7

I am now involved in trying to introduce and popularise Mathematica at the university where I work. Among other things, I am coauthoring a general introductory article about Mathematica for a popular maths and physics journal. I want to mention both the MathGroup and SE and encourage the readers to participate in both, but I have been wondering what to say (if anything) about the differences between them.

I can see just one obvious difference: questions posted to SE appear immediately and so one is likely to get an answer faster, but what else? In particular, are there any questions that are more appropriate to the MathGroup rather than SE or vice versa? Should cross-posting be discouraged?

These seem to me a very natural question but I have not been able to find any discussion of them.

11

Here are some of the differences which come to mind and are in favor of SE:

  • Very fast turn-around (as you mentioned)
  • Community-moderated site
  • Rich answer format: code is specially formatted to stand out, images are allowed within the answer, answer can be structured using the markdown language.
  • Duplicate answers strongly discouraged both by the system and the community culture. Generally, people keep their aswers when they feel that they add something important, missed by other answerers.
  • Voting system, with all its shortcomings, generally works quite well to ensure that better aswers are at the top (but not always, particularly for answers added after a few days).
  • Incorrect answers don't last long, because people downvote them and / or use other means (comments, flagging etc) to make sure that they are removed or corrected.
  • The degree of collaboration / collective activity is much higher than on MathGroup, where each question receives pretty much a number of disconnected answers
  • Unanswered questions do get attention even after a while, which almost never happens on Mathgroup
  • The system tracks questions referred to (linked by) other questions, which is often useful.

But SE model is not perfect, and here are some of the shortcomings:

  • Those who answer first often get a rather unfair advantage just because they enter a "positive feedback" loop - their answers are seen first and upvoted, making sure they are still first (this does not happen always, but the effect exists)
  • Vote count for questions and answers does not have an "absolute" meaning: you can not meaningfully compare vote counts for different questions, or answers to different questions (because simple questions and often simple-to-understand answers tend to attract more votes than difficult questions / perhaps more thorough but more involved answers).
  • The way credit is given serves more to support competition among answerers than collaboration. Often, it is a disservice to everybody, since people are not encouraged to collaborate (no way to share credit, for example)
  • Rep does not mean much, really. It is mostly just a candy, and while people do want to get appreciated and get a feedback (so some form of rep is probably a good idea), the system currently is built in a way which does again encourage competition more than collaboration. Rep really shows the degree of participation on the site in the first place, and not much more (of course, there is some correlation between rep and the level of expertise, but it is less than often perceived: there are experts with very low rep since they just don't have much time to spend here / contribute).
  • Questions have a very short lifetime (arguably even shorter than on Mathgroup) . Questions more than a couple of days old (especially those which received an answer or several) are mostly "forgotten" by the majority of the audience, and the system does not encourage late answers. In particular, great answers coming later are usually unlikely to receive the votes they deserve, relative to the other answers already given (with some exceptions, usually related to things that impress people, such as fancy graphics, or something else that can be quickly appreciated without getting the full and possibly complex context).
  • Questions can not really evolve, like Mathgroup threads can, just because of the SE format (usually, a single answer per person, answers history is irrelevant - a kind of "stateless" model which does not keep any memory - except comments, which are limited). This may be an advantage, but often also is a huge disadvantage.
  • The system makes it hard for intermediate-level answerers or those who don't have lots of time to spend on the site to monitor new questions, to answer, because of the "fastest gun in the West" effect. This is different for Mathgroup.
  • The system makes it hard to share useful information, because one has to ask a question and then self-answer, which is a rather awkward way to do it. I am sure than much more would be shared had it offered a special format for that. There is no problem with this on Mathgroup.

All in all, SE model feels much like http requests: it is intentionally made stateless, prefers short answers, and actually also short and specific questions. While I feel that SE Mathematica is a step ahead w.r.t. Mathgroup in many ways, the system clearly has its shortcomings, and most of those were intentional design decisions and are unlikely to change any time soon. For beginner users, however, most of the limitations I mentioned won't probably matter much.

  • 2
    I think the main difference would be that it literally is not a discussion forum. This can't be emphasised enough. All other differences are secondary to this. – acl Mar 25 '12 at 16:40
  • A very useful answer. "Rep" seems to me rather like "teaching evaluations" - I am not overfond of either. The answer format is certainly much more impressive and convenient here than on MathGroup and I have always had the feeling that for beginners MathGroup could be quite intimidating. However, for deeper problems, requiring longer attention span it, is probably still the place to go. – Andrzej Kozlowski Mar 25 '12 at 16:46
  • @acl I don't disagree, but for me at least, the more detailed understanding of the mechanics of any two systems often helps to understand the main difference better, because the latter is a consequence of the former, and also because the statement "SE is not a discussion forum" is rather general and not very informative by itself: there may be lots of ways to make sites which are not discussion forums, and SE is just one of them (but the one important to us). – Leonid Shifrin Mar 25 '12 at 16:58
  • @Andrzej Thanks - agree with you on all points. – Leonid Shifrin Mar 25 '12 at 17:00
  • No, I do not think it's not informative, in that it expresses the main distinction between the mathgroup and here: one is a discussion forum while the other isn't, and explicitly so. This really is the fundamental difference between them (in my opinion of course). Should this place allow open-ended discussions for questions, I am quite certain that it'll end up like any other forum on the internet. – acl Mar 25 '12 at 17:04
  • @R.M. My point is not that there is no way to do this or that, but that the system does not provide an easy-to-use mechanisms for it, does not encourage this to a large extent /lower the barrier. There is no reason to take my answer as an extreme critique though - SE is a good next (after traditional forums) evolutionary step in systems of knowledge transfer, and also a business model. But it is certainly not the last step, and looking at it without rose-colored glasses (but giving it credit that it deserves) is a good step in understanding what would be needed for the next-generation system. – Leonid Shifrin Mar 25 '12 at 17:10
  • @acl We are always biased by the models we already know. I can imagine that some models which would allow open-ended discussions would balance that with some other mechanisms, which would make those different from the traditional forums of today. This is also a question of personal taste - I left Theoretical Physics already a while ago, and now am more an empirical person :) – Leonid Shifrin Mar 25 '12 at 17:15
  • Well, the "danger" of open ended discussion wasn't my point, but I nevertheless will say on it that I did not reach this conclusion by reasoning from abstract principles but by observing forums and online groups; so I think I'm being empirical enough, although I may be drawing incorrect conclusions. In any case, my point really was just that this is not a discussion forum. Discussing the pros and cons of what it is is just fine, of course. If I ask "what's the differences between a cat and a glass bead", the largest is that one is an animal and the other is not. It's also true that one – acl Mar 25 '12 at 17:33
  • is furry and the other smooth, one is hard and the other soft etc, but these are secondary (apologies for the analogy; it was the first thing that came to my head). – acl Mar 25 '12 at 17:34
  • @acl As I said from the start, I don't disagree with that. I just presented the way I view things (hoping it may be useful to some other folks), and since this is not a discussion forum (which is particularly exemplified by a long chain of comments under my answer), the usefulness of my answer is decided by the votes :) – Leonid Shifrin Mar 25 '12 at 19:00
  • The long chain of comments actually demonstrates my point: some jerk (myself in this case) will always go on and on (and on) about whatever he happens to think is important at that moment :) – acl Mar 25 '12 at 19:03
  • @acl lol :). It does. – Leonid Shifrin Mar 25 '12 at 19:08
  • 2
    I have been here a short time but feel I can already confirm everything Leonid wrote. One problem with the "Rep" system is that it encourages people to attempt to answer question even when they do not have sufficient knowledge or experience to do so, and they gain "Rep" for this. As this is not a "discussion forum" I will stop here, but I have found more wrong answers on this forum ("accepted" ones) than would be possible on MathGroup. – Andrzej Kozlowski Mar 26 '12 at 19:19
  • 2
    @Andrzej I am sorry to learn that your experiences here were not very satisfying so far. But, if the answers are outright wrong, you can leave a comment stating that, or flag it, or put your own answer with a rebuttal, or bring this to chat, so that the community knows and can do something about it. And I really hope that you will keep your presence here despite these annoyances, because it is the presence and contributions of experts like you which can help improve the quality of the answers and fight these negative effects. – Leonid Shifrin Mar 26 '12 at 19:41
  • A related discussion from the old days on SO: What's the difference between Usenet and Stack Overflow? (From an old man that only knows Usenet) – rm -rf Mar 31 '12 at 16:20
9

Perhaps the most important distinction between the two is that this site is not a discussion forum. From Szabolcs's email to the group announcing this site's launch:

This website is not a traditional discussion forum but a collectively edited questions and answers site: ideally each question thread will eventually become a little piece of knowledge that is of use not only to the original asker but also to everyone who finds it using a search engine.

So questions that are mostly rants (e.g. https://mathematica.stackexchange.com/q/1917/5, which already has one vote to delete) or soliciting strong opinions/bullet-list answers (e.g., What is the strength of Mathematica compared to MATLAB, which is now deleted so you need 2k+ in beta to view), which would've gotten an "Amen, bother!" response or led to opinionated arguments on MathGroup are explicitly off-topic on this site.

That is not to say that the same question cannot be asked constructively (e.g.: Implementing efficient multiple undo) and we don't salt the earth if you use the word MATLAB or its concepts either (e.g. How can I improve the speed of eigenvalue decompositions for large matrices?, which ultimately turned out to be an issue with Timing or Why don't * and ^ work as I expected on matrices?). We have also been a little relaxed in allowing some level of discourse rather than an extremely strict Q&A with one definite answer, but these questions are exceedingly rare (e.g. Placement of Condition /; expressions).

In addition, there's always the option of using the Mathematica chatroom if you want to engage in discussions (pretty much about anything, not necessarily Mathematica), ask/talk about stuff that is off-topic for the main site etc. People have also known to help others with small issues without having to ask a question formally on the main site. Your students can also pop in to ask if certain kinds of questions would fly on the main site. Sometimes it might be a dupe and regular users might fish it out for them. The rules in the chatroom are very relaxed and the place is a lot more informal, but civility is certainly a must :)

Cross-posting between MathGroup and this site is fine. There is a higher likelihood of finding niche domain experts (not necessarily Mathematica experts) on MathGroup because of its age, but it would be courteous to users of both if you explicitly mentioned that this has been cross-posted, with links to each, so that people can check to see if an answer has been posted before writing one of their own. In the same spirit, it would be courteous if you let people know if you have gotten a satisfactory solution on the other group. If the solution was on MathGroup, then I'd suggest recommending to the answerer to also post it here. Barring that, the OP can answer it themselves and attribute it to them and link to that thread.

  • 1
    Actually, the MathGroup (which is pre-moderated) does not permit explicit mention of competing products (like Maple or MATLAB) and heated discussions concerning such matters (without mentioning names) tend to be cut short by the moderator. – Andrzej Kozlowski Mar 25 '12 at 15:52
  • 2
    @AndrzejKozlowski I agree, hence the "... we don't salt the earth if you use the word MATLAB or its concepts either" :) – rm -rf Mar 25 '12 at 15:54

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .