Students testing limits of turnitin

The Inquirer notes issues when students use turnitin more than once:

Universities, catching on to the sheer amounts of plagiarism going on in Academia, had been asking students to submit their papers and dissertations through programs like ‘Turnitin’ which checks the ‘work’ for any tell-tale signs of copy pasting or wikipedia-ing.

But against all odds, students are proving that despite the quantities of booze consumed on a daily basis to cope with the stress of a 10-hour week, some of their frazzled brain cells remain alive, and even capable of coming up with sophisticated ways of beating the system.

By sophisticated, we mean the devious degree seekers use Microsoft Word’s thesaurus option to change words and sentence structure around a bit before resubmitting their coursework to the program to see if it still picks up any trace of plagarism. Sounds like pure dedication and hard work to us.

Cranfield University was the first to take issue with the phenomenon, writing a new rulebook that decreed that students could only submit papers through Turnitin once, meaning they’d have to be much more thorough with the use of thesaurus before hand.

IPBiz notes that, at some point, the effort in re-casting old work will exceed the effort required for new work.

But perhaps Mike Masnick would call the re-casting a firm of collaboration.

Note also:

But other academics have leapt to the students’ defence, with some going as far as to plagarise a paper to be presented at the Third International Plagiarism Conference at Northumbria University (we were invited) calling for restrictions on Turnitin’s use to be lifted immediately.

***Separately, on the concept of reviewing things -->

Peer-to-patent reported: Peer-to-Patent attracted more than 2,000 peer reviewers. The first 23 office actions issued during the pilot phase showed use of Peer-to-Patent submitted prior art in nine rejections.

From the detailed report:

Of those 23 applications, in nine cases (1 GE, 2 HP, 1 IBM, 1 Intension, 2 Microsoft, 1 Sun, 1 Yahoo!), the USPTO relied
directly on submissions by eight public members of the Peer-to-Patent project to issue final or non-final rejections of the application. In two cases (1 Intel, 1 Sun), the USPTO relied indirectly on Peer-to-Patent submitted prior art. In at
least one case, the examiner made and mailed his determination prior to reviewing the Peer-to-Patent submission.1
From a comparison of the Information Disclosure Statements, identifying references submitted by the inventor and
references submitted by the public via Peer-to-Patent, examiners seem to be more than twice as likely to use a Peer-
to-Patent-submitted reference than an applicant-submitted reference.

The peer-to-patent report misrepresented the concept of "material":

For example, in a GE application (20070174746) on “Tuning Core Voltages of Processors,” the Peer-to-Patent public
cited 3 references, while the applicant cited 10 references. The examiner used 1 reference each from the Peer-to-Pat-
ent project and the applicant in rejecting the claims for lack of novelty and non-obviousness. In other words, for this
application, 1/3 of Peer-to-Patent references were found to be material to the claims, and 1/10 of applicant references
were found to be material.

There was discussion of patent vs. non-patent prior art: In the examination of the first 40 Peer-to-Patent applications, peer reviewers cited non-patent prior art literature 55
percent of the time. In contrast, applicant references for the same patent applications cite non-patent prior art litera-
ture 14 percent of the time. The examiners’ own searches are even narrower. Bhaven Sampat, a professor at Columbia
University and co-author of the article “What To Do About Bad Patents”, states that patent examiners are only citing
non-patent prior art 10 percent of the time. In a recent study of 502,687 utility patents, examiners were found to account
for 41 percent of the citations to previous U.S. patents but only 10 percent of references to non-patent prior art in the
issued patent.

Also note: As with Wikipedia, where the visitor count far exceeds the number of editors who actually write the encyclopedia
entries, only 365 (18 percent) of the 2,000 peer reviewers who have registered on the Peer-to-Patent Web site are
considered to be “active” participants who did the work of submitting prior art in connection with the public review of
the first 40 applications. (...) Of the 365, 104 came from IBM (...) The survey results indicated that all (100 percent) of the participants found the application understandable or easier than most patent applications. The applications were not difficult to read or understand.

***
Eli Kintisch, not known for accuracy in patent matters, wrote in Science in 2006 [Vol. 312. no. 5776, p. 982]:

In hopes of improving the quality of patents and reducing a backlog that this month topped 1 million applications, the U.S. Patent and Trademark Office is weighing an online pilot project to solicit public input on patent applications.

Not clear how peer-to-patent reduces backlog. Is that why the report (above) made the silly comment about "material"?