Buying authenticity: plagiarism checking and counter-checking

Alex Tabarrok on Marginal Revolution posted about how the software company Turnitin is not just helping schools detect student plagiarism, but also providing WriteCheck, a tool for checking that a paper is non-infringing. Are they providing a useful service for conscientious students to avoid unconscious infringement, or just playing both sides of the fence, profiting from an arms race where they sell all the arms?

Plagiarism seems to be on the rise. Partially this is because the Internet makes it much easier to find sources to steal from, partially it is because it is also more possible for instructors to detect plagiarism by Googling suspect sentences or using software like Turnitin. Reports about plagiarism also spread further, making more people concerned about the problem. The importance of plagiarism may also have increased: once upon a time individual invention in the arts and sciences was not highly regarded, but from the Romantics and onwards regard for the creative work of individuals has been high – not to mention the profits from intellectual property.

What is morally wrong with student plagiarism? There seem to be four main ways in which it can be wrong. The first is that it gives a false measurement of skill in a student. The second is that it is misses the point of education. The third is that it is deception. The fourth is that it breaches the intellectual property of the originators, or at least refuses them credit for their work. Virtue, consequentialist and deontological ethical views would emphasize them differently, but for any view there are at least one or two ways plagiarism is problematic. It might not be a profoundly immoral activity but more akin to breaking the agreed rules of a game, but insofar one has agreed to play the academic game one should abide by the rules and support that they are enforced. So using plagiarism detection software seems to be a good idea.

Someone not plagiarizing because they expect their teacher to find them out is however not necessarily being moral. From a virtue perspective they are still be sloppy or lazy, just more afraid of discovery than they are disinclined to work well (however, by being forced to overcome their inclinations they might learn to become better). The intentions have not turned into good intentions, so a deontologist would be unhappy with them. From a consequentialist perspective there might be less problems, but the effort going into monitoring students might be excessive or misplaced. So plagiarism detection is not necessarily improving the moral situation much.

What about software that allows you to check whether your teacher’s software would cry foul? Conscientious students might use it to avoid unconscious plagiarism, which seems to be a good use. But obviously, lazy students will instead change wording in marked passages. Or check that the papers they bought will not set off alarms.

If enough rewriting is necessary, then the program would actually force the student to do real work and the result would be good. Rephrasing ideas is another way of learning them. Unfortunately it is likely that what is being examined is just surface characteristics of the text, making this more an exercise in word processing than thinking. And words do not matter as much as ideas: it is still plagiarism if the idea is unattributed. It would be easy to “forget” to add citations and then only add them to flagged passages.

Unfortunately software cannot detect proper and improper citations. Consider the following two sentences: “This sentence is in the public domain and may be used by anybody without attributing me”. “This sentence is the intellectual property of Anders Sandberg and may not be cited for any commercial or non-commercial purpose”. Anti-plagiarism software might detect either sentence in a paper (say a paper on the ethics of plagiarism) and notice that they have been copied from the net, but it cannot tell that the first one is always OK even without attribution and that the second is never acceptable. The semantic content matters in a way that software of less than human intelligence is unlikely to understand.

This is why both Turnitin and WriteCheck are both slightly morally suspect, at least from a virtue perspective. They both encourage laziness and merely looking at surface characteristics of text. Typically teachers recognize plagiarism by suspect deviations of language, be it correct usage of semicolons or shifts in grammatical structure, or that the paper doesn’t fit the educational level evidenced during lessons. Software allows them to conveniently run through essays without having to pay attention to language and educational level could of course allow more focus on content, but I suspect that it might just mean speedier grading. The existence of paired plagiarism detectors and checkers is a bit like teaching and studying just for the tests: the form and rules of an activity are allowed to overshadow its intended content.

The paired software doesn’t help ensure accurate skill assessment except penalize cheaters who do not use it. It likely does not improve learning. It will not stop deliberate deception, just make it somewhat harder. And it cannot enforce intellectual property (or just the courtesy of giving proper credit) without deep understanding of the text and subject. So from most ethical angles it is not doing anything good, just exploiting the increased concern about plagiarism.

That is not necessarily bad either. But real solutions to plagiarism involve making students aware of what is expected of them and why – and having incentives (and enough resources) for students and teachers to strive for excellence in their tasks.