There is a large portion of math in the PM forums that could be turned into PM articles. As PM gets more users and the forums start supporting latex this will become even more relevant. It would seem that we need some kind of procedure after a question is answered.
Suppose we have some discussion in the Problem forum. For example
http://planetmath.org/?op=getmsg&id=8122
I think the best way would be to create a request and somehow attach the discussion to that request. This would require some modification to the PM software.
Another way (with no changes to PM software) would be just to create an request with a link to the discussion. A problem, however, is that further discussions can take place at two places: 1) on the request page and 2) on the original problem forum.
In any case, it should be easy to see which threads (in the Problem forums) that are on their way into PM entries. License-wise there is probably not any issues if we add some attribution line "This entry is based on the discussion …". I think it would also be polite to offer the PM entry to the person who originally answered the question on the forum.
More generally, is there a need to rethink the request page? I think the idea of a "seed entry" would be great. There was also an idea (I think a good one) that a new entry is for some time world editable. — matte 07.19.2005
I wonder if just making messages searchable and having "good" threads get weighted highly would have the desired effect. Combined with a reputation system, the reputations of the participants could be used for weighting. Of course we could do weighting right now based on number of replies. --akrowne Tue Jul 19 11:59:00 UTC 2005
An "Advanced search" would be great. However, this only helps if the user knows what he is searching for. I think it would be best to have all the mathematical content in the encyclopaedia. Or, I do not think we should have mathematical content spread out over the whole of PM. (Part in the forums, part in the encyclopaedia.) After all, part of the fun thing with PM is its browsability.
If a discussion merits an entry, why not have a method (or culture, protocol) for turning it into a PM entry? This will make it easy to use in other entries. By entering it into the encycloaedia, the new entry will also be classified, attached to an entry (if suitable), and interlinked with the rest of the entries. Turning a thread into a good entry probably also involves some copyediting, restructuring, etc.
I could imagine that entries based on forum discussions could be quite useful since some of these come from "real life" problems. – matte Jul 19 2005
One cool trick would be to do Noosphere auto-linking in the messages. Not quite what Matte was asking for, but it would be a step, since it would get messages "in" the encyclopedia automatically as long as they match some existing item. People might find answers automatically this way :). Also, in cases where the question needed a new answer, the author could put the body of the response in the encyclopedia and just a pointer to it in the follow up message. --jcorneli PS. I forgot to mention that if articles had backlinking (perhaps with several different "types"), then we could see which forum postings linked to the article. This is a very "scholiumific" approach (as Ray and I agree, via private correspondence!) and best of all (IMO) it should be simple to code up. :)
By auto-linking a thread, I assume you mean that the thread is attached to an existing entry in the encyclopaedia. If we employ this solution, then after a real entry is written from the thread, the same data is in the encylopaedia twice.
I think this approach can be developed further. Here is what I would propose:
Let's add more structure to the Problem forums. Instead of just posting a message, one would "Open a question". Questions would have the same kind of meta data as PM-entries such as related entries, related questions, and classification. However, these can be filled in during the discussion. After the question is answered and thoroughly discussed, it is "closed" (although users can still post to the thread). By dividing questions into Open and Closed, it is easier to present and sort the question forums. Also, since problems would have a "Related entries" field, it would link the problems to the encylopaedia more closely (with bidirectional linking). This would also have the advantage that the Problem forums would be easier to use and browse due to classification.
If the topic of a closed question is of general mathematical interest, one could flag the thread as "Entry material" (or something alike). These threads would automatically appear in the "Request list". After an entry is created, it would have a link back to the original thread, and vice-versa, a link from the thread to the entry. This is a good way to credit all contributers. Also, if there is any problems interpreting the entry, the original thread would be useful.
This is maybe what you meant? – matte June 20.2005
NO! (at least, not quite :)). What I meant is "autolinking in the already existing Noosphere sense of the word", wherein key phrases (like "automorphism group" or "holonomic" or "Oswald's lemma") become hyperlinks, automatically. So if someone says "Why is xyz the case?" an author could reply "See Oswald's lemma in the encyclopedia" and not bother with copying and pasting the link. YES, what's true, is that if an author responds in full to the question in the forum, then we'd tend to see duplication of answers. But tricky & smart authors would be expected to respond to the question in the encyclopedia and then only include a reference to their solution in a followup message (i.e. assuming the answer wasn't entirely trivial). Does this make more sense? You wouldn't see duplication of answers in the encyclopedia and in the forums as long as authors followed the rule I'm proposing here. Furthermore, some people asking questions might find that the autolinking gave them answers that they were too lazy to look up (or couldn't think of the right search terms to find). Such linking wouldn't always take place, of course, because sometimes the question-asker would choose the "wrong" words, but then the answerer can use the right ones. --jcorneli
OK I should have been able to quess that one :-) Linking from the forums into the encyclopaedia could be good. For example with \PMlinkname commands. www-links are already automatically made clickable. However, I am not so sure about auto-linking the whole text. That will generate a lot of bad links. In particular since messages are not classified nor editable. Latex support would be great though.
I'm not here discussing a "quick fix", but more idelogically thinking what would be the optimal way to make full use of the mathematical disucssions on the problem forums. I think the solution I offered in the previous post is a quite good one. For example, if someone has outlined a rough sketch of a proof in the forum, that is a good start when someone – at some point N years later – is writing an entry on the topic. The solution is also completely automatic. Once a problem is posed, it is a start of a chain leading to a PM entry on the topic. Another good thing is that it is distributed. The tasks of 1) posing the problem, 2) solving the problem, 3) setting meta data (classification, related entries) 4) copywriting/editing can all be done by different people. That is a good thing in an environment like PM. matte
Yes, this is a good idea. (I had to explain my idea before I could really get into yours, because I think mine is a good idea too and I wanted to make sure it was clearly laid out :).) In the approach you're proposing, questions are more like "bug reports" from the software development world. This makes sense (given what PM is), and the design of the system could follow the design of standard bug reporting software.
On the topic of "bad links" you brought up above - I don't think this is a problem with my suggestion, but with (a) the autolinker and (b) the fact that, as you mentioned, forum postings aren't editable ;). In fact, the more bad links we see, the more we should be able to fix. I'll stand by my suggestion, but note that it may require more of a redesign of the forum system than I had initially thought.
--jcorneli
Maybe not a bug report. More like a feature request :)
I do not think our proposals are mutually exlusive. I, however, would see links from the encylopaedia into Problems as temporary. Over time, someone should transform these into entries (and the PM software would/should encourage this). Would you agree with this? Some threads can be pretty messy to read.
Here is how I would envision the link between Problems and the encyclopaedia:
Each Problem (Open or Closed) can be classified and it has a 'Related entries' and 'Related problems' lists. After this, the Problem would be linked (bidirectionally) from all the entries/problems in its Related lists. When the Problem is made into an PM-entry, the Problem's Related-lists would just be transformed into the entry's Related-lists.
In any case, (as you said) the forum system first needs to be rethought in order to implement any of these approaches. But, when doing that, one should keep these issues in mind. — matte
I don't think there's any reason to make any of the links temporary, though the way they are handled might change over time (e.g. perhaps they would no longer show on the article's main page, but on a sub-page "Problems (or questions) that link to this article"). The fact that the questions are now encyclopedia entries doesn't change the fact that they are still, in their basic nature, questions. So, someone might want to use this list of questions as an exercise set in order to learn more about the theorem. Anyway, various details like this can be worked out later, no problem. The main thing is to get some development effort going on the Noosphere code! --jcorneli