@DJMcMayhem sorry, but no. I don't like that channel, as his titles are so clickbaity, and he moves so slow in his videos

That video alone didn't give me high hopes either. (I did fail that riddle the first time I saw it as well though.)

I can hardly believe this video was 30 minutes long.

It's been a while since I've played Mario Kart. Last time my friends did a thing where we split off into teams of 2 and shared controllers (one person controls steering, other controls gas/breaks). You'd assume it would turn out horribly but then my team got 1st place on rainbow road without falling off.

Haha wow, sounds crazy and kinda fun.

@El'endiaStarman I remember watching that a while ago

The ending was fairly slow (because you knew what was going to happen)

but I did enjoy it

So, thinking about Machine Learning: It seems to have one fairly large problem: Data seems rather "unexportable"

as an example: YouTube has done a bunch of machine learning to figure out what kind of music to play for me

I bet Spotify does the same sort of thing

Now, it's not that hard to write a program that would "convert" a YouTube ID to a spotify song ID

But if you tried to transfer a user over? That's impossible.

You'd basically have to build another ML algorithm that understands the YouTube data, and can transfer it over

And since the input is complicated, and the output is complicated: it'd take even more data to train than either of the 2 took to train