Summer of Math Exposition

Across the board I think it’s an average math explainer.

Novelty and memorability is lacking. I do think this structure of video thrives as an addition to lecture material, but it is not for the mathematically curious. Too mathematically focused, and although the animations were well done, nothing visually memorable.

Very good

Motivation is well delivered and the explanation is quite clear. Entropy as a topic is very popular on the internet, and content is quite similar to 3B1B’s recent entropy videos.

So many equations! Was there a way to explain this visually or in a simpler way?

This video helped me to understand the concept of data compression. It explained why typical set is small, but appears in high probability intuitively. However, this assumes that each token is independent, which is an unrealistic assumption in practice. It would have been more interesting if this video compared the theoretical efficiency of the encoding system and the practical efficiency in real world application.

This is a good polished looking video. My one concern was the number of equations without true explanation in the video. I felt that these would be appreciated by some of the viewers, but without backing they don’t present much in the way of education for newer to the subject viewers. I think the core idea that I took away was that we can successfully assign short codewords to the family of most likely sequences. The most interesting thing to me was that these typical sequences don’t inherently include the most likely sequences. Overall great video, thank you for making it!

A well made and explained video. While punchy and fast, the quick rate of dialogue can make some words hard to pick up. The biggest thing I would suggest would be a more granular worked example with visual accompaniment, worked to completion, such as with DNA nucleotide distributions or keeping up the usage of that coin flip example. More breaks to explain what the current formula represents in relation to previous formulas mean would be good, especially if they recontextualize prior formulas to fit the current framing here. At ~3 bits the average information per symbol is plotted for arbitrary n sequences. It seems to follow a uniform distribution. This is a great opportunity to introduce, visually, some sequences will have a higher information per symbol, and some will have a lower, and this is related to the frequency of occurrence of that sequence to begin with. It took me a couple of rewatches to be convinced of what was being told, but i eventually got it, and appreciate what went into it and what it did.

Very solid! But overall a bit too technical and mostly just the standard arguments.

Strong points:

• Good speaking voice; well recorded audio.
• Good take on explaining entropy; there’s a million videos about that, but yours works well, particularly in context.

Neutral points (strong?):

• Listened to the video while cleaning the house; didn’t watch video except for some glances. As a computer scientist, but without knowing asymptotic equipartition, I was able to follow along nicely. (Don’t make me take an exam, but I’m happy to have now heard of it.)

Weak points:

• Don’t love the title. but maybe I’m just personally allergic to “ThE GeNiUs IdEa” kind of titles ;)