I taught a junior level CS course in the past, and used Alex Aiken's MOSS (developed in the mid 90s - this is a new version of an old story) to detect cheaters. Found quite a few. Every accusation was based on my inspection of the similarities; I treated MOSS as an aid in helping me spot those similarities but not as proof that the students were cheating.
Tools like MOSS will help you find a certain type of cheater (those who neither grasp the problem nor have the ability to code the solution) but not others (those who can code, but do not understand the problem to be solved). The latter group can look at working code (eureka!) and then code their own solution. The amount of work required to take existing code and make it appear different to MOSS was non-trivial; renaming variables, reordering code, moving code from one function to another - these were all things that were caught. My belief is that for someone to fool MOSS, he would have to be well versed in the language, and would have been able to finish the assignment without resorting to cheating.
As for making the accusations - I consulted with other instructors and faculty and found varying opinions. Some felt it was their duty to report all cheaters; others felt it was too much work and that nothing would happen. There was some consensus, though. I had to be absolutely certain that the students cheated; if I had any doubt I should not make an accusation. Furthermore, I had to be able to prove to others that the student cheated. "Others" refers to a college or university level committee that handles appeals; this group would have members who would not have CS backgrounds. Since my only proof consisted of the similarities in the students' programs, it had to be good enough to convince such a committee.
Even so, I heard of appeals being successful. A student had clearly cheated, and was given an F for that course. This would have dropped the student's GPA below a certain average, which would have meant he lost his scholarship, which in turn would mean he'd have to drop out of school. The committee weighed this heavily and overturned the instructor's ruling.
If you ever wondered why those guys who cheated got a better grade than you did - well, maybe there were other things going on that you were not aware of.
Slashdot Mirror
Bidirectional debugging allows you to step forward and backward one line of code at a time.
A paper published in PLDI 2000 describing this is available here. IIRC at some point they had an implementation of this built on top of gdb.
I taught a junior level CS course in the past, and used Alex Aiken's MOSS (developed in the mid 90s - this is a new version of an old story) to detect cheaters. Found quite a few. Every accusation was based on my inspection of the similarities; I treated MOSS as an aid in helping me spot those similarities but not as proof that the students were cheating.
Tools like MOSS will help you find a certain type of cheater (those who neither grasp the problem nor have the ability to code the solution) but not others (those who can code, but do not understand the problem to be solved). The latter group can look at working code (eureka!) and then code their own solution. The amount of work required to take existing code and make it appear different to MOSS was non-trivial; renaming variables, reordering code, moving code from one function to another - these were all things that were caught. My belief is that for someone to fool MOSS, he would have to be well versed in the language, and would have been able to finish the assignment without resorting to cheating.
As for making the accusations - I consulted with other instructors and faculty and found varying opinions. Some felt it was their duty to report all cheaters; others felt it was too much work and that nothing would happen. There was some consensus, though. I had to be absolutely certain that the students cheated; if I had any doubt I should not make an accusation. Furthermore, I had to be able to prove to others that the student cheated. "Others" refers to a college or university level committee that handles appeals; this group would have members who would not have CS backgrounds. Since my only proof consisted of the similarities in the students' programs, it had to be good enough to convince such a committee.
Even so, I heard of appeals being successful. A student had clearly cheated, and was given an F for that course. This would have dropped the student's GPA below a certain average, which would have meant he lost his scholarship, which in turn would mean he'd have to drop out of school. The committee weighed this heavily and overturned the instructor's ruling.
If you ever wondered why those guys who cheated got a better grade than you did - well, maybe there were other things going on that you were not aware of.