= 27d567bc2d48d9f40e9ccc20ea370b15)
nshepperd comments on Diseased thinking: dissolving questions about disease - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (343)
Look, I have no problem with basing ethics on moral intuitions, and what we actually want. References to right and wrong are after all stored only in our heads.
But in the specific context of a discussion of the Categorical Imperative—which is supposed to be a principle forbidding "categorically" certain decisions—there needs to be some rule explaining what "universalizable" actions are not permitted, for the CI to make meaningful prescriptions. If you simply decide what actions are permitted based on whether you (intuitively) approve of the outcome, then the Imperative is doing no real work whatsoever.
If, like most people, you don't want to be murdered, the CI will tell you not to murder. If you don't want to be robbed, it will tell you not to rob. Etc. It does work for the normal majority, and the abnornmal minority are probably going to be a problem under any system.
Please read the above thread and understand the problem before replying.
But for your benefit, I'll repeat it: explain to me, in step-by-step reasoning, how the categorical imperative forbids me from taking the action "if (I am nshepperd) then rob else do nothing". It certainly seems like it would be very favourable to me if everyone did "if (I am nshepperd) then rob else do nothing".
That's a blatant cheat. How can you have a universal law that includes a specific exception for a named individual?
The way nshepperd just described. It is, after all, a universal law, applied in every situation. It just returns different results for a specific individual. We can call a situation-sensitive law like this a piecewise law.
Most people would probably not want to live in a society with a universal law not to steal unless you are a particular person, if they didn't know in advance whether or not the person would be them, so it's a law one is unlikely to support from behind a veil of ignorance.
However, some piecewise laws do better behind veils of ignorance than non-piecewise universal laws. For instance, laws which distinguish our treatment of introverts from extroverts stand to outperform ones which treat both according to the same standard.
You can rescue non piecewise categorical imperatives by raising them to a higher level of abstraction, but in order to keep them from being outperformed by piecewise imperatives, you need levels of abstraction higher than, for example "Don't steal." At a sufficient level of abstraction, categorical imperatives stop being actionable guides, and become something more like descriptions of our fundamental values.
I'm all in favour of going to higher levels of abstraction. Its much better appreach than coding in kittens-are-nice and slugs-are-nasty.
Is there anything that makes it qualitatively different from
if (subject == A) { return X }
elsif (subject==B) { return Y }
elsif (subject==C) { return Z } ... etc. etc.?
No, there isn't any real difference from that, which is why the example demonstrates a flaw in the Categorical Imperative. Any non-universal law can be expressed as a universal law. "The law is 'you can rob', but the law should only be applied to Jiro" is a non-universal law, but "The law is 'if (I am Jiro) then rob else do nothing' and this law is applied to everyone" is a universal law that has the same effect. Because of this ability to express one in terms of the other, saying "you should only do things if you would like for them to be universally applied" fails to provide any constraints at all, and is useless.
Of course, most people don't consider such universal laws to be universal laws, but on the other hand I'm not convinced that they are consistent when they say so--for instance "if (I am convicted of robbery) then put me in jail else nothing" is a law that is of similar form but which most people would consider a legitimate universalizable law.
If the law gives different results for different people doing the same thing, it isn't universal jn the intended sense, which is pretty much the .same as fairness.
"In the intended sense" is not a useful description compared to actually writing down a description. It also may not necessarily even be consistent.
Furthermore, it's clear that most people consider "if (I am convicted of robbery) then put me in jail else nothing" to be a universal law in the intended sense, yet that gives different results for different people (one result for robbers, another result for non-robbers) doing the same thing (nothing, in either case).
I don't think there is, but then, I don't think that classifying things as universal law or not is usually very useful in terms of moral guidelines anyway. I consider the Categorical Imperative to be a failed model.
Why is it failed? A counterexample was put forward that isn't a universal law. That doesn't prove the .CI to be wrong. So what does?
We already adjust rules by reference classes, since we have different rules for minors and the insane. Maybe we just need rules that are apt to the reference class and impartial within it.
If we have different rules for minors and the insane, why can't we have different rules for Jiro? "Jiro" is certainly as good a reference class as "minors".
Remember the "apt". You would need to explain why you need those particular rules.
When you raise it to high enough levels of abstraction that the Categorical Imperative stops giving worse advice than other models behind a veil of ignorance, it effectively stops giving advice at all due to being too abstract to apply to any particular situation with human intelligence.
You can fragment the Categorical Imperative into vast numbers of different reference classes, but when you do it enough to make it ideally favorable from behind a veil of ignorance, you've essentially defeated any purpose of treating actions as if they were generalizable to universal law.
I'd lovely know the meta model you are using to judge between models.
Universal isn't really universal, since you can't prove mathematial theorem to stones.
Fairness within a reference class counts.
A qualitative difference is a quantitative difference that is large enough.
Sometimes. Not always.