TBH, for many NP-hard problems algorithms exist that "almost" solve those problems, giving an answer that is at worst several times worse than optimal, and slime moulds use the same approach.
True, it's not like P == NP under whatever paradigm the slime mould's using, but it's heuristic is a lot better than our state of the art a lot of the time and it's expensive enough to simulate that it'd sometimes be worth getting an actual slime mould in a lab on a custom lattice instead of trying to book 1000 supercomputer hours or whatever, which is extremely cool.
TBH, for many NP-hard problems algorithms exist that "almost" solve those problems, giving an answer that is at worst several times worse than optimal, and slime moulds use the same approach.
True, it's not like P == NP under whatever paradigm the slime mould's using, but it's heuristic is a lot better than our state of the art a lot of the time and it's expensive enough to simulate that it'd sometimes be worth getting an actual slime mould in a lab on a custom lattice instead of trying to book 1000 supercomputer hours or whatever, which is extremely cool.