It still seems like we can figure out the rules of chess better by observing many games than by sitting down and reasoning about what the rules of a tactics gam

It still seems like we can figure out the rules of chess better by observing many games than by sitting down and reasoning about what the rules of a tactics game played on an 8x8 grid would be; sure we'd miss uncommon moves like under-promotion and en passant until we had a very large sample, but to me the chess analogy is an argument for empiricism, not against it.