Thread by @sh_reya

i bet this isn't the only plagiarizing paper on arxiv. @sirajraval is getting called out partially because he is an influencer, and we don't want fraudulent influencers in ML.

as a TA, i've seen many plagiarism cases. why do people commit academic dishonesty? my thoughts: (1/10)

let's start with a simple case: student in an intro class copying someone else's code on homework. we allow students to admit that they've cheated, and we'll give them a zero for that assignment. in my four years of TA-ing, i've personally never had a student do this. (2/10)

this leads me to believe most students we catch in the intro classes honestly believe they'll never get caught. maybe it's because there are hundreds of students, and it's unlikely we'll catch all violations. (3/10)

well, we can't catch everyone. so when a student cheats and *doesn't* get caught, they're probably likely to do it again or commit academic dishonesty of a higher degree. maybe they'll google the answers to an exam. submit plagiarized code for a final project. (4/10)

as these instances of fraud build up, people can lose sight of how much effort it takes to do work with integrity. they take on too many classes and set unrealistic expectations on how much content they can produce. (5/10)

in @sirajraval 's case, it's inhuman to produce 2 high-quality videos/papers teaching ML a week! his github repos had large chunks of code copy-pasted from other repos (theregister.co.uk/2019/09/27/you…). (6/10)

to many of us, the act of STEALING someone else's months/years of research is absolutely unthinkable. but there is a class of people out there that have poorly aligned their incentives over time. i honestly believe he didn't think too much about the consequences. (7/10)

how do we prevent people from building an immunity to academic dishonesty? double down on the honor code from day zero. professors -- don't tell students not to cheat; tell them why cheating is bad and what happens to cheaters in the long run. (8/10)

i'd love to see teachers reward students for honesty instead of solely penalizing for dishonesty. give credit for showing up to office hours, writing "i don't know," and/or dropping the lowest homework grade. remove multiple choice (it's too easy to cheat). (9/10)

finally, a disclaimer for anyone who thinks i'm defending @sirajraval -- i am not. i'm merely starting a discussion on academic dishonesty. academic dishonesty is bad. but i'm confident we can educate people to follow academic honesty. (10/10)