Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorKevin Lin
    • CommentTimeDec 17th 2009 edited

    I find it kind of weird how people here seem to be so uniformly in agreement with the mantra "the SO/MO format is not good for discussions". On the other hand, I'm ok with "discussion is inappropriate on MO"; certainly the stated goals of the website do not include being a forum for discussion. And I understand that there are many very good reasons to disallow discussion, e.g., so as to not obstruct the actual math from going on, or to not have to deal with inevitable flame wars.

    But I just want to say that I don't think it's necessarily true that the SO/MO format is not good for discussions. If this is not a good format for discussions, then what is a good format for discussions? To have a good system for discussions, what more do you need besides a box in which to type comments and another box from which to read previous comments? It's not really that complicated. What makes message boards and blogs better for discussions? Would you say that the commenting system on, e.g., slashdot is a good format for discussions? It seems that the SO/MO system could easily be adapted to essentially mimic the slashdot system. For example, if you ordered posts chronologically rather than according to +/- points, it would essentially be the same as the system on slashdot, the main difference that I can think of would just be the depth 1 threaded commenting rather than the arbitrary depth threaded commenting. If this small change were implemented, then I would think that the SO/MO format would be even better than blogs and message boards for discussions --- definitely better than, e.g., the as-basic-as-possible commenting system on Wordpress blogs.

    Again, I don't have any problem with banning discussions on MO; I just want to pedantically point out that (with some very small changes) it's not necessarily a bad system for discussions, contrary to what many people seem to think.

    I don't know if this is the reason other people say that, but one reason MO is not conducive to discussions is that, in my opinion, the voting system doesn't accurately reflect the relative quality of different contributions in a long discussion. The problem is that answers posted earlier may receive more votes simply by virtue of having been viewed by more people and that, as time goes on, the number of people viewing a question per unit time decays. Votes therefore don't accurately compare very recent and very old answers.

    More generally it seems to me that discussions and the voting system are somewhat orthogonal. The voting system rewards individual contributions - in the ideal case these are concise and precise answers to the OP's concise and precise question. But the nature of a discussion is to synthesize multiple contributions. For example, if I write an answer that synthesizes the current state of knowledge of the discussion, should I gain reputation for it even if I didn't myself contribute any ideas to the discussion?

    Why take our word if you can get information directly from the designers of the SO/SE system (Joel on Software blog):

    or perhaps these answers from meta:

    In short, the SO/EFU design is tailored to solve a particular problem — getting answers to specific questions and building a community — well. Therefore, it solves some other problems less effectively.

    A discussion site is by definition the one where posts don't result in an answer. While anyone is free to try making such a site for programmers/math guys, it's much more likely to become a decent place to talk about weather, jobs, conferences, girl/boyfriends, and anything else related to math rather then math itself. There's nothing wrong with the existence of such a place, but it's not like something I would get excited about.

    Since it's not likely that a large number of highly professional people will be willing to spend a lot of time every day discussing the topics around mathematics, you'll simply have the average quality of a post go down. On the other hand, building a good site and, more importantly, the community around it, requires lots of work. So this will be lots of effort for less gain.

    Actually, take a look at the Joel on Software discussion forum. This is in fact the first time I ever looked there. It's also the last time. Try, for example, the recent question there called How to present grad school work?. Maybe there is something interesting in that thread, but here's the problem: I absolutely have no time to read a discussion, only an answer.


    As stated above, SO/MO weren't designed as discussion fora, and hence they would suck at it. In fact, as far as my experience goes, every piece of software ever written for this purpose sucks. I don't think it's because their authors were incompetent; I think it is because it is a very difficult problem to solve. It may even be impossible to solve by the means employed today, because users have such different tastes, habits, and requirements. I have my own ideas as to how things could be done better, but then other people have ideas for how to achieve world peace, and they aren't having any success either. I won't bore y'all with my thoughts here.

      CommentAuthorJon Awbrey
    • CommentTimeDec 18th 2009 edited

    All I know is when thinking people hand over power to an unthinking process the results are usually not good.

    • CommentAuthorKevin Lin
    • CommentTimeDec 22nd 2009 edited

    Actually, I just noticed that you can actually already view answers to questions in chronological or reverse-chronological order on this site.

    Qiaochu: Your comments about votes not accurately reflecting the quality of different contributions in a long discussion could also be applied to some non-discussion situations, could it not? At least for "let's make a big list" type questions or "your favorite math joke" type questions, but also for more serious questions, e.g. this one where I feel that Gretar's and Shizhou's answers are not getting as many votes as they should be, probably due to having been posted a bit later than the other answers. I don't really care for points or votes, but anyway I definitely wouldn't ask for discussion-y posts, if they are ever allowed at all, to be anything but community wiki.

    Ilya and hanche: There are certainly lots of inventions that were not designed for doing X, yet are still perfectly reasonable, or even good, for doing X. And ok, you think that discussion sites suck, and you wouldn't use them yourselves. But I have a different view, and evidently lots of others do as well (judging by the large number of people who use discussion forums and who have discussions in blog comments). I don't claim that discussion forums are perfect, but I think they can still be very useful.

    Again, I don't have any problem with disallowing discussion here. But I'm just still not convinced at all that discussions don't work on the SE system. I think they'd work here just as well as anywhere else, and it wouldn't suck any more or any less. If you think that all discussion software currently sucks, that's fine. If you want to disallow discussion here, that's fine. But I'd like to hear why SE is strictly worse for discussion than any other discussion software.

    • CommentAuthorHarry Gindi
    • CommentTimeDec 22nd 2009 edited
    The reason why discussions on SE aren't optimal is because new answers bump a question up, while comments are limited to 500 characters. If there were a tag [discussion] that didn't bump to the top of the question list, or if it were separated from the rest of the site, then that would work fine.

    Meanwhile, Kevin Lin, have you considered posting something about that proposed EGA translation project on meta or somewhere else? It sounds interesting, and I'm sure you could find a lot of people who'd want to help.

    Perhaps we should just start another forum ( running the same software as here, which is explicitly for discussing math (and not discussing mathoverflow; that stays here).

    I'm extremely hesitant to try this. I like mathoverflow just fine now, and don't want to tinker with the formula too dramatically. I also don't want any extra moderation/maintenance duties! :-) But I'd be interesting to hear people's opinions.

    • CommentAuthorKevin Lin
    • CommentTimeDec 23rd 2009

    Harry: I emailed you regarding the EGA thing.

    Scott: Yes, I understand your hesitance to do this. I also don't think it's necessarily a good idea yet. I just think that it's a possibility that could perhaps be explored later and that we might keep our minds open to.


    Kevin: (I'll be brief since this is somewhat off topic.) That many people use X is not evidence that they don't think X sucks. It is merely evidence that they think the value they get from using X outweighs the pain inflicted through its suckage. Heck, I use discussion forums (fora?) myself, but only ones I find intensely interesting, or ones I need to use to get answers to my questions. (Maybe this is a good thing. It stops some time wasting behaviour.)

    I think one reason the format is not good for discussions is the draconian character length limit on comments. This is of course arbitrarily chosen, probably specifically to discourage discussion.

    I can sort of see the argument to "use each tool for one thing." But I don't think there necessarily needs to be a new forum attached to MO to move discussions to. Lots of people already have blogs, whose comments systems are more intended for discussions. Some good discussions also take place in query-boxes at the nLab.

    @Mike, indeed the designers chose to create a clear dichotomy between answers and comments: the first type bumps a question to the top, the second not. Therefore, the reputation threshold for posting answers and the process of posting answers were made deliberately easier: people will notice a bad answer, while a bad comment may stand uncorrected and unseen forever.


    Here's why I think MO (and other SOs) are not appropriate for discussions:

    1. You can't follow a discussion. There's no easy way (that I can see) to say "tell me when someone replies".
    2. The default view messes up the order of a discussion. If in answers, they get ordered by vote count; if in comments, one a small number get displayed and that's chosen by votes as well. So you have to know that a discussion is taking place to know that you should reorder the comments/answers by date.

    These two kill the idea of discussions on MO for me. Of course, if everyone interested knows that a discussion is taking place then they know to keep checking it and know to reorder it. But that destroys the purpose of having the discussion somewhere public: so that more people can join in and contribute.

    Compare this with discussions here or at the n-Forum. When I log in, I get presented with a list of discussions with new replies. I can easily click through and see what those replies were, in the order that they were posted. I can even see them in my RSS feed reader to see if they're worth visiting the site to read!

    Nice n-amedrop! =)
    • CommentAuthorCSiegel
    • CommentTimeJan 15th 2010
    I know off topic for this thread, but Kevin, if you're organizing (or know someone who is) an EGA translation project, I'd be interested.

    Now, back on topic: I agree, that the arbitrary character limit on comments is a big impediment to discussions, but the biggest one I see is, as Andrew Stacey pointed out, the fact that the default view is by votes, plus the fact that there's no internal threading (which can be hacked with setting everything to community wiki if everyone agrees on a format, but still, not so great).

    That said, if I see a discussion question that I find interesting, I would attempt to contribute.