Shortly (a day?) after a PE problem comes online, why not put some developer notes on the solutions forum.
The developer notes should be written by the main contributor to the PE problem with the aim of explaining the intended solution path as clearly as possible. Perhaps the developer notes should be constructed at the same time as the problem itself, and this should be a criterion for acceptance of a PE problem.
Pros:
Good solution documentation (as people have been asking for lately)
But there is not "the intended solution path", so it should be made clear that it's just one (hopefully good) of several methods. It should then, I think, be the first post in the forum, so everybody knows exactly where to find an elaborated explanation.
However, could that dissuade solvers from explaining their approaches?
Il faut respecter la montagne -- c'est pourquoi les gypaètes sont là.
That knowledge is coming anyway from the usual suspects! - but perhaps not always in a very understandable form. So I am hopeful that the promise of developer notes won't put anyone (else) off posting.
I think that developer notes would just neatly round off the problems. I guess as a dev team core member, you don't get this so much: but I often wonder, what did they have in mind when they set this one? I bet that others do, too.
ed_r wrote:I think that developer notes would just neatly round off the problems.
Indeed.
I often wonder, what did they have in mind when they set this one? I bet that others do, too.
Mostly "this might be a good one, let's see" and it either turns out so or not . And as you should know, a good problem not always turns out just like it was originally conceived. So the 'what did they have in mind' changes during development.
Il faut respecter la montagne -- c'est pourquoi les gypaètes sont là.
Additionally:
Quite a lot of problems have several solution methods that solve it within reasonable time. Some faster than others.
It's not uncommon that within the team more than one solution is found.
If we would go to the ultimate limit so that only one method would solve it within reasonable time the problem would get very one-dimensional. We have often tried to avoid that.
You can see that in the forum too: it happens that quite different methods are used.
So please go on and discuss your methods. We will drop in if we think the point is missed ultimately.
But with so many excellent solvers around that would not happen very often.
Let's not get too much classroom like where tests should be explained by the master that knows best.
War ruins the life and health of untold numbers of innocent children.
Awww, c'mon Daniel, you're just playing with words now. Take a look at the forum posts for #175. Lots of interesting stuff there, not all of it well explained, and no indication of how the problem developers (eventually) tackled the question. It would be great to have a commentary from the setters: not just how you solved the problem, but some history/motivation, comments on how hard the problem was expected to be ... just any sort of feedback.
Perhaps you're playing Devil's advocate since I notice you've not yet stated your position on this idea. Is it worth a go? Perhaps as an informal opt-in scheme for those problem creators that have the time/energy to contribute a little more? IMHO it should be forum etiquette to give feedback, but forget I said that and let me be pushy later on ...
hk wrote:Let's not get too much classroom like where tests should be explained by the master that knows best.
Far from it! This isn't a question of PE "marking" people's scripts; it's just one of openness and sharing. It would be unfair of me to characterise the current system as "them & us" (setters & solvers) but there is a tiny element of that that could be alleviated by a little more feedback. And, who knows, sometimes the teachers might admit to learning from their students
ed_r wrote:Awww, c'mon Daniel, you're just playing with words now. Take a look at the forum posts for #175. Lots of interesting stuff there, not all of it well explained, and no indication of how the problem developers (eventually) tackled the question. It would be great to have a commentary from the setters: not just how you solved the problem, but some history/motivation, comments on how hard the problem was expected to be ... just any sort of feedback.
Perhaps you're playing Devil's advocate since I notice you've not yet stated your position on this idea. Is it worth a go? Perhaps as an informal opt-in scheme for those problem creators that have the time/energy to contribute a little more? IMHO it should be forum etiquette to give feedback, but forget I said that and let me be pushy later on ...
I've not yet found my position on this idea. I see some pros and cons and am in the process of weighing them, aided by other people's comments on the matter (so it would be helpful if some others would chime in). One pro would be that our ideas might cause those who followed a different route to broaden their horizon and learn something new (please reciprocate then, we love to learn, too), a con would be that that leaves less room for related problems.
Il faut respecter la montagne -- c'est pourquoi les gypaètes sont là.
I am only a beginner here, so take this with a pinch of salt or two, but I think this sort of posting would have more negative effects than positive -- if it happened early in the thread. It's true that it might spur some competitive urges, but those are really already spurred; I think the larger effect would be to induce a 'why bother' lethargy.
But what if the developer's solution were posted a week later rather than immediately or a day after? Seven days is long enough that it would not adversely affect the first (or even second) rush of excitement, but it would still be an interesting read when it happened, and the archival benefits are substantial.
I agree that some feedback from the developers would be cool. I always liked to read euler's approach to a problem, but he stopped posting in the forums after a while
posting "The Developers Solution" right at the top of the forum is something i would like less, i think it spoils the search for the best/fastest solution a bit, and also there seems to be a match going on every week for being the first poster, which would be less fun if a developer always posts his solution first.
but id love to see a developer post some interesting stuff etc about a new problem, just participate in the problem's forum
edit: the post hk made in the problem 175 forum is more or less what i had in mind
edit2: it would be cool to document the best ideas about every problem. will take quite some time though, but there are very interesting things being posted in the forums o well
The more I think about it the more cons I see to the obligation to post developers notes.
- would'nt it be devastating if we said: we just wanted a very simple problem? Worse still than "Too easy" cries.
-would'nt it be very telltale if for some problem where a follow up is in the pipeline we refrained from posting a developers note?
-haven't we decided that the posts in the forum should be more clear?
So I would make it at this:
We follow the posts in the Forum. (We always did) If a nice extension is possible and is not already posted, we will do.
Nothing as rewarding to test drive your algo with an extension.
If some mathematical explanation is necessary and does not turn up: we will do.
If the wanted solution does not turn up. we will do.
But most of all: after publication the problem is common property of all. So please go on with trying to improve the solutions and elaborate on them. And don't wait on us to provide the final answer. That would be the end of an interesting forum.
War ruins the life and health of untold numbers of innocent children.
Reading this thread has been like following a great game of tennis. After the proposal I thought, "Fantastic idea!" Then as subsequent posts were made I found my own views oscillating from one side to the other. I think that Hans' final post has perhaps distilled the essential points and has drawn a very fair conclusion.
I think i will complicate this all again by suggesting that an idea i have seen elsewhere in a post. That clear and thorough explanations of the solutions are voted/selected for the top of the forum. Since there is usually more than one approach and these might not all have been thought of by the developer. For problem 175 i only understand the approach i used and i didnt explain that well enough.
Alternatively just to throw ideas about a wiki on each problem that can be edited by all. Add sections to the page for each approach and explain it.... yes i know its probably more hassle than the sun. It has advantages - anyone can remove the word easy or add/explain solutions.
Alternatively just to throw ideas about a wiki on each problem that can be edited by all. Add sections to the page for each approach and explain it
i really like that idea won't be easy to make i'm afraid, but it'd be really cool and help get all solutions for each problem explained clearly. the wiki for each problem should only be viewable for those who've solved the problem obviously
edit: also, you don't have to search through hundreds of posts to find some more info about why your approach works, or how you could've solved it in O(n^2) instead of O(n^3), etc
only works if people are willing to add things to the wiki though. but i think/hope that won't be much of a problem
Last edited by stijn263 on Sun Dec 30, 2007 4:03 pm, edited 2 times in total.
There are post rating mods for phpBB but last I looked they didn't do precisely what would be needed here. Writing the code yourselves elegantly is probably a couple of weeks of long evenings. Are there really a lot of problems where people come up with interesting solutions beyond the first 2-3 pages?
If the developers stick to hk's three "will do" rules then perhaps there won't be a need for notes, votes and wiki pages.
Nevertheless -- speaking only for myself now -- if I manage to co-develop a PE problem then I think I will post some developer notes just to "round it off". Besides trying to write clearly, I will endeavour to follow the good advice offered by people here:
Show restraint. Leave it a week or so before posting; don't jump in too soon and kill off an interesting discussion.
Be open-minded. The developer's preferred solution isn't the only one; and there's probably no such thing as "best".
Be sensitive. Probably safest not to call a problem "easy", "straightforward" or (a mathematical favourite) "trivial"!
Think ahead. Don't reveal so much about the development of the problem that the possibility for an interesting extension is killed off.
Euler hasn't banned developer notes; he's just agreed that they should not be obligatory.
or not 
. And as you should know, a good problem not always turns out just like it was originally conceived. So the 'what did they have in mind' changes during development.
