Here's the math: http://lostmathlessons.blogspot.com/2016/03/bouncing-dvd-log...

Only if the tv is a square, which most tvs are not.

Curious to know if you’re under the presumption that TVs are square? Or are you imagining that the logo doesn’t move at 45 degrees?

Both seem obviously false but I’m struggling to think how someone could otherwise have come to your conclusion.

Try on paper originating it from a corner on a rectangle where the initial path is just offset from the diametrical corner.

let's wait and see https://codepen.io/CaptainKeyframe/pen/eYKQVKE

You clearly know more about this than me, because I'm currently Googling to understand what "diametrical corners" are. But in a simple implementation, wouldn't the x and y velocities both be reversed at once with a corner hit, both side and top/bottom collisions having been detected "at once" in the same part of the code between updating positions? And then the logo would just bounce out exactly the way it came in?

"diametrical" is fancy for "opposite".

You are correct.