Positive reinforcement is more effective than positive punishment because it tells someone exactly what to do instead of just what not to do.
But being told exactly what to do is much more restrictive than just being told what not to do. So if you make the above claim, and also make the claim that your way is more pleasant and less restrictive, then Bob is right—there is definitely some trick being played here.
Retracted my previous response due to faulty reasoning.
Yes, positive reinforcement gives a more precise signal than positive punishment. Maybe I didn’t elaborate on this fact enough, but that is the point. I like it when a community precisely signals what its values are. I like it when a community isn’t adversarial to bystanders. These are, among other reasons, precisely why I want a positive reinforcement culture. The point of the post is that these two frames are different. You pointed out a difference in the post that I discussed.
Positive reinforcement, however, is not more restrictive. There is no restriction going on, that’s the point. It is just a more precise signal. If I give you a cookie every time you do the dishes and nothing when you don’t, that is not restricting the counterfactual-yous who don’t do dishes from being in my community. Somebody else just does the dishes and gets the cookie.
I don’t think that you’re responding to what I actually wrote… but perhaps I was unclear. Let me try again, with emphasis on the relevant part:
Positive reinforcement is more effective than positive punishment because it tells someone exactly what to do instead of just what not to do.
The bolded part is what I am addressing.
Being told exactly what to do is more restrictive than just being told what not to do. That is my claim. This claim is not dependent on any purported connection between “being told what to do” and “positive reinforcement”, or between “being told what not to do” and “positive punishment”.
Positive reinforcement, however, is not more restrictive. There is no restriction going on, that’s the point. It is just a more precise signal.
Ethics is the means by which we answer the question “what is the right thing to do”. In ethics, there is no difference between rewarding good things and punishing bad things; these are isomorphic approaches.
That means that only rewarding one specific thing is isomorphic to punishing all things except that thing; and only punishing one thing is isomorphic to rewarding all things except that thing.
(This is analogous to the distinction between permissive vs. restrictive legal systems—or, as in the old joke, the difference between “everything is permitted, except that which is forbidden” and “everything is forbidden, except that which is permitted”.)
Now, you object that you’re not punishing anyone, that all of this stuff is strictly optional. This is the concept of supererogation. There are two problems with this:
First, most rationalists/EAs/whatever are utilitarians, and utilitarianism, famously, has no concept of supererogation. There cannot possibly be any such thing as “good but not mandatory” for a utilitarian. If Bob knows this, then he can see that utilitarians who tell him that he doesn’t have to to do the most good if he doesn’t want to are either hypocrites or liars (or they don’t understand the moral system they claim to espouse).
Second, what Bob in your fictional scenario can probably see, and what most people in real life can also see, is that if status in your community is accorded to those who take all of the so-called “moral opportunities”, and is not accorded to those who turn down those “moral opportunities”, then they’re not “opportunities” at all—they’re obligations.
In other words, Bob asks: “what must I do in order for your community to see me as a good person?”. Possible answers include:
“We don’t think of people as good or bad at all.” (Implausible; and if actually true—very weird. Approximately nobody will believe such a claim, and they’ll be right not to.)
“You’re a good person as long as you don’t do X.” (Highly permissive.)
“You’re a good person if, and only if, you do Y.” (Highly restrictive.)
What Bob hears when you talk to him about “moral opportunities” is that your answer is #3, but you don’t want to admit that your answer is #3, because you feel (correctly) that it’s restrictive and judgmental. Bob suspects a trick because there is a trick.
I mean, it’s right there in the terminology you’re using: “moral opportunities”. Moral opportunities. There isn’t any such thing as moral opportunities! There are moral obligations, and then there’s everything else. If you hear people resisting this sort of terminological sleight-of-hand, it’s because they recognize that calling a thing by a different name does not actually change the patterns of status assignment and other social dynamics related to that thing.
Talking about “moral opportunities” is like talking about the opportunity to pass a class by doing the assigned homework, or the opportunity to keep getting your paycheck by doing what your boss tells you to do, or the opportunity to stay out of prison by paying your taxes on time.
The only way to get out of this would be to actually, genuinely not consider someone who does whatever-it-is (helping to save the world from AI / saving the animals / etc.) to be a better person than someone who does. (In the way that we generally don’t consider someone who makes a lot of money to be a better person than someone who makes less money. If you have a chance to make a bunch of money—legally, without doing anything unethical, etc.—we may cheer you on, we may be happy for you if you succeed, but we won’t think less of you if you choose not to do that, right? That chance is an opportunity—the regular, non-moral kind of opportunity.)
But if your community is built around the notion that doing whatever-it-is is important, valuable, and morally the right thing to do, then of course you will not ever stop considering someone who does whatever-it-is to be a better person than someone who does not. (And why should you?) Newcomers will correctly perceive that you think this. And if you try to convince them otherwise, they will correctly detect a trick.
No, that’s not quite what I’m saying in my post. I’m pointing at the times when a culture could either positively reinforce X or positively punish not-X, which is exactly the same amount of restriction.
Maybe, one could argue, in practice positive reinforcement creates a more precise signal, which on my model is a good thing; I like communities that clearly signal their values and aren’t adversarial to bystanders. I really don’t think this is a trick, I think the post is in fact about the differences between these two framings, including about the reinforcement/punishment distinction.
But being told exactly what to do is much more restrictive than just being told what not to do. So if you make the above claim, and also make the claim that your way is more pleasant and less restrictive, then Bob is right—there is definitely some trick being played here.
Retracted my previous response due to faulty reasoning.
Yes, positive reinforcement gives a more precise signal than positive punishment. Maybe I didn’t elaborate on this fact enough, but that is the point. I like it when a community precisely signals what its values are. I like it when a community isn’t adversarial to bystanders. These are, among other reasons, precisely why I want a positive reinforcement culture. The point of the post is that these two frames are different. You pointed out a difference in the post that I discussed.
Positive reinforcement, however, is not more restrictive. There is no restriction going on, that’s the point. It is just a more precise signal. If I give you a cookie every time you do the dishes and nothing when you don’t, that is not restricting the counterfactual-yous who don’t do dishes from being in my community. Somebody else just does the dishes and gets the cookie.
I don’t think that you’re responding to what I actually wrote… but perhaps I was unclear. Let me try again, with emphasis on the relevant part:
The bolded part is what I am addressing.
Being told exactly what to do is more restrictive than just being told what not to do. That is my claim. This claim is not dependent on any purported connection between “being told what to do” and “positive reinforcement”, or between “being told what not to do” and “positive punishment”.
Ethics is the means by which we answer the question “what is the right thing to do”. In ethics, there is no difference between rewarding good things and punishing bad things; these are isomorphic approaches.
That means that only rewarding one specific thing is isomorphic to punishing all things except that thing; and only punishing one thing is isomorphic to rewarding all things except that thing.
(This is analogous to the distinction between permissive vs. restrictive legal systems—or, as in the old joke, the difference between “everything is permitted, except that which is forbidden” and “everything is forbidden, except that which is permitted”.)
Now, you object that you’re not punishing anyone, that all of this stuff is strictly optional. This is the concept of supererogation. There are two problems with this:
First, most rationalists/EAs/whatever are utilitarians, and utilitarianism, famously, has no concept of supererogation. There cannot possibly be any such thing as “good but not mandatory” for a utilitarian. If Bob knows this, then he can see that utilitarians who tell him that he doesn’t have to to do the most good if he doesn’t want to are either hypocrites or liars (or they don’t understand the moral system they claim to espouse).
Second, what Bob in your fictional scenario can probably see, and what most people in real life can also see, is that if status in your community is accorded to those who take all of the so-called “moral opportunities”, and is not accorded to those who turn down those “moral opportunities”, then they’re not “opportunities” at all—they’re obligations.
In other words, Bob asks: “what must I do in order for your community to see me as a good person?”. Possible answers include:
“We don’t think of people as good or bad at all.” (Implausible; and if actually true—very weird. Approximately nobody will believe such a claim, and they’ll be right not to.)
“You’re a good person as long as you don’t do X.” (Highly permissive.)
“You’re a good person if, and only if, you do Y.” (Highly restrictive.)
What Bob hears when you talk to him about “moral opportunities” is that your answer is #3, but you don’t want to admit that your answer is #3, because you feel (correctly) that it’s restrictive and judgmental. Bob suspects a trick because there is a trick.
I mean, it’s right there in the terminology you’re using: “moral opportunities”. Moral opportunities. There isn’t any such thing as moral opportunities! There are moral obligations, and then there’s everything else. If you hear people resisting this sort of terminological sleight-of-hand, it’s because they recognize that calling a thing by a different name does not actually change the patterns of status assignment and other social dynamics related to that thing.
Talking about “moral opportunities” is like talking about the opportunity to pass a class by doing the assigned homework, or the opportunity to keep getting your paycheck by doing what your boss tells you to do, or the opportunity to stay out of prison by paying your taxes on time.
The only way to get out of this would be to actually, genuinely not consider someone who does whatever-it-is (helping to save the world from AI / saving the animals / etc.) to be a better person than someone who does. (In the way that we generally don’t consider someone who makes a lot of money to be a better person than someone who makes less money. If you have a chance to make a bunch of money—legally, without doing anything unethical, etc.—we may cheer you on, we may be happy for you if you succeed, but we won’t think less of you if you choose not to do that, right? That chance is an opportunity—the regular, non-moral kind of opportunity.)
But if your community is built around the notion that doing whatever-it-is is important, valuable, and morally the right thing to do, then of course you will not ever stop considering someone who does whatever-it-is to be a better person than someone who does not. (And why should you?) Newcomers will correctly perceive that you think this. And if you try to convince them otherwise, they will correctly detect a trick.
No, that’s not quite what I’m saying in my post. I’m pointing at the times when a culture could either positively reinforce X or positively punish not-X, which is exactly the same amount of restriction.
Maybe, one could argue, in practice positive reinforcement creates a more precise signal, which on my model is a good thing; I like communities that clearly signal their values and aren’t adversarial to bystanders. I really don’t think this is a trick, I think the post is in fact about the differences between these two framings, including about the reinforcement/punishment distinction.