Comment by eventualcomp
10 hours ago
Maybe there's an allegory to the sciences. But I rather a model be suboptimal in encoding that can be used to generate hypothetical reasonable flags in its unpopulated states, than having an optimal encoding which doesn't answer questions or reveal possibilities in its unpopulated states. Like what does flag #0xFE look like in an 8-bit encoding? I really like the idea of driving compressed trees in reverse with random data to generate new fascinating (valid) sequences.
Also, for compressing into 8 bits - is this not like saying "I can compress all of Shakespeare's plays in 6 bits" because there are ~39 plays, but then the size of the "shakespearezip" is just the sum of the works?
> But I rather a model be suboptimal in encoding that can be used to generate hypothetical reasonable flags in its unpopulated states, than having an optimal encoding which doesn't answer questions or reveal possibilities in its unpopulated states
This encoding already goes too far towards “support the existing flags”.
As a simple example, look at the pattern of the flag of the Comoros (https://en.wikipedia.org/wiki/Flag_of_the_Comoros). Even using just the top colours (FTA: “red, then white, blue, yellow/gold, green, black and orange”), there already are 7 × 6 × 5 × 5 × 5 = 5,250 ways to colour it so that no two neighbouring colour bands are identical. That alone doesn’t fit into 12 bits.
Even limiting one to all colours different, there still are 7 × 6 × 5 × 4 × 3 = 2,520 colourings using the 7 most popular colours. That’s (just) over 11 bits.
And then, you still have to encode the crescent with the four stars.
Okay, but the article literally uses "Union Jack" as a primative.
That is an orthogonal matter and maybe my imagination is limited. But when I think of realistic new flags I do not think of a union jack missing its right diagonal in the corner (e.g. if we decide to decompose the UJ primitive).
We could also complain that the crescent is a complex primitive. Why not define it with three parameters: two circle radii with a center offset?
Choosing the primitives is hard and requires domain knowledge/taste, and finding a simple enough implementation solving for key usecases (e.g. decent coverage of maps, meaningful bit encoding) and descoping (e.g. deciding on no emblems) is also hard.
But I am personally impressed with the primitives the author created and the scope simplifications done to achieve it.