Something I’ve noticed that I didn’t expect is that people are able to guess a lot of albums in only one or two squares. Some of this is luck and image layout paradigms being surprisingly consistent (titles at the top!), but we also may have better context-based image memory than we think. There’s an obscure setting to increase the difficulty, which makes the squares smaller (or larger) to combat this.
Interesting idea about the reveal! I’ll consider adding that (or some cheap version of placing the image in the success modal at least).
Author here. I was inspired to make this by thinking about how Wordle, a game where a word is slowly unraveled by guessing letters, could apply to images (and specifically album art).
Perhaps clone is a poor word here and inspiration would be more accurate.
The main objection to computer-aided proofs is not that they are more difficult for humans to understand.
The main objection is that most proof-assistants use a different logical foundation than modern mathematics. Modern mathematics is built on ZFC[0] whereas most proof assistants, such as Coq, Isabelle, Agda, etc use different logical foundations, such as the Calculus of Constructions[1].
Many important results in modern mathematics are not easily stated or proven in systems such as CoC[1]. For example, Brouwer's _Fixed Point Theorem_[2], a pretty bog standard result in Topology that is useful to proof many things in Functional Analysis, has a clear statement in ZFC, but does not, to my knowledge, have an equivalent in CoC (and if it does it will be stated radically different).