The two most popular plagiarism-detection programs are Turnitin — widely used in higher education — and CrossCheck — widely used by scholarly publishers. Both programs rely on software developed by iParadigms, a company headquartered in Oakland, CA.

Last week, Inside Higher Ed published a story suggesting that a lesser-known iParadigms product called WriteCheck is allowing the company to work both sides of the plagiarism game. WriteCheck is a student-only version of the plagiarism-detection software. For $6.95, a student can submit a paper to WriteCheck and will receive suggestions for grammatical changes, word choice improvements, and the like. In addition, the software checks for proper paraphrasing, citation, and quotation practices. In short, the student also apparently learns if any part of the paper would be flagged by Turnitin. Students can submit up to three revised drafts. There is complete confidentiality to the service, and the papers are not added to any database.

David Harrington, an economist from Kenyon College, has been looking askance at Turnitin and its ilk. He thinks there’s something wrong with on the one hand selling the guard against plagiarism while on the other selling the way to walk past the guard:

On a fundamental level, I wondered about the ethics of a company that is supposedly marketing this thing that is trying to suppress plagiarism when it sort of cynically offers a product which allows a student to get away with plagiarism. To have this system where you are selling a security system to colleges and universities that is trying to protect against plagiarism and then at the same time selling a system to the students that allows them to subvert the security system that you just sold to universities makes no sense.

Harrington’s skeptical of the effectiveness of Turnitin overall, and conducted an experiment to see how difficult it would be to fool it just by modifying a few phrases and choosing the right sources to plagiarize from. He conducted an experiment — noting that while Turnitin crawls 1.4 billion Web pages, that’s about 10% of what Google crawls. In this experiment, he took a book called “Shoplifting: A Social History” partly because he suspected its author of plagiarizing news accounts, partly out of a sense of irony. After all, shoplifting and plagiarism are cut of the same filching cloth. This becomes clearer when Harrington talks about the similarities between the two:

Just as shoplifters slip the goods they steal under coats or into pocketbooks, most plagiarists tinker with the passages they copy before claiming them as their own.  In other words, they cloak their thefts by scrambling the passages and right-clicking on words to find synonyms.  This isn’t writing; it is copying, cloaking and pasting; and it’s plagiarism.

Harrington found that both Google and open source software relying on Google did a better job of identifying plagiarized passages in “Shoplifting: A Social History.” He asked Turnitin about this, and was told that crawling takes time, and that many news sites aren’t crawled because they’re behind paywalls. Later, Harrington added the New York Times material to the Turnitin database as a student paper, and Turnitin found a similar level of plagiarism in the shoplifting book. The software works, but the database has large limitations for general materials. However, Harrington notes that Turnitin needed his help to work properly, concluding that:

It needs the help of instructors who are willing to investigate suspicious papers; otherwise, greater reliance on Turnitin could lead to more plagiarism.

CrossCheck works a little differently, with CrossRef providing a way for iParadigms to crawl much of the scholarly literature. There are — at least conceptually — fewer holes in the database.

However, the question of creating a service allowing submitters to preflight their papers against the anti-plagiarism database becomes all the more compelling when you accept that the software works; for scholarly publishers, who might be helping to make an even better database, the question gains greater urgency, since the database potentially works for everyone involved.

Apparently, iParadigms (and its iThenticate) product are also working both sides in the scholarly world, as well, with iThenticate offering a version for authors and researchers to pre-screen their papers — or, as iThenticate puts it, “Do your own editorial review before submitting your manuscript for publication.” iThenticate charges $50 for this service, which allows one paper to be checked up to five times.

The analogies used include an arms dealer who’s selling to both sides in a war, or a store security vendor who also sells aluminum-lined shopping bags so shoplifters can avoid detection. Both are apt, but there’s a twist with WriteCheck — you can resubmit a few times, tuning your cut-and-paste work until you get it just-so. The game then shifts from whether you plagiarized or not to whether you plagiarized well or clumsily. Then, the analogy shifts to one based on the ability to navigate barriers — something more akin to the injustice of people with means avoiding prison while those without end up in the hoosegow. Or perhaps the analogy of a protection racket is more appropriate — shame if something happened to your nice little paper.

In short, it doesn’t seem right. Imagine if you caught an anti-virus company also selling hacking protocols, a mirror-image business that undermines the purported goal, whether it’s preventing plagiarism or preventing computer viruses.

With the volume of submissions going up, the high stakes around publication, and the proliferation of outlets, plagiarism-detection software has provided publishers and professors with some sense of security — a sense that it would be harder to scam a publisher or instructor by putting through a plagiarized paper. However, it now seems both sides may be reaching a new level of stalemate, with iParadigms in the middle, happily taking payments from symmetrical channels without providing either with more than an illusory sense of progress.