Should we make the Red Baron hat trigger a permanent badge?

I think we should replace Reversal, for all of the reasons listed in that discussion and then some:

• Current Reversal badge encourages answering awful questions without turning them into good ones.
• Current badge actually encourages downvoting the question you're answering (or, in some unfortunate cases, posting "witty" answers to the questions you're downvoting).
• Current badge pits answerers against moderators and editors; anything that discourages someone from trying to improve a question they cared enough about to answer is pretty lousy.
• Current badge is... More of a consolation prize than a reward.
• Current badge is misnamed. "Diamond in the rough" would've been good (if long). Reversal is actually a pretty good name for this criteria.

The downside here is that we'd effectively be "cheapening" the Reversal badge: right now, it's one of the harder badges to earn. Answers that score 20 or more aren't that rare, but answers that score 20 or more and were posted to downvoted questions are currently uncommon - although I'll wager we'd see a few more if this badge was implemented. Another side-effect of this is that a decent number of these badges would be awarded to answers that sorta rode the coattails of another, much, much more popular answer...

One possibility here would be to make this a two-level badge:

• Silver: Red Baron or Deus Ex or The Cavalry or something, posted answer of +5 to question of -3 that went on to score +3.
• Gold: Reversal, posted answer of +20 to question of -4 that went on to score +4.

This would create one badge that's much more common than the existing Reversal, and another that's much less common.

Other criteria changes we should consider:

• Disallow locked questions
• Disallow locked answers
• Disallow Community Wiki answers

Creating a new badge is potentially interesting (I don't have a strong opinion on it). I don't think, however, you should replace the Reversal badge. (Maybe eliminate; see below.) They seem similar at first, but Reversal and Red Baron are different in some key ways:

• Reversal requires 20 upvotes rather than 5. On most sites, in my experience, that's the difference between "yeah, reasonable answer" and "yes!". (On some sites Reversal is extremely hard to get because of small communities, but nobody ever said badges were consistent across sites.)

• Red Baron rewards improving the question (yay!), but sometimes the question really can't be helped but, yet, it's worth somebody writing an answer to help the OP out. For example (deliberately not linked), there's a highly-downvoted question currently on MSE where somebody asks how to get a job at SE -- closed now as off-topic, but the answer still provided useful information. I think a Reversal for that is fine and we wouldn't necessarily want to try to fix the question.

• Red Baron requires that the question already be downvoted when the answer is posted. Yet we generally try to discourage answering those; downvoting usually means it should be put on hold, not answered. So it seems like Red Baron creates a not-quite-right incentive here; is that what you want? (Reversal, on the other hand, is sometimes awarded for questions that were later downvoted.)

If Reversal is a problem (as suggested in comments and other answers), then maybe we should just retire it. In any case don't redefine it; a badge renaming is just confusing and I'll bet you can come up with another name for what you want.

Finally, if you do implement a badge for Red Baron, you might consider a tweak: only award it where the author of the answer did something to help fix the question -- edits for sure, but arguably close votes too. Encourage fixing the question and not just targeting likely candidates that you can ride the coat-tails of.

After reading the comments and answers, talking to our developers, and cogitating, I'd like to suggest a slight variation on the Red Baron criteria:

Reversal

Answer with a score of +5 or better a question with a score of -3 or less at the time of answer (not including the answerer's own votes) that scores +3 or better within 7 days.

Whew, that's a mouthful. Let's break it down into component parts:

• "Answer with a score of +5 or better"

  A small percentage of answers ever get to +5, so this is a feat in itself. We are looking at pretty good answers here.

  There's no time limit on how long it takes for the answer to reach +5. It'll make the query a bit more complicated, but it might be worth adding a time limit on answer voting. See below.

• "a question with a score of -3 or less at the time of answer"

  When the answerer posts their answer, the question looks bad for one reason or another. But the answerer was able to determine what was asked well enough to provide a quality answer. On Stack Overflow, that might mean duplicating the circumstances of the question and discovering that the asker really did provide enough information to answer.

• "(not including the answerer's own votes)"

  A complication with the Red Baron criteria was that the order of action mattered. If a user asked first and upvoted later, they got the hat which they might not have if voting first and immediately answering. So we ignore the answerer's prior votes altogether. If their downvote pushed the question to -3, they won't get the badge. If they upvote to bring the question to -2 a the time of answering, they still get the badge. So there's no benefit to manipulating the question score prior to answering.

• "that scores +3 or better within 7 days."

  This is the actual rescue. It's also the major change from the hat criteria. Once a user answers the question, it starts a week-long clock (metaphorically) for voters to change their minds about the question. I added the clause for two reasons:

  1. The longer it takes for the voting to come around, the less likely it was the answer that did the trick.

  2. Without some limit, tracking potential badge-winning answers becomes difficult (both for users and in queries).

  In truth, this particular requirement has minimal impact as far as I can tell. None of the Red Baron hats would have been eliminated since most of the question voting occurred a day or two after the answers. (Well, all the voting that wasn't prompted by my publishing the list of questions.)

Now for the bad news: this particular trigger would produce a grand total of 346 badges across the network. (139 of those on Stack Overflow and 77 on meta sites.) It turns out that reversals of this magnitude are fairly rare; gold badge rare in fact. In order to bring this badge trigger into line with other silver badges we would need to adjust the criteria down somewhat:

Rescue

Answer with a score of +5 or better a question with a score of -1 or less at the time of answer (not including the answerer's own votes) that scores +1 or better within 7 days.

This criteria would result in 6,351 new silver badges on Stack Overflow and 11,909 networkwide; still on the stingy side. But I expect that like the hat, a pair of badges will encourage more people to seek out misunderstood questions to answer.

Nice idea, but problematic tracking....

Surprise badges are nice when they arrive, like "whoo hoo, popular question!", but for almost all badges, there is a user-queryable mechanism for tracking your progress to the badge:

1. you can see the score you have on posts, and track that "great answer"
2. You can watch the populist happen
3. you can see reversal
4. etc.

Badges where you cannot track your progress tend to be the ones with the most problems:

• Socratic - you can't see old deleted questions of your own
• Publicist - how effective was that tweet?

This "Red Baron" badge will be the same. How will you know how close you are? You cannot search for answers you wrote to questions that had a score of -3 or less at the time the answer was given

Perhaps a feaure should be added that indicates on each answer what the question score was at the time the answer was posted.... . hmmm...

About the importance of tracking

I wonder if you might expand on why tracking progress is so important that we might skip a badge that can't be tracked.

Specific to this badge

Let's consider the mechanics of this proposed badge:

• find a (community-perceived) crappy question
• answer it
• perhaps 'fix' the question
• get votes for both the question, and the answer
• Profit ???

What we don't want is people answering crappy questions just because they are crappy. We want crappy questions to just disappear, and be deleted. A half-decent answer on a crappy question makes the deletion much less likely (requires community votes - instead of automatic).

The purpose of this badge is (should be) to reward people who turn around bad questions and make them good (in addition to answering them well). The purpose is not to encourage people to answer (m)any old bad question and hope that one of them may get a badge.

OK, so you want to reward people who make bad questions good, not reward people who answer many bad questions in the hope that one of them miraculously becomes good. The reward is for providing a good answer and most importantly, for nursing the question back to health.

Consider three user-types on SE:

1. the user who does not care about badges, who unwittingly stumbles over a poor question, answers it, and then unwittingly gets upvoted, and gets the badge. This user is not affected by the badge until afterwards, and the badge did not modify their behaviour
2. the opportunist who targets bad questions, and answers as many as they can in the hopes that one will result in a badge. This is behaviour we want to discourage. This user will have a search set up for score:-5..-3 is:q answers:0..1 hasaccepted:0 closed:0 views:..100 and then just answer as many as they can. No point in answering anything worse than -5... too much luck needed when there's plenty of -3's out there.
3. the hunter, who decides to identify and repair broken questions. This user will actively find 'diamond in the rough' questions, answer them, repair them, continue to nurse them, and get justly rewarded for the effort put in to the process.

The first and second types of user are not deserving of the reward. Hmm, 'deserving' is the wrong word, they deserve the badge, but they did not earn the badge by behaving in the intended way. They did not satisfy the 'spirit' of the badge. The first type of user is not influenced by the badge at all. The second is influenced in the wrong way. Remember, this is a gold badge, demanding outstanding effort. The third type of user is deserving. How do you encourage the third-type of behaviour?

First, the user has to know what to nurse. I have answered about 1400 questions on the SE network, the first thing I would like to know is which of those answered questions is a candidate for the badge. Which questions can I go back to now, to edit, improve, tweet, bounty, etc. in order to get a reward for desired effort I already did in the past, that just needs a bit more tweaking? Which answers from the past can be edited, improved a bit, etc. in order to make them worthy of more votes?

So, without tracking, I can't identify items from the past that are candidates, but, I can still identify items in the present. I can find crap questions now, and answer them? But, that puts me dangerously close to user-2 type behaviour... just answer crap, and hope it comes right. Perhaps I will make an improving edit to the question, and make it half-decent. But, to be properly a type-3 user, I would need to keep a log somewhere of questions that need nursing, or answers that need improving.

That 'nursing' part is the most important part. Consider AirThomas's meta question on Stack Overflow where he admitted he put a 200-rep bounty on the question to try to give it a "second first impression". I see nothing wrong with that, but, how did he know which question to bounty? By keeping a 'log' (sure, of only 1 question).

This badge will require people to monitor their past activity to identify which questions are candidates for nursing, and where to spend their extra effort. It will require that future activity is directed to the right sorts of questions, and that the user 'owns' the quality of both the question and the answer.

Without the tracking, it will just encourage fire&forget answers on crappy questions in the shotgun hopes that some (one) will just miraculously happen.

General tracking comments

Different people are motivated by different goals. I personally am motivated by a target, and like to measure my progress. Targets that have no measurable progress tend to demotivate me, and I pick a better target to aim for. Socratic is a badge that does not really mean much to me... it's too hard to measure progress, and there's too much uncertainty. I can't see all my deleted content, so I can never measure how far away the goal is. Additionally, it is just too hard to do the math on 100 questions to see id any were asked on the same day, etc. Not even the data explorer can help with that badge.

Badges that are intended to reward ongoing beneficial user behaviour should all be trackable so that users can monitor when their behaviour is helping, and when to adjust their course.

In regard of some of the existing answers complaining about this badge encouraging answering bad questions and similar, how about the following:

To receive the badge, you must:

• When the question is at −3 or worse: Post the answer, perform a (major) edit or leave a comment that later receives n upvotes to the question.
• Answer the question within m hours of the above (m = 12?), with that answer scoring +5 or better at some later point.
• The question scores +3 or better at some later point.

This way, if a question is actually bad and you work on improving it, e.g., by posting a comment asking for the important clarification or editing it such that the actual underlying good question is visible to all (and later answer the question), this gets rewarded. If a question is actually good and people fail to see it, you do not directly need to answer (which may be impossible, if the question is closed), but you can also make people aware of the question’s qualities in an edit or comment.

The criteria for the badge seems to be a great idea and I think it should be made into a badge.

Firstly, should this be a new badge?

Yes, I don't think that it should replace the existing Reversal badge. Though the flaws of the Reversal badge has been brought up, the idea of the current Reversal badge isn't necessarily bad.

As Tim Post pointed out in this answer,

It's occasionally awarded for someone giving a bit of a snarky answer to a poor question, but most that earn it worked hard for it once they saw an opportunity to get one.

Since the Reversal badge is awarded to the answerer, I don't think that there's anything wrong with occasional bad questions. The main purpose of Reversal is to reward and credit the effort of the answerer who tries to provide a good answer to "not so good" questions.

Secondly, should this be a gold, silver or bronze badge?

The hat of this criteria was the most difficult hat for all 3 Winter Bashes that includes it.

• In 2014, this was a non-secret hat known as Red Baron and was awarded 26 times.
• In 2015, this was a secret hat known as Flying Tiger and was awarded 11 times.
• In 2016, this was a non-secret hat known as Maverick and was awarded 17 times.

So, this concludes that the new badge is quite difficult to earn. However, since it's slightly different from Reversal, I think it should be made a stand-alone gold badge.

It wouldn't make sense to make it a "silver version" of Reversal since a user can get Reversal but not this badge due to the "question later becomes +3" criteria.

So,

This badge idea seems quite difficult to earn as seen in the other answers. And it also seems to be the Reversal badge except with a less strict answer criteria and a requirement that the question becomes +3 score. Thus, Reversal should be replaced by the new badge... actually, let's change Reversal's criteria to this new criteria:

Post an answer with a score of 5 or better when the question was at -3 or worse, then rose up to a score of 3 or better after your answer was posted

I'm sure the description could be shortened a bit more of course. Why should we only change the criteria of the badge rather than replacing it all together or making a new badge? One reason is because the answer is causing a reversal of sorts to the score of the question, from negative to positive. That's nearly literally the dictionary definition:

a change to an opposite direction, position, or course of action.

The badge should be gold of course due to the rarity of this achievement. This badge is not that much harder than the original Reversal since you don't have a time limit of 3-4 weeks to get the badge/hat. Some extra improvements to the badge description is that if you go to the badge's page, there would be two definitions of Reversal, the old one (before the change) and the newer one (after the change) so users won't be confused why some users have Reversal even though some didn't get the question to +3 or better like this:

Before {date here}, the requirements for Reversal was to post an answer with a score of 20 of higher to a question with a score of -5 or lower.

After {date here}, the requirements for Reversal are now {new requirements here}.

One of the requirements for the Red Baron hat includes answering the question while the question's score is still -3 or lower.

I don't think this is good. Many times, downvotes signify that a question is unclear/not asked very well; while many people do it, I strongly believe that questions should not be answered while they are still unclear. I think the question should be edited first so that the intent is clear, and answered only after it's clear.
Therefore, I think that it's a Bad Idea™ to make it a badge requirement that the answer must be posted while the question score is still negative.

I propose changing the badge requirement to questions that have had a -3 score at some point in its history, and answered at any point, then being edited by the answerer, with the question score eventually coming back to +3. I think that will give us the benefits of this badge, all the while not encouraging people to answer questions while they're still unclear.

This creates a certain amount of overlap with the Explainer family of badges, but I think if we're going to implement this badge, it's okay, so long as we don't encourage people to answer unclear questions.

(Jon Ericson said that "The hat didn't encourage answering unclear questions." I didn't read through the revisions of all the questions linked to in the question, above; I'm merely writing my thoughts in response to the proposal itself.)