I suppose people use the term “LW ethics” to refer to Eliezer’s moral indexicalism (Is there a name for the position that has actually been adopted into more wide-spread use here?)
Subjectivism.
plus consequentialism,
The consequentialism is of the utilitarian variety , which isn’t particularly compatible with subjectivism/indexicality. So there’s two theories. There’s also the objective-sounding CEV thing, and the
deflationary-sounding tendency to talk about “morality” and “preferences” interchangeably.
Really? In that case, I’m strongly against using that, because the term “subjectivism” seems ill-defined, not very illuminating, and has a bunch of wrong connotations. In particular, as I read him, Eliezer does think that we are all talking about the same thing, that we have a shared referential intention, and that disagreement is therefor substantial and about the truth-value of a definite proposition that has a truth-value. (Which is precisely the reason why I think his theory is wrong.)
The consequentialism is of the utilitarian variety , which isn’t particularly compatible with subjectivism/indexicality. So there’s two theories.
Of course; one is a metaethical theory, the other is an ethical theory. But I think the two are very compatible, simply because they’re orthogonal.
really? In that case, I’m strongly against using that, because the term “subjectivism” seems ill-defined,
By whom? It seems well-defined by professional philosophes to me.
not very illuminating, and has a bunch of wrong connotations
To whom? It is not a force of nature that makes words have connotations. The individual brings whatever connotations they bring.
In particular, as I read him, Eliezer does think that we are all talking about the same thing, that we have a shared referential intention, and that disagreement is therefor substantial and about the truth-value of a definite proposition that has a truth-value. (Which is precisely the reason why I think his theory is wrong.)
Then what happened to the indexicality you mentioned? (And for which there is independent evidence, such as “Yudkowsky!good”)? If the truth values of ethical claims are indexed to individuals, where does the disagreement come from?
Of course; one is a metaethical theory, the other is an ethical theory. But I think the two are very compatible, simply because they’re orthogonal.
Subjectivism, by itself is a metaethical theory which yields object-level results when an individuals prefrences are plugged in.
Utilitarianism, by itself is a metaethical theory which yields object-level results when a societies preferences are plugged in.
What I want is not automatically what the greatest Number wants, so they are incompatible.
By whom? It seems well-defined by professional philosophes to me.
Well, the Stanford Encyclopedia doesn’t give a unambiguous definition of it; it says that it’s about “mind-(in)dependence”, but says that there is some disagreement about what precisely that means. Intuitively, I feel Eliezer’s view is not illuminatingly called subjectivist because for him, moral statements are simply about the properties of a certain abstract algorithm.
Interstingly, I found this piece, according to which Eliezer’s view is clearly relativist.
Then what happened to the indexicality you mentioned?
Imagine two humans standing besides each other and disagreeing about the truth of “it’s bright here”. (Assume they’re both blind and so only have indirect evidence, otherwise disagreement doesn’t really make sense or would necessarily be about the standard of brightness, which is a tricky issue in itself.) Since they’re in the same place, they are using the indexical “here” with the same reference. For Eliezer, as I understand him, the same thing happens when two humans disagree about “one should do X”: the indexical “should” has the same reference when used by either of them.
Not responding to the last section because I’m not following it. I see your point that the whole utilitarianism/CEV/preferences business—in my view, the ethical as opposed to the metaethical part—is somewhat muddled up in Eliezer’s writing, though.
According to such a view, it is possible that when John asserts “Stealing is wrong” he is saying something true, but that when Jenny asserts “Stealing is wrong” she is saying something false.
would seem to be trivially true, if they are in fact talking about sentences there. Just have Jenny be speaking a foreign language that contains the same symbols with a different meaning. By that standard, anything anyone says ever could safely be considered relative.
On the other hand if they’re talking about propositions then it would seem to be trivially false, because propositions have truth values.
They are talking about propositions, and the point is that the sentence “Stealing is wrong” expresses different propositions when uttered by different people, just like “I’m sleepy” does. That’s what indexicality is!
That’s what I meant by “talking about sentences”. Any sentence can express different propositions when uttered by different people. Just have people speaking different languages. So clearly “means different things when said by different people” isn’t half specific enough to have any dire metaethical implications.
We are speaking the same language. Yet we express different propositions when we say “I’m sitting at a table.” This is not trivial; for example, various other sentences do not have this property. So the SEP quote, once interpreted correctly, is non-trivial, too, because it is quite clear that the writer did not intend your “speaking different languages” interpretation.
Also, you can divorce a technical notion of a sentence from that of a string of sounds; sentencehood might be a two-place predicate of a string and a language.
But I could easily claim that the way strings like “me” and “here” change their meanings depending on context just shows that we do not always speak the same language. I could think of a language as just being a mapping from symbols to propositions, in which case any variation in propositions expressed means that it is not the same language.
You could argue that there is some kind of mapping to an intermediate state we have in common: symbol -> intermediate -> proposition where symbol -> intermediate is your “language”. But then I would ask why anyone should care about that particular intermediate state, and whether that intermediate state can be compellingly or uniquely defined with respect to words like “wrong”.
But I could easily claim that the way strings like “me” and “here” change their meanings depending on context just shows that we do not always speak the same language.
Then you are employing the word “language” in an idiosyncratic, confusing, and, in my opinion, not very useful way. Note that we also switch languages depending on where we are (“here”) and, in fact, continuously all the time (“now”). Good luck building a theory on that.
You might want to look up the notion of a Kaplanian character, which is precisely the intermediate level that you’re suggesting. A character is a function from contexts of utterance to propositions, and the (language-relative) meaning of a sentence is such a character, so, as you say, you could think of a language as a relation between strings and characters (not a function, because of ambiguity). In that picture, an expression is called “indexical” when its character is not a constant function.
Why we should care about that? Because it’s useful in explaining why we understand each other despite the fact that we don’t express the same propositions with one and the same sentence all the time, I suppose.
So the claim that “wrong” is indexical is certainly meaningful and non-trivial. Whether it’s correct is another matter. (I think it isn’t.)
I could think of a language as just being a mapping from symbols to propositions, in which case any variation in propositions expressed means that it is not the same language.
If that’s what counts as a language, I think we should deny the existence of languages:
“Rather than take for granite that Ace talks straight, a listener must be on guard for an occasional entre nous and me...or a long face no see. In a roustabout way, he will maneuver until he selects the ideal phrase for the situation, hitting the nail right on the thumb. The careful conversationalist might try to mix it up with him in a baffle of wits. In quest of this pinochle of success, I have often wrecked my brain for a clowning achievement, but Ace’s chickens always come home to roast. From time to time, Ace will, in a jersksome way, monotonise the conversation with witticisms too humorous to mention. It’s high noon someone beat him at his own game, but I have never done it; cross my eyes and hope to die, he always wins thumbs down.” From A Nice Derangement of Epitaphs
Yet we express different propositions when we say “I’m sitting at a table.”
Well, if that’s the way to read the SEP quote, then it does look trivially true. At least as trivial as this claim:
It is possible that when John says “It is raining today”, he is saying something true, while when Jenny says “The dog is brown” she is saying something false.
The non-trivial reading would have to be that John and Jenny are asserting the same proposition, but that Jenny’s assertion is true while John’s is false.
On the other hand if they’re talking about propositions then it would seem to be trivially false, because propositions have truth values.
It doesn’t follow from the fact that propositions have truth values that the SEP quote is trivially false. “Stealing is wrong” as asserted by John and “Stealing is wrong” as asserted by Jenny both have truth values, they just don’t have the same truth values. You need another premise.
By “if they’re talking about propositions” I meant assuming that they mean “Stealing is wrong” to refer to a particular proposition (ie. asserting that John and Jenny are stating the same proposition) rather than a sentence (string of words).
Right, so you’d need a premise like “A given proposition can only have one truth value”. That’s not trivial though, and it’s been the subject of historical debate. For example, “Hen is sitting” is true...and false....and true again. You can work around that by saying “propositions come with time indexes” or something, but that’s far from trivial. Why can’t differences in speaker have the same effect?
So the SEP quote may be trivially false under a certain understanding of ‘proposition’ but I take it the issue of how to understand propositions is part of what the SEP quote intends to raise.
“Indexed to something mind-dependent ” and “indexed to something about an individual” are not both more precise than your “indexical”
I don’t find it illuminating to be told that morality is about an abstract algorithm , since that does not tell me whether the algorithm is defined at the individual, group, or universal level; nor whether it is cognactively accessible; nor whether it is mind dependent
You have said that moral claims are indexical, and also that “should” always has the same referent .hat is it so where does the last indexicality come from...what makes it vary?
Let me take another try at the last section: ethics and metaethics aren’t orthogonal . Not all combinations work. As object level ethics, utilitarianism is incompatible with metaethics that is indexed to or relative to individuals.
I’m using “utilitarianism” in the standard philosophical sense, not in the LessWrong “uses utility functions” sense. As such, it already has a utility function—the utility functions of everyone that exists.
Subjectivism.
The consequentialism is of the utilitarian variety , which isn’t particularly compatible with subjectivism/indexicality. So there’s two theories. There’s also the objective-sounding CEV thing, and the deflationary-sounding tendency to talk about “morality” and “preferences” interchangeably.
Really? In that case, I’m strongly against using that, because the term “subjectivism” seems ill-defined, not very illuminating, and has a bunch of wrong connotations. In particular, as I read him, Eliezer does think that we are all talking about the same thing, that we have a shared referential intention, and that disagreement is therefor substantial and about the truth-value of a definite proposition that has a truth-value. (Which is precisely the reason why I think his theory is wrong.)
Of course; one is a metaethical theory, the other is an ethical theory. But I think the two are very compatible, simply because they’re orthogonal.
By whom? It seems well-defined by professional philosophes to me.
To whom? It is not a force of nature that makes words have connotations. The individual brings whatever connotations they bring.
Then what happened to the indexicality you mentioned? (And for which there is independent evidence, such as “Yudkowsky!good”)? If the truth values of ethical claims are indexed to individuals, where does the disagreement come from?
Subjectivism, by itself is a metaethical theory which yields object-level results when an individuals prefrences are plugged in.
Utilitarianism, by itself is a metaethical theory which yields object-level results when a societies preferences are plugged in.
What I want is not automatically what the greatest Number wants, so they are incompatible.
Well, the Stanford Encyclopedia doesn’t give a unambiguous definition of it; it says that it’s about “mind-(in)dependence”, but says that there is some disagreement about what precisely that means. Intuitively, I feel Eliezer’s view is not illuminatingly called subjectivist because for him, moral statements are simply about the properties of a certain abstract algorithm.
Interstingly, I found this piece, according to which Eliezer’s view is clearly relativist.
Imagine two humans standing besides each other and disagreeing about the truth of “it’s bright here”. (Assume they’re both blind and so only have indirect evidence, otherwise disagreement doesn’t really make sense or would necessarily be about the standard of brightness, which is a tricky issue in itself.) Since they’re in the same place, they are using the indexical “here” with the same reference. For Eliezer, as I understand him, the same thing happens when two humans disagree about “one should do X”: the indexical “should” has the same reference when used by either of them.
Not responding to the last section because I’m not following it. I see your point that the whole utilitarianism/CEV/preferences business—in my view, the ethical as opposed to the metaethical part—is somewhat muddled up in Eliezer’s writing, though.
That SEP page is curious, because this:
would seem to be trivially true, if they are in fact talking about sentences there. Just have Jenny be speaking a foreign language that contains the same symbols with a different meaning. By that standard, anything anyone says ever could safely be considered relative.
On the other hand if they’re talking about propositions then it would seem to be trivially false, because propositions have truth values.
They are talking about propositions, and the point is that the sentence “Stealing is wrong” expresses different propositions when uttered by different people, just like “I’m sleepy” does. That’s what indexicality is!
That’s what I meant by “talking about sentences”. Any sentence can express different propositions when uttered by different people. Just have people speaking different languages. So clearly “means different things when said by different people” isn’t half specific enough to have any dire metaethical implications.
We are speaking the same language. Yet we express different propositions when we say “I’m sitting at a table.” This is not trivial; for example, various other sentences do not have this property. So the SEP quote, once interpreted correctly, is non-trivial, too, because it is quite clear that the writer did not intend your “speaking different languages” interpretation.
Also, you can divorce a technical notion of a sentence from that of a string of sounds; sentencehood might be a two-place predicate of a string and a language.
But I could easily claim that the way strings like “me” and “here” change their meanings depending on context just shows that we do not always speak the same language. I could think of a language as just being a mapping from symbols to propositions, in which case any variation in propositions expressed means that it is not the same language.
You could argue that there is some kind of mapping to an intermediate state we have in common:
symbol -> intermediate -> propositionwheresymbol -> intermediateis your “language”. But then I would ask why anyone should care about that particular intermediate state, and whether that intermediate state can be compellingly or uniquely defined with respect to words like “wrong”.Then you are employing the word “language” in an idiosyncratic, confusing, and, in my opinion, not very useful way. Note that we also switch languages depending on where we are (“here”) and, in fact, continuously all the time (“now”). Good luck building a theory on that.
You might want to look up the notion of a Kaplanian character, which is precisely the intermediate level that you’re suggesting. A character is a function from contexts of utterance to propositions, and the (language-relative) meaning of a sentence is such a character, so, as you say, you could think of a language as a relation between strings and characters (not a function, because of ambiguity). In that picture, an expression is called “indexical” when its character is not a constant function.
Why we should care about that? Because it’s useful in explaining why we understand each other despite the fact that we don’t express the same propositions with one and the same sentence all the time, I suppose.
So the claim that “wrong” is indexical is certainly meaningful and non-trivial. Whether it’s correct is another matter. (I think it isn’t.)
If that’s what counts as a language, I think we should deny the existence of languages:
“Rather than take for granite that Ace talks straight, a listener must be on guard for an occasional entre nous and me...or a long face no see. In a roustabout way, he will maneuver until he selects the ideal phrase for the situation, hitting the nail right on the thumb. The careful conversationalist might try to mix it up with him in a baffle of wits. In quest of this pinochle of success, I have often wrecked my brain for a clowning achievement, but Ace’s chickens always come home to roast. From time to time, Ace will, in a jersksome way, monotonise the conversation with witticisms too humorous to mention. It’s high noon someone beat him at his own game, but I have never done it; cross my eyes and hope to die, he always wins thumbs down.” From A Nice Derangement of Epitaphs
Well, if that’s the way to read the SEP quote, then it does look trivially true. At least as trivial as this claim:
The non-trivial reading would have to be that John and Jenny are asserting the same proposition, but that Jenny’s assertion is true while John’s is false.
It doesn’t follow from the fact that propositions have truth values that the SEP quote is trivially false. “Stealing is wrong” as asserted by John and “Stealing is wrong” as asserted by Jenny both have truth values, they just don’t have the same truth values. You need another premise.
By “if they’re talking about propositions” I meant assuming that they mean “Stealing is wrong” to refer to a particular proposition (ie. asserting that John and Jenny are stating the same proposition) rather than a sentence (string of words).
Right, so you’d need a premise like “A given proposition can only have one truth value”. That’s not trivial though, and it’s been the subject of historical debate. For example, “Hen is sitting” is true...and false....and true again. You can work around that by saying “propositions come with time indexes” or something, but that’s far from trivial. Why can’t differences in speaker have the same effect?
So the SEP quote may be trivially false under a certain understanding of ‘proposition’ but I take it the issue of how to understand propositions is part of what the SEP quote intends to raise.
It’s not trivially false because philosophers take a sentence that appears to be in certain language to have its normal meaning in that language.
It’s not trivially true because identical sentences can have different truth values when spoken by different people, eg “my name is John”
I guess the idea is that assertions about morality are not propositions in that sense?
“Indexed to something mind-dependent ” and “indexed to something about an individual” are not both more precise than your “indexical”
I don’t find it illuminating to be told that morality is about an abstract algorithm , since that does not tell me whether the algorithm is defined at the individual, group, or universal level; nor whether it is cognactively accessible; nor whether it is mind dependent
You have said that moral claims are indexical, and also that “should” always has the same referent .hat is it so where does the last indexicality come from...what makes it vary?
Let me take another try at the last section: ethics and metaethics aren’t orthogonal . Not all combinations work. As object level ethics, utilitarianism is incompatible with metaethics that is indexed to or relative to individuals.
Utilitarianism is an ethical theory, not a metaethical theory.
Utilitarianism with a particular utility function is an ethical theory.
I’m using “utilitarianism” in the standard philosophical sense, not in the LessWrong “uses utility functions” sense. As such, it already has a utility function—the utility functions of everyone that exists.