Bill Dubuque comments on "Related questions" (7529)

Bill Dubuque — Sat, 24 Jul 2010 12:20:06 -0700

I agree that it would be very useful to explicitly support links (as opposed to ad-hoc links interspersed in answers/comments etc). It would also be helpful to allow the links to have metadata to indicate the type of link, such as generalization, specialization, MSC/subject classification ... and other such remarks that convey why the author thinks the link is pertinent. Not only does that allow the reader to decide whether it is worth their time to follow the link, but it also conveys to the reader some idea of the authors intent in supplying the link.

Strangely enough, as one of the heaviest linkers in some other forums (sci.math, ask an algebraist, etc), I've discovered that not too infrequently readers may misinterpret the intent of the author when supplying many links. For example, I happen to think discussion of generalizations, specializations, etc is highly valuable when teaching algebra, so that students can get a really good handle on how mathematical abstraction works. So I often liberally pepper my pedagogical posts with links to my prior posts discussing such (I have far too many posts to recall after a couple decades online). I have Emacs keyboard macros that search for keywords in my database of thousands of posts and automatically inserts links. When pressed for time I'd sometimes insert such links without too much comment. I noticed that such little-annotated links were received far less favorably than those that were more heavily annotated. E.g. occasionally newcomers would misinterpret such links as "showing off". But most longtime readers know that - at heart - I'm a generalist and I enjoy sharing my experiences with the many tantalizing ways that mathematics is interconnected. In any case my point is simply that the richer the link annotation capability, the more likely the link will prove useful. Imho such rich linking capabilities should be a requirement of any software for mathematical discussion.

Another point worth emphasizing is that structure imposed by such links helps serve to overcome the linear, non-threaded nature of the Q&A format. For example, a number of Q&A's linked together with remarks and appropriate metadata could comprise an interesting survey of open problems, or, perhaps, a supplementary reading topic for a course, etc. The possibilities are endless. Thus, even though the current foundation of MO is not designed for such discussions etc. that doesn't mean that we cannot impose such higher-level structure *now*. The sooner the better since it's much easier to annotate posts when they're first processed than to have to revisit all those posts later when MO is ported to a richer platform.

Mariano comments on "Related questions" (7507)

Mariano — Fri, 23 Jul 2010 14:25:20 -0700

David, well, you could have given the exact same argument to Tim Berners Lee a few years ago!

WillieWong comments on "Related questions" (7506)

WillieWong — Fri, 23 Jul 2010 10:37:02 -0700

And presumeably there can be a segmenting off in the display: so the "related to" fields will not be buried among the comments. I can see why it may be useful, but it may end up being another one of those new features we won't get from SE.

Qiaochu Yuan comments on "Related questions" (7505)

Qiaochu Yuan — Fri, 23 Jul 2010 09:34:51 -0700

I think Mariano means if question X links to question Y, there should automatically be a link on question Y to question X, so a large web of questions is easily navigable.

David Speyer comments on "Related questions" (7504)

David Speyer — Fri, 23 Jul 2010 09:17:34 -0700

It does. Just leave a comment on the question saying "A similar topic is discussed at ...." Why would adding specific software support for this help?

Mariano comments on "Related questions" (7502)

Mariano — Fri, 23 Jul 2010 08:09:29 -0700