This paraphrasing of Soros' reflexivity principle from https://allenfarrington.medium.com/the-complex-markets-hypothesis-44c2b2b191d2 is probably the clearest statement of it I've come across.
This also links to Wiener's idea of Augustinian and Manichean demons; problems that, effectively, stand separate from the solver vs. those where you're coupled to the system.
One one extreme, you might have physics; on the other hand, a game of bridge or jiu-jitsu against a highly skilled human opponent. Even if you identify a pattern, the moment you change to take advantage of it, your adversary will adapt.
Actually, that's kind of a fun thought- I wonder how much you can think of the A-M spectrum as just a function of timescale. On a short enough scale, every system is static. On a long enough scale, every system is adaptive.
Boyd's OODAtaktik is trying to shift the adversary toward the linear/Augustinian side of the spectrum, by compressing the timescale and disrupting feedback. A sufficiently Augustinian opponent is just an engineering problem.
Physical laws don't move in an attempt to thwart us, assuming that the idea of shifting fundamental constants doesn't turn out to hold some lurking complexity.