> Is it clear how many people have access? If access is fairly limited, I'm not surprised at the low number of examples. I wouldn't be too surprised either, wh

> Is it clear how many people have access? If access is fairly limited, I'm not surprised at the low number of examples. I wouldn't be too surprised either, what surprised me was the large number of people who explicitly said words like "many examples" and then couldn't link more than one. > Here's an example from their own docs: Thank you! This is definitely a much stronger example than the Quake one. Looking at the git repository it seems like this is generic startup code for a university assignment, so it probably shows up a significant number of times in the training data. While it definitely makes sense to interpret a piece of code showing up several hundred times in the exact same format as being okay to straight-up copy+paste (e.g., imports, some boilerplate in more verbose languages, keyboard input switch statements), this does seem to highlight that Copilot can't distinguish between code snippets that are always identical because that's just the correct way to do it, and code snippets that are always identical because only one person did that thing, and it just happens to be in hundreds of repositories for one reason or another.