To add something a little more practical than "it's fun" or "it's philisophical" -

For someone packaging things with Nix, it's tempting to think of derivations as just a builder - you write your scripts, accept your inputs, and your result ends up in the store.

The article suggests, taking advantage of lazy evaluation, that you can put a whole state machine in Nix (i.e., lift the DSL for describing states into the Nix language). If you had a team of developers that need various customizations of something, they could describe it in Nix, and benefit from the caching of intermediate steps. If someone's just changing something at the end of the long string of customizations, they only need to build their changes.

A more concrete, if very niche, example - Back in the days when we were all training our own deep learning models, my company had assembled a list of facts for every product we could find on the internet. Our data scientists wanted to be able to test datasets with different aggregates of the lists - we needed to both build the training set, and compute it at runtime for new products. There was some compositionality to the aggregates - one run might use the mean, another would subtract the mean from every element, etc. It was time consuming and annoying to manage. One of our developers made a DSL in Haskell to describe the set of aggregates desired for a given dataset, and needed to build caching for each column. The technique the article suggests could have done this out of the box (at build time, even) - build a function and a vector for each aggregate, throw them in the nix store, and then wrap them up as a training set and a server.

The implementation is superficially silly, but I think this can also be read as a meditation on exactly what distinctions we're making when we use terms like "program" vs. "process" or "input" vs. "dependency". Perhaps a cousin to the idea that compression formats represent specialized instruction sets for bitstream generation, or to static single assignment. It makes explicit what is usually implicit in emulator "movies", that each frame has the state of the prior frame as a dependency, even though that relationship is normally thought of as a side effect of the passage of time.

The underlying principle of translating those implicit temporal dependencies into explicit data dependencies very well might have applications in optimizing high-accuracy computer/console emulators. The usual implementations iterate over some variable that represents the passage of time and compute all possible side effects, which is expensive. Many compilers already do a related static optimization by eliminating intermediate variable mutations that can be proven to have no side effects.

(also, if you're not familiar with the "tool-assisted speedrun/superplay" tradition of optimizing video game input sequences as a flavor of metaprogramming, see https://tasvideos.org/NewcomerCorner)

It's 'fun', for some folks 'nostalgic', possibly even 'interesting', in a casual sort of way. You apparently aren't familiar with these concepts. No need to hate on it.

It's just a fun hack. I think it's neat.

Answer honestly: what is the point of yourself or anyone for that matter?