Problem Set Problem: Cheating

Let’s be upfront: cheating—most commonly in the insidious form of copying or other illicit collaboration—happens quite a bit at Harvard on take-home assignments, especially problem sets. Worse, Harvard’s prohibitory policies on inappropriate collaboration, while explicitly stated in the Handbook for Students and often in course syllabi themselves, are enforced about as effectively as anti-terrorism legislation is enforced by the Palestinian Authority.

Trusting Harvard students never to cheat on problem sets is like trusting us not to use false identification to obtain alcohol (I don’t know anyone who would do that, do you?) or trusting us not to download copyright-protected music on Kazaa (I also don’t know anyone who would do that, do you?).

True, individual instructors often grant permission for varying degrees of cooperation on assignments. But the trust placed in students not to cross the line is too easily, and too often, abused. If you keep your eyes open, indications of the prevalence of cheating rear their ugly heads quite readily.

Look, for instance, at the example of CS 50—the computer science department’s introductory course—which is one of the few Harvard classes that, using special software, automatically checks all assignments for plagiarism. Because of this practice of checking every student’s work, CS 50 manages to garner, its website reports, an “extraordinarily high percentage of the cases of undergraduates required to withdraw from Harvard College for disciplinary reasons.”

This has some very disturbing implications. If this is the case in CS 50, then simple reasoning dictates that the situation is in fact far worse in many other classes. CS 50 is a required course and / or a building block for a number of concentrations in the applied sciences. As such, many of its students have an external incentive to learn the material in addition to simply wanting a good grade—an incentive that almost all Core classes utterly lack, as they are generally the first and last course in that field which most of their students ever take.

Add to this that most classes have no way to check for copying, and you have fertile territory for potential cheaters. Students are usually placed on the honor system not to cheat—and, as the example of CS 50 indicates, that means trouble.

The status quo would be acceptable, though still unattractive, if cheating were a victimless crime. But most large science classes—incidentally the places where copying on problem sets is most apt to occur—are graded on a curve. This creates a zero-sum situation, where any points people gain by cheating do not simply help them, but hurt others.

If you are not a Harvard student, you may be surprised at the phenomenon I have just described. If you are a Harvard student, your reaction is, perhaps, threefold.

First, you likely have a good idea of what I am talking about (“Of course people cheat on problem sets. It’s so easy to do so with complete impunity, there’s little incentive not to.”). Second, you may think I am a complete jerk for mentioning it in such a public forum (“I need those points!”). And third, you may be wondering if I am going to make a suggestion about how to fix this seemingly intractable problem (“I wonder if he is going to make a suggestion about how to fix this seemingly intractable problem.”).

I am indeed going to offer one potential solution, but before I do, let me address one oft-suggested idea—the implementation of an honor code—that would not be effective.

The biggest problem with honor codes is that, well, they don’t work. If a student is a cheater, is it really feasible that putting his signature on a piece of paper is suddenly going to infuse him with integrity? I don’t think so—and neither, I suppose, do the nearly 50 ex-University of Virginia students who are now looking for a new school, after they made a travesty of their school’s vaunted honor code by cheating off each other in a physics class last year.

If not a feckless honor code or some other similar nostrum, then what?

There is a simple and elegant idea that I have not yet heard suggested, but which would solve many problems while creating few: reduce the weight of problem sets.

In many, if not most, large science classes, problem sets count for at least a third of the course grade—and the figure often is closer to a half. Capping the weight of problem sets at, say, 10 or 15 percent of the course grade would instantaneously solve almost all of the troubles associated with Harvard’s cheating problem.

If implemented, this proposal would drastically undercut the benefits of copying friends’ answers on assignments. More importantly, people who do not or cannot find others to work with on problem sets would no longer be at such a tremendous and unfair disadvantage come grading time. If necessary, weekly quizzes could be used to replace the feedback lost by lowering problem sets’ importance.

Ultimately, though, it is immaterial whether this specific solution is satisfactory; either way, the administration needs to find a way—some way—of stymieing cheaters. To turn a blind eye to this problem—and yes, it really is a problem—is to ignore a situation which causes real, tangible harm on the students who least deserve it: the ones who don’t cheat.

Zachary S. Podolsky ’04 is a classics concentrator in Currier House. His column appears on alternate Thursdays.