short-term memory. But being able to put all the answer content in one place, up front, avoids a combinatorial explosion of attempts to explain various subsets of the Ways To Do It (and of various aspects thereof). It also avoids confusion over whether to answer with a summary of all ways, or in detail about one way - more importantly, it avoids worrying about coordination problems between answerers.

The potential downside I see is that (as things stand, to my understanding) articles could not be directly used to close duplicate questions, at least not if they're in their own category (which would IMO definitely be desirable for organizational reasons). But then, a lot of the time it would be better to answer questions with a "stub" that refers to main article content (and perhaps points out a specific relevant section) and then in some cases it could be closed after that (as no other answer could reasonably be necessary). Possibly in some cases, those questions could become really good duplicate targets. The stub would give a direct answer for a very specific question, and then relevant detail is behind the link (just as it would be for answers that quote and cite documentation and then give a one-sentence explanation of how the quote answers the question).

As for specifics about article writing: I envision people taking advantage of the threaded comment feature, using comments heavily to point out missing, poorly organized or otherwise inferior content. Separate comment threads could discuss any given issue, and by convention OP would be responsible for significant content edits - to maintain a consistent style. Others could edit for the usual grammar/typo/copyediting reasons. I imagine articles making heavy use of <details> tags, so that they can avoid taking up space with a separate "table of contents" (although it should be possible to link those up internally with HTML), and make it easier to scroll over sections that don't interest the reader.

posted about 1 year ago

CC BY-SA 4.0

1y ago

Karl Knechtel‭

2020 reputation 33 48 258 74

I hear the walk-before-we-run argument. I think this would be a good thing to try once we reach running speed, though.

Personally, I don't like self-answered questions; I think they're an awkward fit for Q&A sites. I'm primarily here to scratch my itch to help people—when I see a question that's relevant to my interests, only to realize that the asker never wanted help with the question at all, I feel a little let down. Spreading knowledge proactively is a noble goal and I don't want to discourage it, but I think there ought to be a more fit-for-purpose way to do it than the self-answer approach, which I think is kind of a hack.

The proper comparison with articles here would not be Wikipedia articles, IMO, but articles on GitHub wikis for specific projects. Wikipedia is a bad place for rando project documentation because it doesn't meet notability requirements. GitHub wikis are a good place to put that documentation, but GitHub issues are, in some projects' opinions, bad places to ask questions, and so you see a lot of README.md instructions asking users to consult both the wiki on GitHub and Some Other site when they have questions. Enabling articles would let projects define a one-stop shop for knowledge base articles and Q&A, which I think would make a certain amount of sense.

posted about 4 years ago

CC BY-SA 4.0

r~~‭

2785 reputation 3 60 274 33

There are a few ways to understand "wiki".

1. Like wikipedia - an interlinked web of articles in a standardized format.
2. Like SO's documentation project or Github wikis - a stripped down wiki that is kind of sort of like a wiki but also weirdly shallow.
3. A post without answers, where the post itself is the point.
4. Question that does not have a single right answer, but is open ended, every answer is one possible item, which the community votes up or down, and the "true" answer is the ranked list of all answers.

1 is a clear no - wiki sites are too different. If the CD team wants to try a separate wiki project, I would excitedly follow it, but I think mixing the two would be like oil and water.

2 - I don't have much experience with. I remember when SO tried it, as an observer it seemed neat and interesting, but disappeared out of nowhere. I would love to hear a detailed retrospective from someone who used it heavily - but I think it's more complicated than CD needs to be. Again, as a separate project, might be great, but it's at cross purposes with CD's simple Q&A format.

3 - This is an interesting one, I guess https://proposals.codidact.com/ is this model too. What I like is that it's a nice, logical extension of the Q&A principle. What's weird to me is that it feel very restrictive to have only comments, not answers, to elaborate on the question. But that restriction goes away once you realize you can just ask questions in the other category and link to it. So perhaps this is something that feels new and scary, but as people get used to it over time, will become a much appreciated aspect of Q&A sites.

4 - This was the DIY way on SO to do "wikis". The recipes example on pre-2010 SO would have been "what are the best dessert recipes?", with each answer a different recipe (sometimes alternates of the same). Soon after 2010 SO decided these were no longer okay, and I can sort of see the argument about open-ended, never resolved, etc. But if it was up to me I'd use it. Notably, this format survived on some niche sites, like code golf (especially the popularity-contest tag), where it seems to be working quite well.

I feel like the ideal solution would be something like 3 or 4. But instead of "questions without answers" it would be "answers without a question". The difference being that people can post multiple alternate version of the same thing, and let the best one win. For example, I posted https://proposals.codidact.com/posts/288625 - but maybe someone feels it's all a bunch of crap and they want to completely rewrite it. It's a bit hard to do that with edits sometimes, because edits are not as exposed to community consensus and dialogue. But if they could post an "answer" to my "question" which is basically the same post, but rewritten to be better (according to them), I think that would be the best solution. Recipes are another great example of this - it's certainly valuable to be able to post a few alternative takes on veggie stuffed peppers. And it's not a matter of "the best recipe" - often it's nice to have a few variants, and it's not about eliminating all but the best.

tl;dr: The articles are nice, but they are too biased towards the OP's take on the subject. We should allow posting alternate versions of articles as "answers" to them.

posted over 1 year ago

CC BY-SA 4.0

1y ago

matthewsnyder‭

2209 reputation 52 59 259 92

I am not categorically opposed, but I currently don't see a use case for articles.

For one, wikis already exist. What would be the benefit of hosting our own rather than participating on Wikipedia? After all, Wikipedia has a very mature software and policies for this very thing. It also has a much higher reach thanks to its impeccable google ranking.

Also, at least at the current maturity level of the implemenation, self-answered questions seem to have some important advantages over articles:

• Having to write a question makes it easier to tell what that article is about, allowing readers to quickly assess whether it is relevant to them.
• If other people think they can write a better article, they can.

In summary, I don't currently see a usecase, and would leave this feature disabled until we do.

posted about 4 years ago

CC BY-SA 4.0

meriton‭

2375 reputation 7 56 220 35