I don’t think you’ve acknowledged the reasons that equivocation is so common, and the benefits it brings. There’s the obvious motte-and-bailey advantage for adversarial or partly-adversarial communication. There’s also a big advantage in generalization, as it’s often impossible to be truly literal in any reasonable amount of time.
I’m a big fan of speaking plainly and making most statements as propositional predictions, among trusted people, on topics we’re cooperating on. Even then, my tendency to generalize and equivocate gets in the way. Heck, even in my private mind, I shy away from some parts of complex questions, and let myself handwave past them.
Policy debates (at any level—household to global) are orders of magnitude more complex than simpler examples, and there’s probably no avoiding both rhetorical tactics and model compression that intermix as ambiguity, with implicature as a common way of dealing with that ambiguity.
Policy debates (at any level—household to global) are orders of magnitude more complex than simpler examples, and there’s probably no avoiding both rhetorical tactics and model compression that intermix as ambiguity, with implicature as a common way of dealing with that ambiguity.
I’m not really against rhetorical tactics in the sense of being strategic with one’s words; I’m against shortsighted rhetorical tactics which stop the speaker themselves from realizing the truth, or burn the commons by inserting lies into the public discourse, etc. Most rhetorical tactics are actually quite transparent if you take a moment to try and apply logically similar arguments to politically-different examples, etc. (I’m not saying rhetorical tactics are automatically OK if they don’t fit into that category; but I am saying that most rhetorical tactics just wouldn’t work at all in a world where they were uncommon and listeners applied ordinary truth standards to discourse rather than applying the current rhetoric-friendly rules.) Or to put it a different way: imho, all one needs to do to (mostly) eliminate (most) empty rhetoric is establish common knowledge (or close to common knowledge) about those rhetorical patterns.
(On the other hand, I agree that it’s practically impossible to eliminate all ambiguity, particularly what’s due to model compression.)
It’s an important point. Equivocation can also be framed as the heuristic inference “if A and B are highly correlated, replace A with B” (but made implicitly, without even making a distinction between A and B, because A and B have been clustered together at some previous point).
The real point is to have the flexibility to be able to make fine distinctions or big generalizations as appropriate for a situation, rather than being stuck in whichever way you think about it.
However, making distinctions seems especially powerful somehow. You want to be able to do both, and you want to actually use bot, but if it were only a matter of moving one direction or the other, it seems somehow better to move in the “distinctions” direction.
One point is that distinctions tend to be more concrete and objective, and therefore, more “portable” between brains, and more scientifically testable.
Also, making more distinctions temporarily for the sake of putting things back together later (in a hopefully better way) can be a great way of making progress. The other direction (temporary abstraction) is incredibly useful as well, but seems a bit weaker (it’s like analogy as opposed to logic).
Another point is that we have to cluster a lot to fit things into our limited working memory. So it’s more probable that by making extra distinctions, you can see things that you’d miss otherwise. On the other hand, since everyone is abstracting all the time (to handle limited working memory), the probability that you’ll hit new insights that way seems lower.
I don’t think you’ve acknowledged the reasons that equivocation is so common, and the benefits it brings. There’s the obvious motte-and-bailey advantage for adversarial or partly-adversarial communication. There’s also a big advantage in generalization, as it’s often impossible to be truly literal in any reasonable amount of time.
I’m a big fan of speaking plainly and making most statements as propositional predictions, among trusted people, on topics we’re cooperating on. Even then, my tendency to generalize and equivocate gets in the way. Heck, even in my private mind, I shy away from some parts of complex questions, and let myself handwave past them.
Policy debates (at any level—household to global) are orders of magnitude more complex than simpler examples, and there’s probably no avoiding both rhetorical tactics and model compression that intermix as ambiguity, with implicature as a common way of dealing with that ambiguity.
I’m not really against rhetorical tactics in the sense of being strategic with one’s words; I’m against shortsighted rhetorical tactics which stop the speaker themselves from realizing the truth, or burn the commons by inserting lies into the public discourse, etc. Most rhetorical tactics are actually quite transparent if you take a moment to try and apply logically similar arguments to politically-different examples, etc. (I’m not saying rhetorical tactics are automatically OK if they don’t fit into that category; but I am saying that most rhetorical tactics just wouldn’t work at all in a world where they were uncommon and listeners applied ordinary truth standards to discourse rather than applying the current rhetoric-friendly rules.) Or to put it a different way: imho, all one needs to do to (mostly) eliminate (most) empty rhetoric is establish common knowledge (or close to common knowledge) about those rhetorical patterns.
(On the other hand, I agree that it’s practically impossible to eliminate all ambiguity, particularly what’s due to model compression.)
It’s an important point. Equivocation can also be framed as the heuristic inference “if A and B are highly correlated, replace A with B” (but made implicitly, without even making a distinction between A and B, because A and B have been clustered together at some previous point).
The real point is to have the flexibility to be able to make fine distinctions or big generalizations as appropriate for a situation, rather than being stuck in whichever way you think about it.
However, making distinctions seems especially powerful somehow. You want to be able to do both, and you want to actually use bot, but if it were only a matter of moving one direction or the other, it seems somehow better to move in the “distinctions” direction.
One point is that distinctions tend to be more concrete and objective, and therefore, more “portable” between brains, and more scientifically testable.
Also, making more distinctions temporarily for the sake of putting things back together later (in a hopefully better way) can be a great way of making progress. The other direction (temporary abstraction) is incredibly useful as well, but seems a bit weaker (it’s like analogy as opposed to logic).
Another point is that we have to cluster a lot to fit things into our limited working memory. So it’s more probable that by making extra distinctions, you can see things that you’d miss otherwise. On the other hand, since everyone is abstracting all the time (to handle limited working memory), the probability that you’ll hit new insights that way seems lower.