What should be done about answers posted as comments?

[pnuts]

tl;dr

Once an a “A” in a Comment has been established as a suitable solution a ‘proper’ A (including attribution) should be posted, preferably by whoever posted the Comment but otherwise by anyone prepared and able to do so. Later the Comment should be deleted. There should be NO penalisation.

The (very) long version:

Over the years I posted many thousands of Comments on SO (of which just 8 remain). Almost all of these either disappeared along with the Q or A, or were deleted by me. Amongst them were many of the kind:

" Try X. "

(the second of Lundin’s four bullet points above). Often I could have said:

" If the dates you mention are text format try “X”, otherwise try “Y” – and by the way, because of automatic coercion (see https://// ), it is too late now to expect to change the format simply by selecting your date data and HOME > Cells > Format > Format Cells… > Number > Date for Category and something like “14/03/2012”. What you would have to do instead is Z (see https://// ), or speak to the person providing your source data to get them to change their output. "

…had a Comment allowed me more characters.

I think the theory was that I should have asked for clarification in a different way. Say:

" What is the format of your dates? "

And then all the follow up to responses such as " American style. ", " Some are bold. ", " The years are only two digits. ", etc. , and, of course: " I wasn’t sure but I have just formatted them as Date and no change. "

Leading in turn to: " Is your data left justified? ", " Which version of Excel? ", " Are you seeing little green triangles? ", " Was the data scraped off the web? ", " Have you checked for trailing spaces? ", etc.

But " Try X. " is a lot simpler. This is less confusing for OP and left me more time to help others. I consider that a Win/Win, especially when, whatever my “X” was, it seems it did the trick at least three times out of four.

Why are there no “Try X.” type comments of mine remaining? Well, one reason is because I tried to ensure that no Q (that remained on site) in my areas of interest was left without an A (as such). I regularly returned and posted a ‘proper’ A (if nobody had done so already) once a solution had been found. This I think is what Monica has suggested as the correct approach. Naturally, I strongly approve. And, it may be worth noting, I often later posted a ‘proper’ A where clearly a suggestion in a Comment had worked and whoever wrote that Comment had not revisited the site (or been absent a very long time) and not posted their own A to the Q concerned. (In such circumstances I usually remembered to make ‘my’ A a wiki and gave attribution.) In some cases I merely copied the Comment, in others I elaborated a little. Sometimes I also edited the Q to include key facts that had only become evident through the interchange of Comments.

Actually, since the issue is so often one of identifying what the problem is, rather than how to solve it, the Gardening & Landscaping site is a good example, since this has a tag [identification] (it is the most popular one there at present). In my short spell of activity on that site I offered many suggestions for plant names. Occasionally as a ‘proper’ A, for example Lantana is hard to confuse with anything else, so I offered a ‘proper’ A (that commenced " This is distinctive enough for me to offer: "). More often however as a Comment of the kind: " Maybe this is Z ". I think I was right more often than I was wrong and there were indeed several complaints that I was “Answering in a Comment.” In fact my last two actions on that site (other than tidying up my other Comments there and offering some bounties) were two such Comments. Both disappeared very rapidly and I assume because considered attempts " to circumvent the will of the site by posting an “answer” as a comment. "

But some plants can only be differentiated with close-up, even microscopic, photographs of key differences. And distinguishing features may include the texture of leaves, the smell, and the growth habit. Even when at least one picture is available (and that was not always the case!) it does not provide those clues, even if critical. Pictures were often not very sharp and frequently did not indicate scale. Veining and leaf edges were frequently not shown clearly (but could make all the difference). Rarely did images follow the plants through the seasons, so flowers might not be shown. Whether or not deciduous could be unclear. Habitat, soil and climate conditions were not always provided.

HOWEVER, a guess that gave the OP some idea where to look could be all they needed, with a little research and what else they knew about the plant that they had neglected to mention. They could look up and see, say, that the leaves of the suggested plant turn yellow in autumn but remember that their plant’s did not. Or they might check and confirm, yes my one really does smell of toothpaste! And so forth.

OP’s in general seemed happy with my Comments but they ceased to receive any suggestion from me after my last two Comments were deleted. I think that is a shame and that it would be a shame to penalise “As in Comments” on Codidact IN ANY WAY. Punishing the expert for deficiencies in the Q is no way to retain contributors.

[manassehkatz]

@pnuts - I agree 100%! What you describe is not the typical “here is how you should do things on SE” recommendation, but I often do something similar. Actually I do a lot more commenting now because I have imposed a “no Answers” on myself in protest of the “situation” yet still can’t resist trying to be somewhat helpful by way of comments. But even before, when I would actively answer, I would often hedge my bets a bit by commenting first, seeing what kind of response I got, and then Answer if it looked like my idea could be turned into a quality answer. And I also try to delete my own comments if they are no longer relevant (whether I posted an Answer or not).

[Olin]

@manassehkatz

I would often hedge my bets a bit by commenting first, seeing what kind of response I got, and then Answer if it looked like …

This is exactly what we don’t want people to do. Even on SE, the right method is to post an answer. I’ve found that if you’re clear about the answer not being definitive or directly answering the question, you generally don’t get dinged for it. Of course, if you’re suggesting is outright wrong, it will get downvoted, as it should. In that case you can simply delete it.

The problem with using a comment as you describe is that you might be answering a question that the community has decided should not be answered by closing it, there is no way to downvote for people that believe your suggestion is wrong, and it adds to the clutter below the answer, thereby making comments that are there for the right reasons more likely to be missed.

[cellio]

Reminder: this is a community issue, not a technical issue. There is a lot of discussion here, but maybe it’s time to stop since it doesn’t affect the platform. Feel free to ask a question on https://meta.codidact.com, cast as how a community that sees this as a problem can address it. Also remember that comments in general will be less visible by default on Codidact than they are on SE.

[manassehkatz]

I think you’re basically right. This seems to be something that varies a lot between communities, including:

• How “complete” an answer should be before it is posted.
• What kinds of things are allowed in comments. (“requests for clarification”, “pseudo/partial answers”, “related-but-only-tangentially chit chat”, etc.)
• How much comments are cleaned up after a question is answered (or even after a clarification is added to a question in response to comments).

Since we (I think) all agree that there is a use for at least some comments, the rest is pretty much policy rather than technical.

[celtschk]

@Olin

This is exactly what we don’t want people to do.

Who exactly is “we” here?

[Olin]

@manassehkatz

Since we (I think) all agree that there is a use for at least some comments, the rest is pretty much policy rather than technical.

As long are there are mechanisms for enforcing proper comment use. Obviously mods need to be able to delete any comment, and users probably need to be able to flag them.

Some flags also need special handling by the software. For example if enough (number is site-specific) users flag a comment as “answer in comment”, then the comment should be automatically deleted. There should then be some negative consequence to the comment author, like a site-specific decrement to the commenting privilege trust level.

[pnuts]

@Olin

The problem with using a comment as you describe is that you might be answering a question that the community has decided should not be answered by closing it, there is no way to downvote for people that believe your suggestion is wrong, and it adds to the clutter below the answer, thereby making comments that are there for the right reasons more likely to be missed.

“you might be answering a question that the community has decided should not be answered by closing it”.

I have seen this (As in Comments to already Closed Qs) only very rarely. Though I have sometimes offered an A (or, more likely, a strong pointer – as Monica has said, there was rarely room in a Comment for all I would like to have written in such cases (though very many of my As were one-liners prefaced just with "* Please try:* "). They were all instances where (i) I thought the OP did not deserve closure and (ii) I considered myself likely much more of a “subject expert” than whoever were the ones downvoting (eg when I was the only user with a gold badge for the main tag). From the OPs point of view, getting an A regardless should be some consolation for being Closed unfairly and, for everyone else, maybe the OP will learn (by example, obviously Help was not enough) for next time that for a full A they need to write a ‘good’ Q.

On the other hand, if the Closure were deserved then, as it stood, the Q itself was just clutter – and decluttering that (via Delete votes or Roomba) automatically deleted all associated Comments anyway.

It seems you did not mention the situation where the Q is still Open at the time but later Closed, so I am not addressing that here.

“there is no way to downvote for people that believe your suggestion is wrong”.

In my experience I doubt I have seen a single case where such a suggestion was clearly wrong. When not a likely duplicate and not for one or other of Lundin’s last two bullet points, most often the “A in a Comment” (to an Open Q) is because the person trying to help is not quite sure what the issue really is. That is, the Q is not as clear as it should have been (often also exactly why Closed Qs with “As in Comments” were Closed). Where there is room for doubt about what is being asked then, pretty much ‘by definition’ there is not enough certainty to justify downvoting an attempt to assist anyway, in my opinion.

If a suggestion in a Comment is wrong then yet another Comment giving such warning should be effective and, in my opinion, worth the extra ‘clutter’.

“making comments that are there for the right reasons more likely to be missed.”

The only “right” reason on SE is to seek clarification of the OP’s Q, and the prime audience for that is the OP of the Q. I don’t see much harm (given the backup measures I have already mentioned in a separate post in this thread) in the OP failing to notice requests for clarification when that is because they already have their solution.

[pnuts]

It might help your case (at least with me!) if you could expand upon:

@Olin

For example if enough (number is site-specific) users flag a comment as “answer in comment”, then the comment should be automatically deleted. There should then be some negative consequence to the comment author, like a site-specific decrement to the commenting privilege trust level.

Maybe you have already explained, but confused me by not differentiating correct “A” from incorrect “A” (in a Comment):

• Q is Closed. With very few exceptions (eg historical lock) on SE sustained closure equates to “not wanted on site”. Hopefully therefore it will (with all its As and Comments) eventually disappear. Is there significant benefit in accelerating this process but only for Comments (correct or incorrect), even were it simple to achieve (and I’d not consider site-specific number of flags to be so)?

• No downvoting option. Why would anyone want to downvote an A in a Comment? The “A” is either correct, and easily moved to the proper place, or incorrect and eligible for a flag (for deletion as “No longer needed” (nor ever was!). I suspect already on SE if enough such flags are cast the Comment is autodeleted. (If not already the case, what makes “A in Comment” flags worthy of special handling?)

• Clutter. I am much in favour of reducing clutter [but prolix myself nevertheless]. I would be very pleased not to see incorrect As, whether or not properly classified. However, if the “A” is correct it is definitely useful – surely it is arguable more useful than other Comments to the same Q (which on SE should only be asking or offering clarification … which the correct “A” has proved no longer necessary). If really to reduce clutter, should not the Comments that don’t answer that Q be the ones to go (once the “A” has been posted)? “making comments that are there for the right reasons more likely to be missed.” Missing what is no longer relevant is good, arriving for the right reason is not the same as ‘needed in perpetuity’.

If not clear whether or not the “A” is correct then offering it can be an effective way to seek clarification, which is exactly what Comments (on SE) are for.

But, mainly, why penalise people that have offered a correct answer? In the wrong place does so very little harm and is easy to fix. Punishing people who know the correct answer and are prepared to post it is bound to discourage contributions from such people. What makes punishment worth that?

[tuggyne]

@pnuts

for everyone else, maybe the OP will learn (by example, obviously Help was not enough) for next time that for a full A they need to write a ‘good’ Q.

Very few OPs need or want “full answers”, so they are not at all likely to learn what you hope. They’re not the reason we want to have them. Full answers are for the vastly larger number of later readers.

An OP in this case will logically learn the exact opposite of what we wish: that they can get an answer that’s good enough for them without having to meet question quality standards.

@pnuts

The “A” is either correct, and easily moved to the proper place, or incorrect and eligible for a flag (for deletion as “No longer needed” (nor ever was!).

This entangles flagging with determining subject-specific correctness, making moderation dependent on specific knowledge in that particular question’s field. Generally not a good idea; as much as possible, moderation should rely on nothing more than site-general knowledge.

[pnuts]

With regard to this thread, here, I have just seen that Monica wrote “maybe it’s time to stop since it doesn’t affect the platform.” I’m not sure that proposals for a special flag, automated transfers, automated deletion of Comments when transferred, and automated punishments are merely a community issue but am not proposing to post any more to this thread (you will be relieved to hear). Please therefore do not feel any need to answer any of my questions in this post.

---

@tuggyne

Very few OPs need or want “full answers”, so they are not at all likely to learn what you hope.

I agree ‘need or want’ for their immediate problem. But I believe it is possible for a community to have good natured users who will realise that a quick fix in a Comment is not ideal (e.g. for later readers). The OP of a Q has no control (subject to sock puppets!) over how they receive an A. If it is in a Comment it is likely the Q was either a duplicate or for the reasons mentioned by Lundin earlier here. IF users are good natured [I tend to favour the Y of McGregor’s X and Y theories] they should, gradually, learn the sites mores and how to ask a good Q. I think most “As in Comments” are to Qs from new users, or those very short of experience. Depriving them of an A is not likely to encourage them to return, losing Codidact valuable contributors (as well as nuisance ones).

@tuggyne

An OP in this case will logically learn the exact opposite of what we wish: that they can get an answer that’s good enough for them without having to meet question quality standards.

Indeed that is very possible. But (based on SE) only for a short while. Repeat offenders (of bad Qs) were already dealt with (on SE) with bans. Surely a much more effective approach than depriving them of available answers.

@tuggyne

This entangles flagging with determining subject-specific correctness, making moderation dependent on specific knowledge in that particular question’s field. Generally not a good idea; as much as possible, moderation should rely on nothing more than site-general knowledge.

I fully agree with both of these two points of yours. But offer a simple solution: “if a mod has any doubt about the technical validity, delete the Comment.” I happen to believe that, for almost all the cases I have seen of As in Comments, there has been very little doubt over correctness. (Mind you, I was mostly looking at spreadsheet Qs and so many people have spreadsheets the chances of being able, easily, to try - if even necessary - a solution proposed in a Comment to a spreadsheet problem are probably relatively very high. But again, if in doubt, delete.) Yes, “Generally not a good idea” but with emphasis on the first word. I have found As in Comments often require exceptionally little technical knowledge to assess (as Lundin mentioned, frequently “so simple to answer”; easily be found “in RTFM”, or “at an external site”).

Consider a typo for example. Would you advocate keeping from OP that their answer is not to call, say, [patner] when it was [partner] they dimensioned? (Hence maybe just leave OP with " a problem that can no longer be reproduced or that was caused by a simple typographical error " to go on.)

Or an OBO error. A duplicate but nobody can be bothered to find of what. Should the only clue given to the OP be downvotes (no closure so no Close reason because no one has bothered to find, of what)?

Or unbalanced parentheses. Unlikely a Q of any lasting value. Merely downvote and let OP stew?

Would you ban a user who habitually offers correct "A"s in Comments. I wouldn’t, but if “yes” then there is your solution. If OPs won’t take a hint and become more inclined towards bad Qs I would happily ban them.

[klutt]

@pnuts

Indeed that is very possible. But (based on SE) only for a short while . Repeat offenders (of bad Qs) were already dealt with (on SE) with bans. Surely a much more effective approach than depriving them of available answers.

True in theory. Not in practice. You can register an account on an SE site and ask a question within a minute. I think there has to be more benefits in using an established accounts. I know it’s a bit tricky and all things have their pros and cons. And we don’t want to scare people away. Some ideas I have:

• When you post your first question, you’ll receive a message telling you that it will be published in 24 hours, and until then you’re encouraged to make sure you’ve done everything you can to learn the rules and improve the question.
• A clear drawback of the above is that then it’s VERY likely that the asker will not be around to answer comments upon publication. A way around this is to not autopublish win 24 hours, but instead tell the asker that they should come back in 24 hours to publish it manually.
• The first (or maybe first three) questions from a new user needs to be reviewed before publishing them. This has the potential of requiring a huge amount of work from the community.

In order for methods for dealing with “repeat offenders of bad Q:s” to be effective, there has to be something preventing them from just registering a new account.

[Olin]

@pnuts

It might help your case (at least with me!) if you could expand upon:

@Olin

For example if enough (number is site-specific) users flag a comment as “answer in comment”, then the comment should be automatically deleted. There should then be some negative consequence to the comment author, like a site-specific decrement to the commenting privilege trust level.

First, keep in mind that the purpose of this thread is to decide features that the software needs, not eventual site policy. Ideally, the software has enough tweak parameters so that a wide range of site policies can be implemented. The part you quoted discusses two possible mechanisms to support sites with strict “no answers in comments” policy.

The first is a flag that causes special handling in the software. When some number of these answer in a comment flags are cast, then the comment is automatically declared to be such and dealt with accordingly. The number of flags required to cause that would be a per-site parameter, with 0 disabling the mechanism altogether. The most obvious action for dealing with the offending comment is to delete it.

The second point was additional mechanics if a comment was deemed to violate the “no answers in comments” rule, regardless of how that is done (by a mod, by flags, or some other mechanism not discussed here). This second mechanism provides a dis-incentive to break the rules. Without it, there is really no reason a user can’t continue to answer in comments. As SE has abundantly shown, if there is no downside to doing something, people will do it regardless of the rules. The proposed downside in this case would be for the offender to eventually lose the commenting privilege and/or lose some rep. Again, the exact amount some privilege level or rep is decremented would be per-site parameters.

[klutt]

@pnuts

With regard to this thread, here, I have just seen that Monica wrote “ maybe it’s time to stop since it doesn’t affect the platform . ” I’m not sure that proposals for a special flag, automated transfers, automated deletion of Comments when transferred, and automated punishments are merely a community issue but am not proposing to post any more to this thread (you will be relieved to hear). Please therefore do not feel any need to answer any of my questions in this post.

You raise a very good point here. In order to discuss what the platform should support, we also need a picture of what the policies on the different communities would look like.

So I think that such discussions are needed, but it’s also important that they don’t take over.