Slashdot Mirror
Creating a High School Programming Competition?
goatmon asks: "A few high school computer teachers across Tennessee have organized an annual "Geek Games" competition. We started last year with a great competition in PC repair, patterned after the VICA competition. This year we are trying to expand to add a competition in computer programming. The question is, how do you create a language-independent computer programming competition? Is it fair to have a timed test if one group of kids is hacking in Python and the other in C++? And who wins, the one with the shortest code, the fastest programmer, the one with code that works? Can anyone offer insight from experience or a pattern that we might be able to follow?"
See the International Olympiad in Informatics model...
Its at http://olympiads.win.tue.nl/ioi
My suggestion to you would be to set a goal that is generic. Decide on a system that whould be built. This can be something graphics based (e.g. a simple calculator that does addition, subtraction, division and multiplication), information based (e.g. building a time tabling system), networking (e.g. a system where input occurs on a client, and processing on a server).
However, the trick is not to specify technology. In fact, you could make it quite clear that it doesn't matter what development paradigms that are used. What is important, though, is that the goal be described as something that is useful. (I remember being brilliant with my repeating print out on my ZX81, but realising that no one wants to see "Eibhear was 'ere" 10, 100, 1000 or any amount of times.)
In short, decide what you would like to be produced by the entrants, and let them decide how they do it.
With respect to judging it, you could devise a set of qualitative criteria that can easily be assessed by judges with a modicum of technical experience. Criteria could include:
These criteria are somewhat subjective, and to ensure fairness for all entrants, there would have to be a panel of judges, like in the figure skating competitions.
Then there is a set of quantitative criteria, which would enclude:
The point is to make the competition such that one geek with a linux and a C/C++ compiler and years of programming experience will not have an advantage over a kid with access to and a preference for HTML/javascript/VB/perl/etc. By making the requirements of the competition non-technical, the entrant will focus on the means best used by herself, and judgement will not be prejudicial.
Hope this helps.
Éibhear.
If I'd done it, I would have allowed them to use any language/environment. I think the ones using Perl would have had an advantage on certain problems, but that's just because we gave them many limited scope problems. If you gave them a single large problem, the ones using a stricter language like Java or Ada would probably have the advantage.
Anyway, I think you need to avoid subjective judgments like elegance and concentrate on measuring things like correctness and speed of completion.
Check out my college's programming contest website for some ideas.
Good luck
The ACM (Association for Computing Machinery) has held college-level programming competitions for many years. Though the actual problems may be at a level beyond what you are looking for, they have long ago solved the issues of how to hold such a contest.
Here is the 2001-2002 ACM International Collegiate Programming Contest page which lists, by region: a report, the standings, and the programming problems.
The basic format is to have a number of problems for the contestant teams to try and solve in a fixed amount of time. (The first one I checked out listed 8 problems for a 5-hour contest.) It included explicit problem specifications, sample input, and sample output for each problem.
I'm sure that by looking down the list, you'll find a baseline of informationt that will go well towards helping you to design your own competition. Good Luck!
There is already something similar to what you speak of. Its called the American Computer Science League or ACSL. I don't participate in it but i think my school got like 3rd or 4th nationaly last year. It works by sending in the answers to porblems for the 'regionals'. and then the highest scoring teams get invited to the national competition.
You should leave the language choice up to the students. Someone who picks C to do pattern matching and string operations instead of Perl deserves the buggy code they'll get (and the time they'll waste). Actually, this will end up enriching the contest, so far as students who learned more than one language will be able to pick the one best suited to the task at hand, a skill which can make or break a good design.
Join the NFSNET. Our prime goal is making little numbers out of big ones. http://www.nfsnet.org/
This isn't how to set up a computer copetition. It's a true story from ~6 years ago when I was at one in high school.
At the local commmunity college they had a science and math day every year and there was a computer programming category. After the opening ceremony the students were being led to the different types of labs where they could compete.
Let me first say that at the time Windows 3.11 was state of the art (Ok, Mac's were but we weren't using them) and the AP computer classes used Pascal as the language of choice. It changed the next year or so to C or C++.
Back to the story, as we were walking most people were talking. For no reason everyone when quite except fo this one guy who said "God programs in QBasic."
That's it. Kinda weak when I re-read it but it was funny as hell at the time. Not that we were experts. We knew that we didn't know much relativitly speaking but we knew God didn't use QBasic.
You might try the American Computer Science League. Apparently, they specialize in High School level competition.
My high school participates in and loses at a smaller, local competition every year. As far as I know, students can work in C++, Pascal, or Basic. They're given a set of problems to play with for several hours. The team with the most correct implementations in the shortest amount of time wins.
My information may be shaky. I have no first hand experience just yet.
We had 3 hours to complete 4 problems. You work in teams of (at most) 4 people, and I think you can use any language you want (you could use a different language for each problem if you wanted). But you could only libraries that were included with the language (no add-on packages). We used QBasic (the only language all 4 of us knew), some people used Pascal or Visual Basic, and I think there were people using C and Turing. The problems were best suited to text-mode programming, so VB just made it harder (people had to create a form with a start button, and some labels for output). One team used a computer running GNOME on Linux (now that I think about it, that was when I first heard about Linux). I'm not sure what language they were using.
The programs usually used input from a file on disk, with given names (i.e. "A:\DATA31"), and output to the screen. We usually had sample input. When we finished any problem, the judges would come to our computer and give us a disk with new input. We'd run the program, they'd check if the output was correct, and give a mark out of 100.
Time bonuses were also assigned. The time bonus started at a given number, and decreased linearly to 0 by the end of the contest. When you submitted a problem, you'd get a time bonus. You could resubmit once, but you'd obviously get less of a time bonus, and if you got a lower score, you were stuck with it (only 2 tries were allowed). Of course, there was different input data for the second attempt. With 4 questions marked out of 100, some teams managed to get scores around 490 because of time bonuses.
Code quality was not a factor in judging. It's too subjective, the judges didn't have time to read code, and the judges can't be expected to understand every language students use. Efficiency was usually not a factor, but some languages (QBasic) did put limits on memory usage. There was also a limit on how long a program could run for, to prevent people from using purely brute-force solutions that take an hour to run. IIRC, the limit was around 1 minute (if it took longer, you got 0), but wasn't timed with a stopwatch or anything - I don't think anyone ever ran into that limit anyway.
I don't think it was unfair that students got to use different programming languages. The alternative is to force everyone to use one language, which isn't really fair either. However, there was a small problem one year. One of the questions was based on factoring numbers (or something like that), and one team had a library that could do this for them. The library was included with the language, so the judges ruled that this was legal. The other teams felt this gave them an unfair advantage, so you should probably say something like "you must write the factoring code yourself, don't use a library function" in the question. OTOH, you can't just ban libraries, because some languages (like C) are nearly useless without their libraries.
The International Conference on Functional Programming sponsors a contest every year. The prizes typically include a free trip to the conference if the winners are university students.
I am always impressed by the contest and the winning entries.
The fourth contest (2001): http://cristal.inria.fr/ICFP2001/prog-contest/
The third contest (2000): http://www.cs.cornell.edu/icfp/
The second contest (1999): http://www.cs.virginia.edu/~jks6b/icfp/
The first contest (1998): http://www.ai.mit.edu/extra/icfp-contest/
Since you're in the USA, you might as well check out the USA Computing Olympiad, the competition which leads to the previously mentioned IOI. It's language-specific (C, C++, or Pascal), but that doesn't matter - the problems are hard enough that you should be spending significant amounts of time thinking and working stuff out so that you can hopefully code in 20-30 minutes (per program) regardless of what language you choose. The language won't have a big effect on programming ease or time if the problems are well written.
I participated in the ACSL contests mentioned above when I was in high school. We also have a contest run annually by the University of Waterloo called Canadian Computing Competition
The way they work is this - you get a certain amount of time (forget how long - maybe 2 hours?) to write 5 programs to their specs. You have sample input and sample output, and at the end you get scored based on how correct your programs execute on different input files. So I guess they value correctness of output over cleanliness of code or efficiency - the ends justifies the means in this case. This round is done in the high school and usually administered by your computer teacher.
Because it's a timed competition, the design of the code is usually pretty ugly as submitted by the programmers. But it's a great exercise - most programmers are able to finish 3 or 4 of the programs (sometimes correctly). Only the best ones are able to finish all 5 (though there's always something that breaks when it comes to getting the new inputs).
They pick 20 or so of the top programmers by hand (now it's score + code style I think) and move them all to a single location to go head to head in a national competition to come out with the winners list.