It depends what kind of maps. Multiple consistent maps are clearly a good thing (like switching from geometry to coordinates and back). Multiple inconsistent ad-hoc maps can be good if you have a way to choose which one to use when.
Wilson doesn’t say which he means, I think he’s guilty of imprecision.
I think he means that people choose not to think about any map but their favorite one (“their way of looking at reality is the only sane way of viewing the world”), to the point where they can’t estimate the conditional probability P(E|a) of the evidence given not-A.
The link with Aristotle seems weak. But the problem obviously makes it harder to use “the logic of probability,” as Korzybski called it, and Wilson well knew that Korzybski contrasted probability with classical “Aristotelian” logic. (Note that K wrote before the Bayesian school of thought really took off, so we should expect some imprecision and even wrong turns from him.)
Or accept that each map is relevant to a different area, and don’t try to apply a map to a part of the territory that it wasn’t designed for.
And if you frequently need to use areas of the territory which are covered by no maps or where several maps give contradictory results, get better maps.
It depends what kind of maps. Multiple consistent maps are clearly a good thing (like switching from geometry to coordinates and back). Multiple inconsistent ad-hoc maps can be good if you have a way to choose which one to use when.
Wilson doesn’t say which he means, I think he’s guilty of imprecision.
I think he means that people choose not to think about any map but their favorite one (“their way of looking at reality is the only sane way of viewing the world”), to the point where they can’t estimate the conditional probability P(E|a) of the evidence given not-A.
The link with Aristotle seems weak. But the problem obviously makes it harder to use “the logic of probability,” as Korzybski called it, and Wilson well knew that Korzybski contrasted probability with classical “Aristotelian” logic. (Note that K wrote before the Bayesian school of thought really took off, so we should expect some imprecision and even wrong turns from him.)
Or you could always just average your inconsistent maps together, or choose the median value. Should work better than choosing a map at random.
Or accept that each map is relevant to a different area, and don’t try to apply a map to a part of the territory that it wasn’t designed for.
And if you frequently need to use areas of the territory which are covered by no maps or where several maps give contradictory results, get better maps.
Basically, keep around a meta-map that keeps track of which maps are good models of which parts of the territory.
Yeah, that should work.