Calling this a step backwards is patently unfair. The author is explaining a simple algorithm, simply enough for even beginners to implement, for building layouts of a given connectedness. They are not making any kind of argument that the results can be used as-is in any particular genre of game.
Honestly, you might as well attack an A* pathfinding article on the grounds that game units shouldn't necessarily take the shortest path to their goal. Algorithms like this are tools in a toolbox - obviously tweaking them is what makes any given game unique, but beginners still need to start somewhere.
(Also - Nethack? The game where you find potion shops eight levels underground that can only be reached via three rooms full of soldier ants, etc?)
> game units shouldn't necessarily take the shortest path to their goal
They shouldn't? This can in specific cases lead to weird looking behavior, but nobody is suggesting it is an 'abstract movement strategy' that isn't supposed to correspond to anything a person would do. The reason shortest paths are used is that they do correspond to the way people move, to a first approximation.
> simply enough for even beginners to implement
Really? Using delaunay triangulation, minimum spanning trees and relaxation? With no sample code? This isn't a beginner tutorial.
> can be used as-is in any particular genre of game.
Really? You don't think this is specifically for roguelikes/roguelites?
> Nethack
Nethack definitely has crimes against mimesis too, definitely. In fact it has that 'old adventure' feel. I was using it just to show that the challenge spaces aren't abstract.
"Shouldn't necessarily" means shouldn't in some cases, should in others. The point is that something like a pathfinding algorithm is not meant to distinguish between those cases, just as the worldgen algorithm in this article is clearly not attempting to create genre-specific dungeons. It gives you raw information, which you'd need to refine if you wanted to have a feel that's specific to a given game.
> This isn't a beginner tutorial.
Beginner gamedev. Is this a distinction that you think negates my argument? It sounds like you're listing up everything in my post that you think people could disagree about.
> You don't think this is specifically for roguelikes?
Whether it's about roguelikes is neither in evidence nor at issue. The parent post took issue with the article for not making maps that felt like a prison/mine/fortress/etc., and I'm saying it clearly didn't attempt to.
> Nethack ... the challenge spaces aren't abstract.
Every game is abstract in some ways and not in others. But within that universal truth, Nethack is most definitely not a game that even gestures in the direction of realistic world generation, and is a counterexample to the overall point of your post.
Honestly, you might as well attack an A* pathfinding article on the grounds that game units shouldn't necessarily take the shortest path to their goal. Algorithms like this are tools in a toolbox - obviously tweaking them is what makes any given game unique, but beginners still need to start somewhere.
(Also - Nethack? The game where you find potion shops eight levels underground that can only be reached via three rooms full of soldier ants, etc?)