pba-content

pba-content copied to clipboard

Proactive policies & practices

• https://www.turnitin.com/blog/what-is-an-honor-code-and-why-is-it-important-to-academic-integrity
• https://www.turnitin.com/blog/starting-the-year-how-to-establish-a-culture-of-academic-integrity
• https://opensource.stackexchange.com/questions/9254/whats-the-correct-way-to-include-licence-for-part-of-the-code-from-another-proj
• https://opensource.stackexchange.com/questions/11631/how-to-properly-attribute-others-people-work-on-github
• https://towardsdev.com/how-to-cite-programming-code-from-open-source-software-github-and-stack-overflow-22c7e7f231a2?gi=32d91fc22ccd

Tools & techniques

• https://www.turnitin.com/resources/plagiarism-spectrum-2-0
• https://www.turnitin.com/static/plagiarism-spectrum/
• https://www.forbes.com/sites/dereknewton/2021/04/19/a-powerful-new-solution-to-collusion-cheating-and-why-we-need-it/?sh=10021e451d18

I hope PBA won't use turnitin. This article, A Guide for Resisting Edtech: the Case against Turnitin, makes the case better than I can. From the first paragraph:

Plagiarism detection software, like Turnitin, has seized control of student intellectual property. While students who use Turnitin are discouraged from copying other work, the company itself can strip mine and sell student work for profit.

By blatantly copying students are only hurting themselves. They are learning less. If they are not blatantly copying, but simply studying others' work, they may be learning more. I learn by studying others' code all the time.

In terms of blatant copying, any competent grader can detect it. I detected it in Cambridge. We don't need a software to justify our claims that there is cheating.

At minimum, if PBA starts using turnitin or something similar, they need to disclose this to students clearly before students arrive, or even put effort into an entrance exam so that students who don't consent to turnitin's terms of service can opt out.

Agree that turnitin is not very likely to be the tool we choose, the posts here are more blogs on topica that have relevant points imho to consider.

The fear I primarily have is that we have multiple uncoordinated graders on the same assignments that miss student collusion that we have for sure spotted in the past. This actually is most pressing to me in the rust entrance exam that is soon to be OSS with 100% automated grading IIUC, with many hundreds and soon to be thousands of submissions to evaluate for plagiarism, especially once solutions inevitably start surfacing publicly.

In the Buenos Aires wave we also had a few "close calls" with assignment 3 where at least 4 students were required to redo assignments that too closely resembled a preexisting solution (a uniswap v1 like dex pallet) that would have been easy to miss if Shawn didn't know that existed and to look for too close looking patterns in submissions strongly smelling of copy&paste and modify just a bit to work.

So some tooling to help us screen for and at least flag for review possible issues would be good imho. This doesn't mean we explicitly need to use that as definitive evidence or a requirement to act on it even. Just something to assist grading.

Stumbled on this pet-project platform for grading I am liking the features of! Seems like it's not a good direct fit for our needs and Rust support is lacking... It includes a mention of anti-plagiarism tooling: https://help.codepost.io/en/articles/3324264-faq-does-codepost-do-plagiarism-detection

They used a dedicated bit of tooling we could consider (if it's maintained and works with Rust, probably not :frowning_face: ) https://theory.stanford.edu/~aiken/moss/