A leading examination board is launching a pilot project to scan pupils’ GCSE and A-level coursework in an attempt to prevent plagiarism undermining the value of school qualifications. source.
Professors have been doing that for years, especially on computer code projects where it is relatively easy. This was mostly local and plagiarism was still hard to prove unless it was a 1-to-1 copy, or near to it. Today, many on-line services provide ready or custom made essays (“remember to change my names, dates and IDs to yours”) and some provide the comparisson service. I was surprised to find that even MA dissertations were available (for about $8000!)
turnitin.com seems to be the big-dog of plagiarism detection. These services are good in that they deter otherwise honest people from cheating in a moment of weakness. They may also help catch perpetual cheaters that shouldn’t get away with it. Here are my further observations:
1. I think that universities/institutions would rather ignore cheating if they can, even if it is right in front of their eyes. This is because they are either delusional about it by refusing to believe there could actually be cheating going on in their super-duper-establishment, or they are afraid that it would taint its reputation if reported. If major cases of plagiarism or data falsification occur that can be ignored no longer, they are reported. If there are minor cases they will slap someone on the wrist and make the issue go away as fast as they can with the least amount of people knowing about it.
In short, institutions would like to verify that the people they admit are not cheaters, but will be light on accusations once they are already in.
2. As the database increases, more innocent people would be accused of cheating:
Turnitin already claims more than 7 million subscribers worldwide, and compares submissions with a database of books and journals as well as more than 4.5bn web pages. It also checks them against its own library of more than 10m previously submitted papers.
It is inevitable that some unfortunate, hard-working souls would have their lives ruined by an algorithm. When people blindly rely on technology, it would be hard for the accused to prove that the work was actually theirs.
3. Since this is a profitable business, there is an incentive to figure out the comparison algorithm and create services that make sure you will not get flagged. I’m sure we’ll see these in coming years. If the process of modifying the “base work” takes more, or equal, time to writing it from scratch, then the detection system should be considered a success. However, given the problem, this would be almost impossible to achieve.