Too many upvotes for poor quality answers

Question

I was looking through the Stack Exchange hottests questions and stumbled on one of the Programmers questions:

When do you start documenting the code?

Because I'm reading Code Complete and I hoped to find another insightful post to help me decide when I should be doing this. Only to find an answer like:

If you don't comment the code as you write it chances are minimal that you will come back and do it later (let alone do it properly).`

Honestly? This answer only tells the OP he should do it, because else he has to come back and do it anyway. No explanation about how writing your code as pseudocode, which can be converted to comments might lead to fewer errors. Or how commenting what you write helps you think harder about the code you just wrote. Nothing.

Why does this answer have 25(!) upvotes? That's a lot of rep for something that doesn't answer the question and clearly doesn't teach anything to users who are interested in it.

Now if this were just incidental it wouldn't be a problem, but I stumble over your 'hot' questions a lot and more often than not I've come back disappointed. Don't believe me? Here's another example from the top 50 Laptop ergonomics: How do you position yourself when working on a laptop for long hours?

Normally, like I would if I was working on a desktop computer. Regular desk, a sturdy wooden chair (I don't like ergonomic chairs or anything of the kind) and a normal posture. Worked for me. YMMW though!`

Nothing about using an external keyboard, a mouse rather than a trackpad, using a laptop stand for a better viewing angle or height? There's a whole industry devoted to this and there's actual science behind the ergonomics, yet an answer that says: 'Worked for me' gets 5 upvotes.

Now I don't have a solution to offer, but as a fellow Stack Exchange user I think upvoting such answers is giving a bad example. Worst of all, legitimately interested users like myself are repelled from even looking at Programmer questions any more because of the bad experiences, which I hope isn't something we're striving here for.

Should we be rewarding poor quality answers like this?

Answer 1

Respectfully, the problem with your first cited example may not be the answer, but the question itself (which the top-voted answer answers cleanly, succinctly and correctly).

Ask a specific question, get a specific answer. Ask a general question, get a general answer (which, by its nature, will attract lots of upvotes).

This is why we try to encourage people to write questions that are specific and answerable.

Answer 2

I've seen something similar to this (on stackoverflow.com) even for very technical, answerable questions. The problem is that the really hard questions either don't get answered, or when they do the general community doesn't care or doesn't recognize that an amazing answer has been posted. Because you aren't going to upvote something you don't understand yourself or aren't involved with. On the other hand, when someone asks something like 'why are singletons bad practice?' or 'why does jquery use the $', everyone wants to either jump in and answer or at least upvote something that sounds about right. The result is that reputation is awarded to popular, general, easy, and well-known questions/answers.

Having a high reputation is probably a function of cherry-picking easy questions and not (necessarily) being a guru.

Not sure how to "fix" this sort of thing, but it's definitely been laughable to me when I answer a difficult question and get maybe (maybe!) 1 upvote. And then later I'll answer some trivial c++ question posted by a complete beginner. And, because I happened to be the first person to see it, I end up with like 10 or 12 upvotes.

Answer 3

Why does this answer have 25(!) upvotes?

Because it is:

1. Correct. It may not be a 20 page treatise on the subject, but you cannot deny that it is ultimately correct.
2. Succinct. Some people prefer answers that are brief. Indeed, I sometimes find that overly flowery and highly detailed answers are less useful than a quick sentence that points the person in the right direction. Long answers are not better even if they are more detailed.
3. Sufficient. Any of the techniques you discuss ultimately involve something that has little to do with comments themselves. Those are general coding techniques; you could just as easily keep the "comments" in a separate text file or even on paper. This answer provides exactly what the user asked for: the time to start commenting. Namely, as you write the code.
4. Flippant. For many programmers, the answer is so obvious that even asking the question shows that the person didn't spend much time actually thinking about it. And therefore, a flippant response is entirely appropriate. Generally, quite a few people will upvote flippant responses to seemingly dumb questions.

Note: I'm not saying I agree with any of these reasonings for upvoting. But all of these are properties of the answer.

Are there times when a poor answer is upvoted? Absolutely. So what? It happens. You deal with it and move on.

You cannot make people agree with what you consider to be a poor quality answer. Everyone has their own idea about what constitutes a good answer. Some people only consider multi-paragraph essays to be good. Others feel that if you can point someone in the right direction, that's a good answer. And so on.

You need to accept that other people are not going to agree with what you consider to be poor quality.

Answer 4

Should we be rewarding poor quality answers like this?

The problem is determining if the quality is poor.

This answer only tells the OP he should do it, because else he has to come back and do it anyway. No explanation about how writing your code as pseudocode, which can be converted to comments might lead to fewer errors. Or how commenting what you write helps you think harder about the code you just wrote. Nothing.

I don't think that the quality of the code documentation/commenting answer is poor at all. The asker wants to know when he should be documenting code. The other information, although true and possibly helpful, does not necessarily answer the provided question. Should it be included? I would probably include it. Must it be included for an answer to be good? No, as long as the answer addresses the question asked.

If you re-read the question, the person asking doesn't ask anything about how to reduce errors (which is a good thing) or thinking harder about design and implementation (another good thing). He just wants to know when you should document your code. The answer says when you should, and says why you should write it at this point in time.

Would I include more information, about producing pseudocode as a form of documentation, for example? Yes, probably. But that doesn't explicitly answer the question. I would consider upvoting this answer (although I typically read all the answers, and upvote the top couple).

Nothing about using an external keyboard, a mouse rather than a trackpad, using a laptop stand for a better viewing angle or height? There's a whole industry devoted to this and there's actual science behind the ergonomics, yet an answer that says: 'Worked for me' gets 5 upvotes.

As for the ergonomics answer, that is a little lacking. More details about the environment would be extremely helpful. However, it's not a bad answer at all. The answerer should discuss his entire work environment - desk, chair, keyboard, mouse, monitor size and position, and so on. Another good answer would link to research or studies on ergonomics.

The "works for me" approach is common here and Programmers, and it should be given the nature of the majority of the questions. However, the answer you mention doesn't really delve into why it works for the person writing the answer, which is key. Without the why, it becomes harder to tell if it will also work for me.

I wouldn't upvote this answer, but I also wouldn't downvote it.

Answer 5

I've noticed conflicting and uncomfortable attitudes toward the Programmers users. On one hand, we need users to generate the content that makes the site interesting and enjoyable. On the other, we don't trust those users to make good decisions. When you ask, "Should we be rewarding poor quality answers like this?", you question the legitimacy of more(assuming downvotes) than 29 users' opinions. You're bypassing the voting process and appealing to Meta's perceived higher sense of quality.

I don't blame you. I, too, find many Programmers questions and answers poorly written, irrelevant, obnoxious, or just plain stupid. Despite that, I recognize in a democratic system there are bound to be times I'm in the under-served and dissenting minority. My disagreement doesn't make everyone else wrong. The democratic dynamics are even harder on Programmers (compared to other StackExchange sites) since its questions tend to elicit answers that are more persuasive exercise than objective truth.

I can put up with highly-rewarded, weak questions and answers. What I'm more afraid of is a small group of highly-motivated, well-intentioned users controlling the site's policies through Meta advocacy and aggressive editing. I know all users are welcome on Meta, but relatively few participate. These non-participants show approval and disapproval with their votes. It would be a shame to diminish the value of those votes with new rules or policies. Users won't participate if their input is ignored or they sense condescension.

Answer 6

How about an "I Demand a Recount!" feature? It could work like this:

• After X users indicate the question/answer landscape has changed enough to warrant a recount, the question goes into recount mode.
• Each user who cast a vote is notified and encouraged to revisit the question.
• For X days, all votes are unlocked and may be modified/reallocated to better reflect the new question/answer landscape.