Summer of Math Exposition

This was seemed a bit confusing to me (and my field is information theory). I think that a lot of emphasis should have been put on the fact that the reason why this does better than Huffman is because it considers dependence between characters. If one encodes i.i.d. symbols from a distribution, the Huffman code is optimal, here the fact that the distribution changes at each symbol indicates that the distribution is actually a conditional. Finally I think that the argument about "patents is bad" is either too short or should not appear. I think it is a very important point but is slightly underdeveloped in this exposition.

1. Identify your intended audience. E.g., should they understand the tree diagrams shown during the first minute? 2. Explain "compression" and "coding" (don't ignore "decoding"). E.g., compare "I WISH I WERE A BIRD" with "IWISHIWEREABIRD" and "WSH WR BRD", etc. 3. Summarize your key result(s) before embarking on prolonged analyses involving various discrete-to-continuous transitions.

This is a very competent video on a somewhat technical topic. I personally appreciate it a lot, because I like information and coding theory and love to see neat topics like this being explored on youtube. You do a good job at the beginning of talking about the social and technological context behind the algorithms you're describing. Your insight that the fundamental idea behind the codes you're discussing is "just encode it as a big binary number" is very catchy and memorable, that's going to stay with me. I like that you acknowledge Reducible's video on Huffman codes, I think it's both good practice to point to the good work of others, and also it helps to connect your video to other things a math-interested youtube viewer may have seen. I also think it's cool that you included your python implementation of this stuff, I think including a resource like that is definitely above and beyond. Also, I liked all of your example texts being silly memes. I felt like your exposition got more difficult to follow in the section on asymetric numeral systems. I felt like connection between the algorithm you were describing and the basic idea of "add a 1 to the beginning and encode the number" got obscured. You also raised some questions you didn't really answer, for example, you brush aside the question of how you produce the labeling needed for ANS, simply saying you "use some heuristic". On the other hand, you also take care to make a note of details about renormalizing X if C(X) is above the threshold T, and I'm sure those details matter, but I felt like I was lacking a little bit of the motivation for why they were coming up. Overall, I felt like this section of the video needed some more refining to make sure it flowed more coherently, and didn't leave you hanging on some details and motivation. I see in your video description on youtube that you ran out of time to work on this video more, but I hope that you do work on it more! I think it's a great topic and I'd love to see what it can become.

I almost wanted you to go in harder on the law part. This was a great explainer though and had great visuals and music choice.

I really have to watch it a few times to understand it. It's highly compressed content:-)

super information dump - brain might explode. good accessibility of content, nice topic introduction and memorability. everything followed, however jargon wasn't as easily addressed as it could have been

interesting informative and excellent animations

really excellent work! got this organically in my recommended feed prior to registering as a judge and i loved it

Goal Orientation: 4/10 Novelty: 2/10 Thought-Provoking: 2/10 Comprehensibility: 1/10 Technological: 6/10 Overall Average: 3/10

Excellent

Motivation: Good. The video has a hook of non-mathematical reasons why this class of algorithms is being suppressed, as well as a promise to explain why those algorithms are actually better. That invites both novices and people with a bit more experience. Clarity: Probably the weakest point of the four here. The video jumps in a bit too early with technical language, in contrast to the invitation laid out from the introduction. Novelty: As someone without a lot of cryptography background, but generally strong math comprehension, I think I was the target audience for the added value of your animations on this topic. That said, it's hard for me to compare with existing sources. Memorability: I was really hoping for an example that I could take home, perhaps running through the video. While the video wasn't lacking in examples, none really stuck out.

Very well presented and clearly explained.

This is a very interesting topic, and your tone and style are pleasant and enjoyable. I think that more simple examples to clarify the encoding algorithms would have helped to make the content easier to follow and remember.