Specifically: delete the UI, delete all the controllers, delete the data, purge any evidence that anything was ever voted on.

I think we've talked about this before, but I'm a fan of Quora's model for Ponder, question and answer are wiki (openly editable by anyone).

You'd favor removing answers entirely, and just have a wiki page with "Answer"?

Or a wiki "summary" page, in addition to individual answers, like Quora?

I'd try to remove them outright, just have Q&A at top as wiki, timeline discussion below. I think it'd take a little thinking to making it clear where to provide an answer.

What I like about this model ^^ is it's easy for places like here to seed common questions and answers.

The second thing is for companies, it's much easier to correct information. Maybe it's "how do I install my dev environment" and instead of new answers on the bottom, when procedures change, it just gets immediately corrected. Someone could also just comment "this no longer works" and all the people who edited the answer would be notified and update the answer as well. Lastly, it makes it more quickly obvious (top of page) what / where the most important information is.

If this is appealing I could do a 10 minute mock on it? Or did you have different paper napkin ideas?

At least for me, a huge part of the incentive to use Quora (and, in theory, Ponder) is that people can see that I wrote a hilarious answer.

Do you have like 3 examples of questions that you'd want to seed in Ponder?

Some questions that I think are a good fit for Ponder are:

What are some of the best image macros?
How can I convince my company to use Phabricator?
How does your company organize work in Phabricator?
What are some funny stories about Phabricator?

These questions do not have a singular answer, are anecdotal/contextual, the identity of the user answering the question is important, and being identifiable partly incentivizes users to answer.

I think these are bad for Ponder (real questions we have today):

How do I enable prototype applications?

If users are having trouble with this, the UI isn't clear enough.

Getting Error on Text Panels in Dashboard: "Call to undefined method ProjectRemarkupRule::isFlatText()"

This should be a task.

Sorting Countdowns By Countdown Date

This should be a task.

Can I reference the current user in custom queries?

This is already a task. If we had the capability, this would be a UI problem.

I would basically close all of these, and try to change Ponder to discourage asking these kinds of questions.

I don't think "how do I install my dev environment" should be in Ponder. It should be in Phriction, for the reasons you describe.

In T6920#90853, @chad wrote:

I'd try to remove them outright, just have Q&A at top as wiki, timeline discussion below. I think it'd take a little thinking to making it clear where to provide an answer.

Isn't user rating what makes Ponder Ponder? If we leave it in title editable description and sequential comments, then we are talking about a simple Maniphest.

My words don't have much weight since Wikimedia is not using Ponder and there are no plans to use it. However, there have been some discussions, and rating (or validating / supporting) answers was a factor considered. I'm asking just as a user.

This is why I'm not a super fan of Ponder. It's either too much like Maniphest or too much like Phriction. My concern is removing voting makes it just a discussion board, which if that's all we want, is fine. I just have no idea what it'd take to make it useful here.

Basically, we have zero real use for Ponder as a Public install, and zero real use for Ponder as a Company install. Even if it moved towards 'user to user help', that'd also be maybe better served with 'Public Conpherence Chat' for us anyways.

Are there any arguments for keeping Ponder? Do you have an idea what questions should go there?

Isn't user rating what makes Ponder Ponder?

If it is, we should delete it. Almost no installs will ever have enough users to make voting meaningful. If the application requires something that only a tiny fraction of installs will ever have, we should just get rid of it.

sequential comments

Removing voting doesn't preclude us from ranking by tokens awarded, or letting users be marked as "topic experts", or having "Mark as Needs Improvement" or "Mark as Good Answer" to demote or promote answers, or other things I haven't thought of yet.

Sites like StackOverflow and Quora have a class of moderator tools like closing questions as duplicates, closing them as poor questions, marking answers as needing improvement, etc.

They also have a class of user-facing tools (voting, reporting). In the case of StackOverflow, moderation powers become available as you gain internet points.

Ponder has only the voting tools. This is backward: all installs are small, but Ponder does not have the moderation tools for a small userbase. It only has the meta-tools that let you scale moderation to a large userbase, which no installs have.

we have zero real use for Ponder as a Public install

I'm not convinced of this yet. Here are some questions that were asked in our Ponder which I think are actually pretty good fits:

What is the best way to plan a tasks in phabricator?

I'd rephrase as something like "What are some ways to use Phabricator for project planning?", but I think this is basically a good question that a lot of users could give different, useful answers to that will still be pretty relevant in a couple of years.

Best practice for Aduit

I'd rephrase as something like "What are some ways to use Phabricator's code audit tools?", but same as above.

How to/recommendations for organize projects/teams?

Similar to the two above.

How to manage Differential revisions spanning multiple repositories?

This might have a "right answer" a couple of years from now, but could reasonably have a few creative workaround answers today.

How can I properly setup harbormaster with my working-copy resources?

A slightly more general version of this ("What are some ways to use Harbormaster to manage builds?") seems like a reasonable question to me.

Mixed public/private phabricator?

Seems reasonable with a slightly more general rephrase.

I don't know that these justify Ponder, but all of these questions are: open ended, with no "right" answer; experience, anecdotes and context are relevant; answerer is relevant; answers are unlikely to change much over time.

These questions don't feel like a good fit for me in Maniphest (too much crosstalk, no way to promote good answers or hide bad ones), Phriction (no attribution), or Diviner (no attribution, experience/anecdotes aren't well suited, high edit barrier).

While we don't have an internal use case for the tool today, I'd imagine these questions being good fits if we had ~15 employees:

What are the best text editors for working with Phabricator?

I'd give a TextMate answer, Bob would give a vim answer or whatever, the new guy we hire would give an Atom answer, etc.

What are the best places to eat near the office?
What did you build in the Jan 2017 hackathon?
How did you get hired at Phacility?

These questions all have the same properties as the questions above.

Unsure if helpful, when I used ponder previously we basically shoved all of the "happens every 3 months" or "happens with every new employee" discussions into it. If I said "skype sucks" (sorry ms!) then the irc bot would put a link to the discussion where replacements had been swatted down in the past :) Sometimes, not often, but sometimes a person would update one of these discussions with a really good answer that essentially something had changed since the previous stalemate. Ponder actually was super great for this. Long, ongoing organization wide debates on best practice were housed fairly nicely and to decent effect over the "long" term.

With the ability to mark the last best-effort "conclusion" it would have been even better. "Go here to see why we use this and read the last 100 replies on why every other solution is more evil".

Ponder is probably the prototype app I've been most excited to see grow and integrate into our use case :)

In my case voting in Ponder is rather good thing. In my use case, we plan to use Ponder as intermediate tool for actual questions in a way that lets voting and/or better skilled people answer.

I shall elaborate on this: in our install, Phabricator is set as developer/admin tool + last line of support. We are doing whatever we can to make room between devs/admins and support guys. We plan to have it set up so that 1st line of support uses support tools that are easy to use for customers (think something like zen support) and contact directly with customers. 2nd line guys very rarely contact with customers directly, but use support software and can answer to customers if 1st line fails. If they are failing to answer they should ask 3rd line guys or another 2nd line gyus.

That's where I see Ponder in our install - on 3rd line questions are asked by 2nd line guys, and are answered by 2nd or 3rd line guys OR even those that actually wrote the system. From that answers we could have either one-off question, KB entry for customers/1st line, KB entry for 1st/2nd line, KB/Wiki entry for 2nd/3rd line, or Maniphest task...

I don't think that in above case Conpherence would fit the bill that Ponder can.

In my installation I even used the "voting" feature in a gamification mechanism in order to increase the willingness to answer questions.
The points you get for an answer are multiplied by the number you get from the voting (+1). In my case I also give points for questions (even if a bit less) since I want to spread the utilization of this nice application.

At my company, we use Ponder very MUCH like Stack Overflow. When there are multiple answers, the voting is essential for determining which one is the "best answer" by group consensus. Removing the voting increases signal noise, and basically makes it like Phriction. If we simply wanted a single group-edited answer, we'd put it on Phriction in the first place.

In fact, Ponder is an essential part of our workflow! In its present form, it has a role separate from Maniphest, Phriction, Slowvote, and Pholio, and the delineation is perfectly clear and natural for us.

Does this task tracks the removal of Ponder (uninstall) in the https://secure.phabricator.com site ?

No. There is no task tracking that. T3578#90847 has a small amount of discussion.

Going to "poll" the audience here, those who are actively using Ponder, is the voting used/helpful?

I wonder if I can turn on Ponder here.

omg

So I'm likely to remove voting and in it's place will be some lightweight tags users can give to questions and answers, like "helpful", "interesting, "offtopic", "such wow" (or some other Phabricator Flavor). Mostly, these feel more "Phabricator" like to me than just voting up and down, and more helpful on smaller installs.

At least, even if I get one "helpful" tag, I'd feel more awesome than just getting an upvote.

Yeah, I generally agree with that direction. I'm not really even sure if we need anything, or could maybe get away with a binary thing (like "good", which makes the answer gold and promotes it to the top of the page) since a lot of the "such wow" stuff can probably be covered with tokens.

Well "Mark as Helpful" as a start. Will see what motivation I have.

If it's helpful, there was an entire debate on Stack Overflow about whether downvotes were useful. My conclusion (FWIW) it that downvotes only serve as an anonymous "I dislike this because I'm a jerk" for a lot of people, as they don't actually bother to read the tooltip "Not Useful".

Thus, having several positive and negative tags specifically for Ponder questions would be very helpful. Things like +Helpful, +Clever, +Hack Value, -Dangerous, -Incorrect, etc. would be more helpful than up/down votes, and more specific than tokens. My logic is that one needs to quickly assess whether a solution should be followed, and a bunch of -Dangerous tags on it make it clear that the solution is NOT something that should be followed, while a bunch of +Clever tags would make the answer stand out above other valid answers.