Dan said:
Turnitin's strongest argument is that their database is transformative and contains no copies of the papers, only relational data *about* the papers. The hashes have no separate value other than for plagiarism detection -- the hashes are not substitutes for the papers in the market (assuming there is any market for student papers). There is a long line of cases holding that intermediate, temporary copying in the process of creating a transformative work is fair, bolstering the argument that the intermediate copy is fair.
Actually, Turnitin has posted an extensive analysis of these issues as PDFs (pertaining to each country of operation) on their website at http://www.turnitin.com/static/usage.html. I've only read the 10-page Australian version, but it seems that Turnitin's strongest argument is, firstly, to claim "Service Provder" status and therefore "Safe Harbour" provisions under the DCMA and secondly, to ensure that all submissions of student works are covered by a non-exclusive licence. In other words, all student work submitted is correctly licenced and hence THERE IS NO INFRINGEMENT OF THE STUDENTS' COPYRIGHT when the copy is made. *Of course* Turnitin makes a copy of the paper - they don't seem to dispute this - but their argument is that they're entitled to because they're a service provider and that in any case, the students give them permission to copy the papers. Whether or not it's a hash and whatever a hash is are completely irrelevant. Stop ignoring the obvious. What is interesting, though, is the reliance of Turnitin on the universities informing students that their work will or might be submitted and what this means ... that scares me a bit. They also rely on jurisdictional issues by claiming that any "reproduction if it is stored on the Turnitin server while it is subjected to the checking process" performed is "not performed in Australia" (it's done in Oakland, California), hence Australian Copyright laws cannot apply. That might be interesting to challenge. There is also a supplementary opinion directly addressing the issues I raised in my last post of the possibility of the copy (ie the "hash", which they refer to as a "digital fingerprint" being an "adaptation" (or "derivative work" as Dan discussed) and of the derivation of commercial benefit from the students' work. This opinion is a response to comments by the President of the Melbourne Postgrad Students Assn in The Australian newspaper and they conclude it is unlikely that the hash is an adaptation for the purposes of the Copyright Act 1968, and that, in any case, it happens in Oakland. In all of this, the lawyers' only reservation seems to be over the question of whether, in the absence of an explicit licence granted by the student on submission of their assignment, an implied licence is granted and what this might cover. They conclude that it is most likely that "an implied license exists for assessors to utilize a plagiarism service such as the one conducted by Turnitin." They suggest that the best risk-minimisation strategy is to get the students to sign a licence on their submission cover sheet ... If a licence exists for a copy to be made, there is no infrigement. QED. Which brings us back to the point of whether a uni can reasonably and ethically expect students to grant such a licence as an administrative proceedure - but that's discussed elsewhere. Cheers, Hughie