How to deal with an accepted answer that seems completely incorrect?

Question

On the question Why aren't average speeds computed over distance?, there is an accepted answer: https://bicycles.stackexchange.com/a/40672/8273.

The answer is written by the OP, in such a way as to give the impression that the answerer and the OP are different people, so it took me a little while to realize. Móż pointed it out also.

The problem that I see is that the question asks, in essence, why a given technical term does not have different way of calculating it. Many users have given answers that essentially explain why it's the way it is. The average is the distance / time.

People have perhaps been misled by the question in the title. The body of the question asks something quite different. It's a rambling discussion that ends

I don't care about time anywhere here (minus speed) -- the definition of average speed doesn't bother me, I'm just wondering why this isn't a more widespread metric.

In retrospect, I think this question should have been flagged as unclear what you're asking, and closed.

The current issue, as I see it, is what to do with this answer. In my view, it takes a straight forward mathematical calculation and coerces it into a statistical framework, which is then miss-applied to support the OP's thesis. In my view it's factually incorrect.

In my understanding, this site is meant to be striving to provide the best possible answers to answerable questions. The answers to some questions are a matter of opinion. Many answers cite experience. But overall we strive for the best possible answer.

In my view a factually incorrect accepted answer should not be on a SE site.

So the question is - what to do. Delete the accepted answer? Get off my soap box and ignore it? Delete and close?

Answer 1

Writing generally here..

No, I don't think you should act as some kind of censor even under these circumstances. Leave your powers of deletion in the box.

For me it would boil down to how fussed you could be. Ideally, you'd write a competing answer and rely on the community to upvote/downvote, such that over time one would expect the answer's score to reflect its accuracy. Or, you could simply vote the thing down, to register your disapproval, and hope other users do the same. I think we're quite lucky on Bicycles, as there is a decent pool of experienced users, and that we don't get a lot of material posted which falls into the "totally incorrect" category.

But the whole thing about SE is the community ethos, and I wouldn't really be comfortable using "privilege tools" to censor what other users can see, even if that means you end up with dumb answers.

Answer 2

I think this situation reflects the fact the OP doesn't have a clear understanding of what they are trying to ask.

I suspect the OP is actually interested in power (measure of performance) that should be largely independent of environmental conditions. Rather than realizing that, the OP is trying to use instantaneous speed as a proxy. Looked at factors influencing instantaneous speed as being "random" and therefore their realized speed as a random variable drawn from a distribution of possible values, where the central tendency of the distribution represents some theoretical such as OP's instantaneous speed in a vacuum on flat land. The OP, then went about trying to estimate this using different ways to sample instantaneous speed, then fit it distribution to it (more on that below).

In retrospect, I think this question should have been flagged as unclear what you're asking, and closed.

If my interpretation is remotely correct then the original question is not clearly worded and Andy's original proposal (closing as unclear) would be the logical action.

So the question is - what to do. Delete the accepted answer? Get off my soap box and ignore it? Delete and close?

I don't think deleting the answer is the right decision as SE doesn't guarantee that an accepted answers are empirically or factually correct, only that it is suitable. In this case because the OP accepted their own answer there is no reputation awarded and the answer does not float to the top of the list.

---

Statistical Aside

The OP's statistical approach is problematic to say the least. I also think the OP is conflating (or not distinguishing) between process (i.e., riding a dirty bike one day or riding up mostly hills another day) and observation error (e.g., gps error). The process error will cause covariance between observations and likely sampling bias. When estimating a PDF (probability density function) from these raw observations, most techniques will assume IID (Independent and Identically Distributed), which will not be the case. Even if you could estimate the PDF, I am not sure what you gain as you will likely use its expected value, which is a fancy way of saying the mean value the random variable will take.

Answer 3

I think deletion is appropriate. It is too easy for answers to get upvotes, even on small sites like this. So while in theory wrong answers will get downvotes, as we saw in this case that doesn't always happen, and often happens very slowly. You'll note that even this wildly incorrect answer that disagreed with every other answer still took days to get even one downvote, and is on -3 right now.

This is perhaps not the ideal example, because Bicycles.SE is a site where safety questions do come up. I hope that no-one here would argue that we should leave up "should ride on the freeway" with an accepted answer like "of course, bicycles are vehicles, you need to assert your right to the road". It doesn't matter whether it's sarcasm, it only takes one more idiot and things could get really quite ugly.

I am all for leaving in place opinions that are weird, or just unjustified, if the OP likes that answer. We have a few of those.

Somewhere in the meta.stackoverflow discussion of the reviewing is a series of questions that touch on this. I can't find it now but I was reading the other day. Their problem was that even while reviewing atrocious answers, those answers would get upvotes. Even spam gets upvotes. There seemed to be a consensus that deleted the worst of it was not just a good idea, it was necessary.

edit: As with the "don't edit spam" theory, what works when there are 5 active mods and 1000 users an hour, often takes days here or does not work at all. So the solution of "if you just wait a week the bad answer might get a downvote" really isn't a good solution. There are a few times when looking at search results that I've found old answers that are blatantly wrong but have, say, +3 votes, while the correct answer is still on zero. In those cases I just have to vote -1 and +1, then edit something so that the question moves to the top of "active questions" and hope that other people vote to fix the ranking problem. Sometimes they make it worse (in which case the charitable explanation is that they're just blinding upvoting the top answer).

Answer 4

In my understanding, this site is meant to be striving to provide the best possible answers to answerable questions. The answers to some questions are a matter of opinion. Many answers cite experience. But overall we strive for the best possible answer.

I don't understand the site to work that way at all. The answers generated are a direct extension of knowledge the user base. For any particular site (especially those were the majority of users are not industry professionals) there may/will be large gaps in knowledge of the subject matter. StackExchange does not have a teacher or judge that awards points for the "correct" answer. In many cases there may not be a correct answer. Rather, we provide only the answer that garners the most points for the answerers. In this case it's completely appropriate to delete an answer that has been largely downvoted and was obviously shady to begin with (self answered and all). However, the factual inaccuracy has nothing to do with the deletion. Had the answer been moderately upvoted or most popularly upvoted, it should have stayed.