The idea of auto-attribution if copilot surfaces licensed code is best because then it keeps the copilot user honest where the code is coming from and honor the original license.
The idea of auto-attribution if copilot surfaces licensed code is best because then it keeps the copilot user honest where the code is coming from and honor the original license.
I’m genuinely disturbed to see how many people in this thread think that casual plagiarism is the norm for “usual devs”.
I'm disturbed it is likely the reality.
None of the code I work on is public, so attribution is pointless in the first place.
Devs often look at code to see basic syntax, understand algorithms, etc. There is absolutely nothing wrong with this. One should draw a line somewhere, but to say I need to attribute […somevar] every time I use it because I happened to see it one time on a blog post is silly.
A thought experiment may help: Scrape Github for all unique strings longer than X and store in a file with a timestamp and owner. How large does X have to be before attribution is required? If not length, then how do you determine whether attribution is required?