A flood of high-tech plagiarism
More university students caught borrowing computer code from friends or internet sources
For a simple assignment, writing code that would allow one computer to visit web pages located on another, two students in H.E. Dunsmore’s class at Purdue University turned in nearly 100 identical lines of code. Was it a fluke? Or had they cheated?
As he looked over their work, Dunsmore, a veteran computer science professor, saw what he called the smoking gun: boolean done = true; while (!done) {
Because they had written
!done — the exclamation point means “not” — the program translated it as “not true,” which made the program ignore the code that followed, causing it to fail. In a class of about 450, they were the only ones who made that fatal mistake.
“This is pretty strong evidence that one had copied the other,” Dunsmore said. “They later both confessed to collusion.”
College students have flooded into computer science courses across the country, recognizing them as an entree to jobs at companies like Facebook and Google.
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 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.
At Stanford, the alma mater of the founders of Google, Snapchat and countless other internet wonders, as many as 20 percent of the students in one 2015 computer science course were flagged for possible cheating.
And at Harvard, where Computer Science 50 is practically its own brand, with T-shirts, slickly produced videos and an online audience of thousands, the class distinguished itself last fall in a more dubious way: According to The Harvard Crimson, more than 60 students were referred to the university’s honor council, a committee that reviews allegations of academic dishonesty, such as plagiarism, and violations of the honor code.
In interviews, professors and students said the causes were not hard to pin down.
To some students drawn to the classes, coding does not come easily. The coursework can be timeconsuming. Troves of code online, on sites like GitHub, may have answers to the very assignment the student is wrestling with.
“You’ve got kids who were struggling with spending a third of their time on their problem sets with the option to copy from the internet,” said Jackson Wagner, who took the Harvard course in 2015 and was not accused of copying.
Complicating matters is the collaborative ethos among programmers, which encourages code-sharing in ways that might not be acceptable in a class. Professors also frequently allow students to discuss problems among themselves, but not to share actual code, a policy that some students say creates confusion about what constitutes cheating.
The executive committee at Yale, where five students were accused of copying code last fall, though a couple had charges withdrawn, acknowledged as much.
“It is often such a complex task to read these guidelines,” wrote the committee chairman, Paul North, a professor of German, “that the code to be written seems simple in comparison.”
Though coding is a foreign language to most people, the principles of plagiarism are the same as with papers written in English.
Sometimes it is acceptable for lines of code to be identical, if the code is performing a routine task or one that cannot be done a different way. But other times it is a red flag.
Usually, anti-cheating software can uncover these tricks. One, developed by Aiken, is called MOSS, for Measure of Software Similarity. Another program, developed by a British company, Codio, monitors students’ keystrokes; a sudden burst raises questions about where the code came from.
At Harvard, David J. Malan, the CS50 professor, introduced a “regret clause,” letting students who cheat and admit it within 72 hours receive an unsatisfactory or failing grade on the assignment, and avoid further discipline — unless they do it again.
But in last fall’s CS50, an unusually large number of cheating allegations — involving more than 60 out of 655 students, The Crimson reported — were reported by Malan or his staff to Harvard’s honor council.
Malan said one reason for the large number of cases was that instructors were scrutinizing students’ work more carefully than before. He also said that since students could use the regret clause, instructors felt more comfortable going to the honor council when students had passed up that chance.
Two Harvard students who said they were familiar with some of the violations, and who wanted to be anonymous for fear of repercussions from the university, described one of the more blatant examples: A student submitted code using a computer language, PHP, that had been taught in a previous year, but not last fall.
Some students suggested that in other cases there was a fine line between reasonable sharing and improper copying, particularly because the professor encourages students to discuss problems with one another.
“There is a gray area as to how much help one can get,” said Javier Cuan-Martinez, the president of the Harvard Computer Society, who took the course two years ago.
Malan, in an email, said the guidelines in the syllabus were clear. One rule says that when asking for help, you may have your code viewed by others, but you may not view theirs. “Whenever a line is actually crossed,” Malan wrote, “it’s often that one.”