# Should we discourage the use of multiple redundant @username pings in comments?

EDIT $\$ There is now an "experimental" SE software revision blocking comments with multiple @name's, see Jeff Atwood's answer below.

As was mentioned in this recent question, only the first @name notification works in comments (besides those recipients notified by default, e.g. the author of the question/answer). However, I recently learned that some folks are explictly using redundant multiple @name's to emphasize to the author that the comment is also explicitly directed at the author - even though the author is always pinged by default, e.g. see the comments here.

This has the following very unfortunate consequence. New users see these multiple @name pings and they assume that this implies that such syntax works generally to ping as many people as they desire. When they don't receive replies from the non-initial pingees (who were not notified) the newbie may make false negative inferences, e.g. my comment wasn't interesting, the pingee isn't sociable, or perhaps even worse. I've encountered this on at least a few occasions, where folks wondered why there was no reply to their never-received ping. No doubt many other cases went unnoticed, leading to who knows how many false negative impressions (alas, comments are not searchable, but perhaps a database dump can be employed?).

Here's another example. Suppose that you posted a difficult problem that has been bugging you your whole life. Nobody answers it. A year later a newbie expert adds a comment to a related question with a link to a paper that completely solves your problem, thinking that there is no need to duplicate the comment in your post, since he pinged you already. But in fact you don't get pinged because you are the second pingee in the comment. As a result, you may never know the solution to one of your favorite problems.

Obviously there are many other analogous problems resulting from folks mistakenly misled to believe that multiple pings work when in fact they don't. Half-implemented and/or broken communication systems lie at the root of many serious problems.

The correct fix would be to have the software warn about such incorrect syntax. But we all know how difficult it is to get the SE folks to make enhancements. Should we instead explicitly discourage use of this misleading syntax, and put a warning in some high-visibility place (faq?).

If folks desire to catch the eye of the author (in addition to the default ping) then why not simply begin the comment addressing the author by name? (without a leading at-sign). This achieves that goal without misleading others into believing that multiple pings work generally.

-
I was completely unaware of this, I had previously thought that multiple @'s meant multiple pings. –  Eric Naslund Apr 27 '11 at 1:21
You'll be interested in this feature-request: In the inline comment help, please mention that only one @reply is allowed. –  Hendrik Vogt May 10 '11 at 13:25
Can someone explain to me why, if my comment references ATuser1, ATuser2 and ATuser3, that all three can't be notified? Is this for technical or other reasons? I understand that if the back end doesn't support it, there is no reason to allow the syntax if it will mislead users into believing ATuser2 and ATuser3 will be notified when they aren't. But I'm not sure I understand why all three users can't be notified. –  Aaron Bertrand Jul 19 '11 at 17:02
@Aaron: The back end doesn't support it, but that's on purpose. The idea is that allowing multiple notifications would lead to too chatty comment threads. (Note that I'm only reporting here, not taking any sides.) –  Hendrik Vogt Jul 19 '11 at 17:11
@Hendrik so the workaround is for me to post two separate comments if I want to make sure both people notice my comment. That doesn't seem less "chatty" to me. :-) I'm not sure why extended comments are such a bad thing, and users are encouraged to move to chat. What's the difference to the site administrators? As a user, I would rather see the evolution of a conversation than an edited answer and try to figure out what discussion led to the changes (including chats I may not be privy to). –  Aaron Bertrand Jul 19 '11 at 17:16
@Aaron: I agree with what you're saying. (As I said, I was only reporting.) I also prefer comment discussions to chat discussions if they are closely related to a particular post; then you have the background of the discussion at hand. The point seems to be that they're trying to discourage lengthy comment discussions to keep the Q/A content of high quality. –  Hendrik Vogt Jul 19 '11 at 17:20
@Hendrik yes, I'm not trying to convince you, just expressing my opinion for anyone who is reviewing this. :-) I agree with the theory about keeping quality of Q/A content high, but then why allow comments at all if they're so dilutive? Because they often add a heck of a lot more context than the original or edited answer. Besides, at least on SO, they only show the top 5 comments; the rest are hidden and only shown to people who are interested. I still contend that those conversations can often be just as valuable, if not more so, than the question and answer alone. –  Aaron Bertrand Jul 19 '11 at 17:22
@Aaron: Again, I can do no better than just agree :-) –  Hendrik Vogt Jul 19 '11 at 17:25
No amount of help text is going to fix this, because people don't read: codinghorror.com/blog/2009/10/treating-user-myopia.html –  Mark Ransom Jul 19 '11 at 22:30
I have to say it takes a fair bit of willpower on my part to resist those chat invitations... bleh. :P On the matter of having to keep @lerts separate per user, I came from a forum culture where multiple posts were actively discouraged, and my habit of trying to keep things in a single comment/answer is a carry over. –  Ｊ. Ｍ. Jul 20 '11 at 11:47

We are experimentally blocking comments that contain multiple @name mentions in comments now.

### Only one additional @user can be notified; the post owner will always be notified

If you wish to bypass this, be sure the comment contains a backtick .

-
Thanks, great idea in fact. Now it's indeed deployed, but in my opinion another tweak would be great. I just tried "@Bill, @Jasper: I'm testing Jeff's answer here" on one of Bill's answers here, and this was blocked although it would have notified both Bill and Jasper. I know that I could use a backtick, but I think in this case no blocking is needed. (And also here I need backticks :-() –  Hendrik Vogt Jul 19 '11 at 10:03
@Hendrik: Pinging you, just to make my point. @Jeff: Why are you blocking this comment? There are situations in which I need to say something expressly to Hendrik before I say something to you. Do you want me to submit two separate comments? Also, where do you want me to put the backtick?  –  t.b. Jul 19 '11 at 14:23
@Theo: That's exacty the point I was trying to make. Thanks for letting me know that I'm not alone :-) –  Hendrik Vogt Jul 19 '11 at 14:47
-1 : There are 14 votes on this thread for "there is no need to discourage multiple name pings" and 6 in favor of discouraging them. That seems like a pretty clear indicator of community opinion. –  Carl Mummert Jul 19 '11 at 14:48
@Carl If you discard the (double-dipping) downvotes, and count the votes on the question, then it seems there is no clear consensus on the matter. I doubt this current "experimental" implementation will persist since it will probably prove annoying to many folks. –  Bill Dubuque Jul 19 '11 at 15:42
@Bill: if you discard down votes (which you and I can see by clicking on the vote count) the ratio is 15 to 8, almost 2 to 1 against. –  Carl Mummert Jul 19 '11 at 15:45
@Carl But that's ignoring votes to the question itself which serves to balance things out. In any case I doubt the very small number of votes here will have much sway since this change will probably generate much more feedback on higher traffic sites such as SO. –  Bill Dubuque Jul 19 '11 at 15:50
@Bill: I wouldn't count the question itself after setting up two answers to explicitly let people choose their options, it seems like more double dipping. Even so, at best there is no consensus about multiple pings, which is hardly support for actively blocking comments. Separately, I don't view feedback on very SO as relevant to what should be implemented on this site, SE sites are supposed to be led by the community. –  Carl Mummert Jul 19 '11 at 16:05
@Carl Alas, because of the big delay between the question and the answers for voting, it's not clear how to interpret the results. With hindsight I would do it differently. –  Bill Dubuque Jul 19 '11 at 16:14
@Jeff: I am testing this blocking functionality on you.@Theo It seems that it wasn't blocked. I don't understand? Is it just because I didn't include them at the beginning? –  mixedmath Jul 19 '11 at 16:54
@mixedmath: I guess this is due to the missing space before @Theo. –  Hendrik Vogt Jul 19 '11 at 17:08
@Hendrik: That's exactly it! Very observant. –  mixedmath Jul 19 '11 at 17:10
@mixedmath: There was a reason why Theo Bühler called me the ping-master :-) –  Hendrik Vogt Jul 19 '11 at 17:15
@Jeff: In my previous comment, the new feature was indeed a bit annoying. –  Hendrik Vogt Jul 19 '11 at 17:24
@msh210: Exactly. The @ before the post author is never needed, but sometimes helpful. For example, several comments in this thread are not so interesting for Jeff. So I might want to use @ to make it clearer that I'm really addressing him. –  Hendrik Vogt Jul 20 '11 at 18:06

I disagree, there is no need to discourage multiple @name pings.

Please vote this answer up if you think that there is no need to discourage the use of multiple @name pings in comments. If you have a proposed alternative solution please post it as an answer that can be voted on independent of other issues.

-
Note: please don't downvote the other option too, please either upvote this or the alternative answer. It would also be helpful to leave comments as to why you didn't find the above reasons persuasive, so that I could modify the proposal if need be. –  Bill Dubuque Apr 26 '11 at 22:30
@Jasper Someone did downvote the alternative choice (and presumably upvoted this). That's why I added the above comment. I'm very curious to know why some folks don't find the above reasons persuasive. Perhaps I am missing some alternative viewpoint? I confess I am baffled. Communication systems that are broken but perceived as working can wreak havoc. I thought that such dangers would be obvious. Perhaps someone else can elaborate on these topics more convincingly than I. –  Bill Dubuque Apr 26 '11 at 23:15
I most strongly believe that multiple pings should be supported. If they are not, I think a warning that explains what will happen is a good idea. Mulitple pings are still a good way to indicate that your comment is responding to more than one other comment, or this answer and a comment (though I guess you could dispense with the @). You just have to hope the addressee will see it for some reason. –  Ross Millikan Apr 29 '11 at 13:24

Can I make a suggestion here that all SE sites include a generic correct use of the comment system in the FAQ to each main site? This will not just deal with the issue of multiple pings but other pinging issues as well, so this is more of a generic solution than a specific one to this question. It could be simply stated in a few words (or be provided as a link) polishing and summarising what is found at How do comment replies work?

When I first joined SE I was only aware of the ability to comment, not how the notification works.

Later I started to see strange @ symbols everywhere, with varying syntaxes.

Later I started to notice comments appearing in my inbox.

Later I asked myself how all this works.

Later I found the post How do comment replies work? describing this.

Later I realised how many were misusing it (including myself) and not realising it.

Later I posted this thread on meta.

Update: The post How do comment replies work? is now linked from the FAQ.

-
this is possibly a reasonable question to bring up on Meta.SO. It would be a good idea for some documentation of the @-notification system to be included in the standard boiler plate FAQ for SE sites, especially since it is something that concerns the basic operation of the site and effective use of the SE engine. –  Willie Wong Apr 27 '11 at 2:01
we are looking at it –  Jeff Atwood Apr 28 '11 at 7:23
@Jasper, @Willie, just to point this out: As of July 1, the second @lert will still notify a commenter if the first @lert was invalid or directed at the post owner. –  Hendrik Vogt Jul 14 '11 at 11:41
@Hendrik: thanks! (And I see that you used your comment as an example.) –  Willie Wong Jul 14 '11 at 13:51
notice that the "help" link under [Add Comment] contains quite a bit of help now. And this help is pre-expanded for new users... –  Jeff Atwood Jul 19 '11 at 9:07
I would prefer making the system more understandable to throwing more help text at the problem. Sorry to repeat myself, but I think this is required reading: codinghorror.com/blog/2009/10/treating-user-myopia.html –  Mark Ransom Jul 19 '11 at 22:46

I'm the one who explicitly used a redundant @name in the comments here cited in the question. Personally, I don't see a real problem in this: People use multiple @name anyway, not knowing that it won't work, so if I do it knowing what I'm doing, it won't increase the number of comments with multiple @name substantially.

My suggestion: If someone uses multiple @name, and you suspect they don't know it doesn't cause multiply notifications, leave a comment saying this. This will educate not only this one commentator, but also other people reading the comments. To make this easier, it would be best to create a faq post here on meta explaining things. Then in a comment one can just link to that faq post.

There's a reference post on meta.SO, and also a post on meta.tex.sx on that topic.

-
I think you may be missing my point. Newbies learn by osmosis. When they see you using such syntax they will assume it works generally. When they use it as such there may be no experienced users who notice such, and that may lead to false negative impressions, as described above. Why take this risk for such a small gain? –  Bill Dubuque Apr 26 '11 at 19:16
I think you may be missing my point :-). Newbies learn this by osmosis, too, if others use multiple @name by mistake. There are so many things about comment notification that go wrong that you need pointers anyway. Yes, I'm increasing the risk a bit, but I'm willing to take it. If, however, my post gets seriously downvoted, or other people leave comments that I shouldn't do it, then I'll never do it again on (meta.)math.sx. –  Hendrik Vogt Apr 26 '11 at 19:20
I don't understand your claim. How is a newbie going to learn by osmosis that the non-initial @name didn't notify the recipient? (possibly leading to an important remark being missed) –  Bill Dubuque Apr 26 '11 at 19:25
@Bill: That's definitely not what I'm claiming, no. I'm saying that people are using multiple @name by mistake anyway. So even if I don't, new users can still see that others did it, and they can still believe that it works. I just don't see that me not doing it will make much of a difference. –  Hendrik Vogt Apr 26 '11 at 19:31
@Han But I think they are using them by mistake because they see other users doing so. If we discourage that behavior then this should stop, which would go a long way towards preventing some possibly very negative scenarios. Why do you think it is so important to doubly notify the author. They get the ping either way. –  Bill Dubuque Apr 26 '11 at 19:37
@Bill: Right, other people making mistakes may make you follow and make the same mistake. But it's very easy to make mistakes without that guidance. So it will never stop, new user will invent that mistake over and over again - I'm pretty sure of that. –  Hendrik Vogt Apr 26 '11 at 19:41
@Han I'm not so sure about that. If they only ever see comments using one @name in active questions then they might infer the proper syntax, esp. if its mentioned in the faq or elsewhere. –  Bill Dubuque Apr 26 '11 at 19:49
@all; As you can see, I don't have a problem with people using the @ sign to indicate to whom their comment is made; this is irregardless of its use in sending messages. –  Carl Brannen Apr 27 '11 at 2:55
@Bill: Just to point this out, my name is H_e_ndrik. And I like it most if I'm addressed with @Hendrik, so my full first name :-) –  Hendrik Vogt Apr 27 '11 at 8:24
@Jasper: If you find a way to emphasize one letter within one word, please let me know! (I expected it not to work, but left it like this.) –  Hendrik Vogt Apr 27 '11 at 11:59
@Hen Sorry about the typo. Usually I type only the first three letters of names that are not programmed into my typing repertoire (min needed for match) to avoid typos. –  Bill Dubuque Apr 27 '11 at 13:05
@Bill: No problem, and I like @Hen much better than @Handrik :-) –  Hendrik Vogt Apr 27 '11 at 18:44
Leaving a comment to teach someone how to properly use the site just increases the noise level and distracts from the relevant comments. I would strongly discourage this and look for an alternate solution. –  Mark Ransom Jul 19 '11 at 22:42

I agree, multiple @name pings should be discouraged.

Please vote this answer up if you think that we should discourage the use of multiple @name pings in comments due to the mentioned problems. If you have a proposed alternative solution please post it as an answer that can be voted on independent of other issues.

-

It seems to me that the solution is simply to use a different character than "@" when we want to indicate that our message is aimed at a particular user, and we don't care whether they are pinged or not. I propose !, as this is already used on IRC in a similar way to mark messages aimed at a certain user.

For example, if someone wanted to show a comment is for me, they can write "!Carl Mummert" at the beginning, using the same number of characters as the "@Carl Mummert" syntax without invoking the messaging system.

-
Rather than using some obscure convention why not simply say "at Egbert" or "to Edwin" or anything else that is obvious? –  Bill Dubuque Jul 19 '11 at 14:50
I fear that it won't be too long before any mention of a superfluous name will be stripped out of a comment, just to keep the internet "clean". –  t.b. Jul 19 '11 at 14:51
!Bill Dubuque: One reason is that way takes two extra characters. If several people start using the ! convention, it won't be obscure. We could even explain it in the FAQ: "You can only ping one other user, using an '@' symbol. If you want to point out the comment to other users, you can use the '!' symbol, although this will not cause a notification to appear in their inbox". –  Carl Mummert Jul 19 '11 at 14:52
@Theo Perhaps you might be more understanding if you missed a very important remark due to not being notified by such a multiple ping - a not too unlikely scenario mentioned in my question above. –  Bill Dubuque Jul 19 '11 at 14:54
@Carl But I think that has the same problem as the original syntax. Folks may be misled into thinking that it is specially parsed by the SE software, something that probably would not occur if one instead employs natural language. As is clear from many discussion here, many folks don't read the fact, instead making (possibly false) inferences by lurking. –  Bill Dubuque Jul 19 '11 at 14:56
@Bill: it seems to me the best solution would be for the system to follow the principle of least surprise and ping all the users who have an "@" before their name in the comment. That would avoid the problem of people not being notified, which I agree is a problem but which is the same whether we write "at user" or "!user" or anything else that does not ping them. –  Carl Mummert Jul 19 '11 at 14:57
@Bill: I said "superfluous". I understand the intention. I disagree with the way all this is handled. –  t.b. Jul 19 '11 at 14:58
At Carl: Yes, I agree an unrestricted @name implementation would be best, but the SE folks have already argued against such on MSO. –  Bill Dubuque Jul 19 '11 at 15:01
Above should say "don't read the FAQ" not "fact". –  Bill Dubuque Jul 19 '11 at 15:05