Slashdot Mirror
As Computer Coding Classes Swell, So Does Cheating (nytimes.com)
An anonymous reader shares a report: College students have flooded into computer science courses across the country, recognizing them as an entree to coveted jobs at companies like Facebook and Google, not to mention the big prize: a start-up worth millions. The exploding interest in these courses, though, has coincided with an undesirable side effect: a spate of high-tech collegiate plagiarism. Students have been caught borrowing computer code from their friends or cribbing it from the internet. "There's a lot of discussion about it, both inside a department as well as across the field," said Randy H. Katz, a professor in the electrical engineering and computer science department at the University of California, Berkeley, who discovered in one year that about 100 of his roughly 700 students in one class had violated the course policy on collaborating or copying code. Computer science professors are now delivering stern warnings at the start of each course, and, like colleagues in other subjects, deploy software to flag plagiarism. They have unearthed numerous examples of suspected cheating.
Often it is extremely easy to determine, namely from patterns of mistakes. I once, as a TA, had a case where I refused points for some exercises to several students. They complained, and then I showed them how a comma had become a dash and then had become a quote over several generations of copying (with the dash and the quote making absolutely no sense in the given context). That shut them up pretty fast.
Of course, for simple and correct code fragments you cannot actually detect cheating reliably. The easy solution is to make the exercises more complex. Students are coddled far too often in coding courses anyways.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
There may be a limited number of ways to solve the problem, but there are an unlimited number of variable names that can be leveraged to solve a problem.
No two should be identical. Though all will be similar.
Sums it up. Or they get a friend to write it for them. I remember in college I was the only one in my computer science class to get a week long project right and have unique code. Everyone else either got it wrong, or all had the same code.
But I agree, kids are coddled in class. This is a very tough mental field. The campaigns to teach everyone to code make it seem like it's house work. Anyone can do it. I think they are finally realizing it's not so.
The key is that there's nearly unlimited ways to solve a problem incorrectly, but when the exact same incorrect solutions keep coming up within the same sets of students it becomes very obvious to the teaching assistant what's going on.
Often it is extremely easy to determine, namely from patterns of mistakes.
It's even easier when it's a direct copy. I've had students who copied homework directly, including the same errors, and the same comments. One time, I had a student use another student's file and they didn't even bother to remove the other student's name out of the comments section. They both got 0's for the assignment, and it was a pretty hefty assignment to get a 0 on.
Taking guns away from the 99% gives the 1% 100% of the power.
I mean at some point in your career you're going to have to figure out some problem on your own - and if you cheat during all that high priced training you might as well not even have gone.
if something like that happens, you just pull both students in independently and ask them to explain how the code works and ask a few pointed questions. In general the students who copy do it because they don't understand the problem or how to solve it, some won't even understand the language they are supposed to be working on.
Funny story in college I had an instructor end class early because he was so enraged by a cheating student. I think it was the complete lack of effort that did it, the student turned in the 1st assignment from last year for this years first assignment, despite the fact that the problem had changed.
Borrowing, or reusing code, has always been the norm and is the basis for libraries of routines and procedures. Blatant ripoffs should be obvious but smaller scale plagarism (your word) is hard to determine.
In my classes, the students are expected to complete the assignments using the techniques covered to date. That is because what we learn in the next class often relies on understanding what we covered in the previous when. So when they use a solution that is beyond what they have already learned, they are putting themselves even further behind the curve.
You can't imagine how many times I've had students make a hamfisted attempt to insert something they found on Stack Overflow that uses techniques not yet covered in the course. Since they don't understand what that code is supposed to do, it often doesn't work correctly, or crashes because they thought they could copy & paste without even compiling. I have no problem giving those students a 0 on their assignments.
Taking guns away from the 99% gives the 1% 100% of the power.
I used to run diffs on everyone's code, piped to wc -l to see who had the smallest number of different lines. I'd check those. Usually they were identical down to the whitespace with only variable names changed. Sometimes only comments. Rarely, but it happened, no changes except the filename. Students heard that I automated grading so they assumed I wasn't checking. it was more work to make their code look different than to write it from scratch.
Heh, I'd have been the dick who wrote an automated obfuscator. Flip the indentation from spaces to tabs ( or tabs to spaces ), randomly change ctime/mtime ( within acceptable range ), camelcase to underscore ( or reverse ), use a dictionary to change variables to their synonyms, add generic comments ( ala "palm reading" ), randomly placed returns ( where language appropriate ).
Figure that wouldn't take more than an afternoon to code up.
Could get even crazier by adding the ability to swap out loops ( foreach to while/for ), but those can impact overall grade and is language specific so it'd have to wait for v2.
Mod me down with all of your hatred and your journey towards the dark side will be complete!
Some things you only realize when you have to grade students' programs. Clueless programmers cheat in clueless ways, of course, because if they were clever at hiding that they copied, they could just use this skill to write the programs on their own. Also, the anti-cheating programs we use are way more sophisticated than inept programmers. I use the excellent VPL Moodle plugin (https://moodle.org/plugins/mod_vpl) which has this built-in.
I see this among my students as well.
First, on the side of the students: It is perfectly fine to copy code snippets. How do I safely hash a password? Unless it's a computer security course, students shouldn't be reinventing code like that. That's when you go to StackOverflow and find the canned answer from an expert. Some students (and professors) are confused about this.
Ok, with that out of the way: When plagiarism does happen, it is generally pretty blatant. Two solutions submitted, identical except for the renamed variables. It's almost insulting, that they think I won't notice. Alternatively, they pay someone else to write the program, and then cannot answer even the simplest questions about how it works.
But even if they manage to sneak a plagiarized solution through: how stupid can you be?!?! If students aren't writing the programs themselves, they will fail the exam, where copying isn't an option any more. Or, even worse, they manage to scrape through the first year exams. If they get into their sophomore year, they are allowed to fail a course and repeat it a second time. This is horrible, because they drag out the pain for 3 or 4 or 5 years before failing out of the program.
What a waste of their lives. If they can't handle the material, they're only doing themselves damage by dragging things out. Plagiarism in a technical field, where ultimately you either have the skills or not - and this will be discovered - is just unbelievably dumb.
Enjoy life! This is not a dress rehearsal.
Same here. I used to have a group of students (from a part of the world where plagiarism is more culturally accepted) sit in the lab and work together. With me sitting just a few feet from them, I watched them as they copied code back and forth from each other's screens, so I made it clear that while they could work together in figuring out the algorithm, it was up to each of them to actually implement the algorithm. They nodded, made a gesture of following what I said, and then went right back to it again.
So, since their code ended up being identical, I gave them 0s.
After that, they stopped copying from each other while sitting in the lab, and instead took their cheating out of class. Thankfully, that was even easier to detect, since they'd have signature traits, such as the same number of irrelevant trailing spaces at the ends of specific lines. They started changing variable names and intentionally adding whitespace in irrelavant places, just to make their code look distinct. What they didn't know is that I had a plagiarism detection tool that accounted for those sorts of changes, and sure enough their code always came out as a 100% match for someone else in the class, while everyone else's would be around 20-50% (which was expected, given that there are only so many ways to implement simple tasks).
Much to my disappointment the professor refused to act on any of it, so the best I could do was give them 0s when it was abundantly clear they were cheating. Eventually they started changing the implementation enough that the tool I used no longer showed a 100% match, but at that point they were effectively re-implementing it themselves anyway, which was exactly what we wanted from them in the first place.
They should be taught both, but that means cheating will still be a problem. Collaborating etc is all well and good but students also need to know how to actually write code themselves. Not everything can be assembled from libraries and stackoverflow code with trivial glue logic.
SJW n. One who posts facts.