I don’t think that’s quite how it works. I’d guess that a court would consider the substance of the law and not just whether it is phrased as a “tax”. I think legislation like this would have a substantial risk of being declared unconsitutional on takings clause grounds.
TFD
Karma: 342
I’m not arguing against deontological bars? I’m just saying that common deontological bars often have elements of intent inherent in what is acutally barred. So in the murder example, I agree that there should be and many people follow a deontological bar against murder, but that murder (as it is often defined and is generally thought of colloquially) requires intentional killing. Its not a question of whether the line is hard or soft, its a question of what is actually barred. Do you agree that when people speak of a deontological bar against killing, they really mean something more like what I have called “murder”?
I think this example actually demonstrates why most deontological bars do in fact depend on intent. I agree “thou shalt not kill” is commonly used because of how snappy it is, but I don’t think people realistically treat it as a literal deontological bar. For example, people do things that have some probabilistic chance of causing the death of others all the time (driving, utilizing products that cause harmful exposures, etc.), yet people don’t usually act as if this is deontologically barred. Presumably they would argue “that’s not really killing”, and in a sense I agree. When people think about the deontological prohibition against killing, they usually mean something closer to thou shalt not murder, where “murder” means intentional unjustified (i.e. not in proportional self-defense) killing. In order for the deontological bar to really have meaning, we need to import some notion of intent, or at least some additional context/assumptions, exactly as is the case for the examples in my post. In this sense, “thou shalt not kill” is actually extremely unclear, but in a way that is hard to recognize sometimes. I think the view I offer actually substantially improves clarity, and this clarity is a leading reason why people should utilize this approach more often.
I’m not sure I entirely understand your point, but I think its worth noting that properly understood, this doesn’t have to mean that people follow “different rules” just that the one rule references the actors beliefs or intent. For example, if I have a rule that people who come over to my place shouldn’t intentionally destroy my things, I can acknowledge that someone who intentionally take a glass and smashes it violates the rule, while someone who accidentally drops a glass does not. Many commonly advocated rules also are some version of “don’t intentionally say false things”. Two people can say the exact same statement, one knowing its falsity, the other believing it to be true. The first violates the rule, while the other does not.
Beliefs and values that move one sufficiently far from endorsing a rule overall should then be thought of as first pushing the person out of some coalition of rule-followers
Can you explain this more concretely? What would be an example of a belief or value that you have in mind here? Do you think this applies to either of the AI safety examples referenced in my post?
Deontological bars should reference the actor’s beliefs
I disagree. As I understand the situation here, the person in question genuinely believes in this incorrect meaning. If they use it in a way that they believe is correct, this is not intentional deception. I think that just follows directly from what “intentional” means, its a property of the agent in question, not of their social context. On the other hand, people in the group who are teaching the wrong meaning could definitely be acting in bad faith, and within a legal context if the group is some type of legal entity (like a corporation) that entity could be acting in bad faith.
As I mentioned on another comment, not all bad conduct in discourse is bad faith. I think part of the why this confusion exists is due to the temptation to apply a label for bad conduct to other conduct that doesn’t strictly meet the definition but is in some way analagous. I generally think it is good to resist this urge. It is good to maintain distinction and clarity about what is being claimed, and doing so doesn’t mean we have to endorse bad conduct that doesn’t fall under “bad faith”.
I think its worth noting that not all bad things you can do in an argument are bad faith. With that in mind, your first example I think straightforwardly isn’t bad faith, its just a bad thing to do. Bad faith needs intentional deception. For the second one, I think it meets what I call the “general meaning” of bad faith in the OP, so I’m not opposed necesarily to people calling that bad faith. I do think there is a more specific meaning that I note under “good faith discourse” that covers a specific thing that gets called “bad faith” very often but I think the general meaning and this more specific meaning kind of blend together and I agree it reasonable to all something that meets the general meaning “bad faith”. I just think the more specific meaning really hits upon the “vibe” so to speak that is often being refered to when people use “bad faith” vs just deception or dishonesty.
Correct, but they would have to actually make such a representation somehow, not just do something where other people think you should be willing to change your mind if you do it. So if you say “I am willing to change my mind” but you actually aren’t, that is bad faith. If you participate in an intellectual discussion even though you aren’t willing to change your mind, that isn’t sufficient in my view. Basically, debating while not being willing to change your mind isn’t sufficient to be a bad faith debater. Similarly, if you come into a negotiation and say “I’m not willing to move at all, its my way or the highway” and that is in fact true, you haven’t negotiated in bad faith. It would only be bad faith if you claim to be willing to compromise but actually aren’t.
I agree that the natural solution seems to be having a new term/phrase so that both can be pointed to while maintaining clarity.
Thanks for the clarification! I confess that this is how read your post, and part of my desire here was to get comments like this. I kind of expected that when I wrote a comment on Habryka’s post but I figured I should write the OP when my comment didn’t quite bring that out. Not sure if Habryka would also be of the same mind or if he feels that I’m wrong that this is the “standard” usage.
On the question of whether the standard concept embads a confusion, I don’t think it does but I get why you do. I think part of the problem is that “assume good faith” is subject to multiple interpretations. One can read it as rounding off to saying that you should assume people are generally good/truthful/rational, which I think is what you are getting at. I agree that this is wrong as is a danger of people leaning into “assume good faith” too heavily. On the other hand, I think there is a meaning of “assume good faith” that is more like “have a rebutable presumption of good faith analgous to the presumption of innocence”, that is in fact helpful for similar reasons as Richard Ngo mentions in the comments on that post.
I am generally of the view that following this “rebutable presumption of good faith” idea/norm actually is one of the primary things that helps actually implement your “stick to the object level” advice (which I agree is very good advice). The thing that throws people out of an equlibrium at the object level is often accusations of some type of bad conduct during the discussion, bad faith being a common example. Having a presumption of good faith can create a dynamic that discourages this, and in particulardiscourages the common occurrence where one person making a kind-of accusation quickly escalates as the other person joins in. I give my version of “stick to the object level” type advice here.
Also, I’m not clear Habryka’s definition is incompatible with the one you’re offering. Serious discussions, almost by definition, do have the pretense of being about mutual truth seeking. If serious discussions usually aren’t actually about that, then they’re not in good faith by your definition as well as Habryka’s.
I think in order to be an intentional misrepresentation, this would have to be a case where someone literally says “I am engaging in this discussion primarily for purposes of truth-seeking” when that isn’t true, to qualify under my definition. I agree that would then be in bad faith. On the other hand, I don’t think someone seriously discussing a topic when their primary motivation is something like persuading other or increasing their own status would qualify under my definition. I don’t think that is intentional misrepresentation. Can you explain more why you think that would qualify perhaps?
If there is a commonly understood meaning of the phrase “good faith,“ it’s nowhere near as clear as the distinction between a faux pas and a broken law.
My example might be a bit hyperbolic but I don’t think its that far off. My primary difference with other posts on this is the idea that there needs to be intentional (as opposed to accidental etc.) deception. Some of the definitions I quote and all the Wiki examples make the need for intentionality pretty clear in my view. The faux pas vs law analogy is more obvious/uncontroversial which is why I use it, in the hopes that that helps people understand my own view. For clarity, I’m not saying that my position is as obviously correct as that analogy may make it seem, just that from my point of view it isn’t really a close call on the intent question.
“Bad faith” means intentionally misrepresenting your beliefs
here meaning “conversation in which the primary goal of all participants is to help other participants and onlookers arrive at true beliefs”
This doesn’t seem to be the normal meaning of “good faith” to me. This post also mis-defines the concept in my view, so it seems it might be a common misunderstanding on lesswrong.
My understanding is “good faith” doesn’t require a person to be primarily motivated by truth, it simply requires that they aren’t misrepresenting their motivation or beliefs.
A classic concept that demonstrates why this definition is incorrect is the concept of “negotiating in good faith”. This doesn’t mean that the parties aren’t motivated by their own profit. A buyer and seller can negotiate a contract in good faith even while they both seek primarily to maximize their own benefit from the contract. The “good faith” part simply means they won’t take this maximization to the extreme degree where they misrepresent their willingness to abide my the terms (e.g. a seller who agrees to provide a certain amount of goods knowning they can’t fufill the contract, but who thinks they can force the buyer to pay once they have them over a barrel on delivery).
Do you view this as the normal/standard meaning of the phrase?
Will AI make everything more correlated?
I only addressed the defund the police issue because it is used as an example in the OP, I’m not trying to describe critics of the slogan in general. “Inappropriate use of strategy”, is my way of describing the view set forth in the OP, because I think it relates to people “speaking strategically” vs “just saying what they think”.
Apologies, I tried to make this clear: I am referring to “high-dimensional discourse chess” that requires asserting or assuming “we can model how public acceptability shifts and cleverly intervene to steer those shifts.” That’s not about communicating an idea, it’s about the goal of convincing people of something in order to have them react in order to change the pubic acceptability of another thing.
No need to apologize, the point of having discussions is th hash these things out, right? And any error or misunderstand may also me mine rather than yours.
The thing that I find confusing is I feel like I get the vibe that is meant, but I don’t understand what the brightline or criteria are for when something is acceptable vs violates some norm. “High-dimensional discourse chess” to me kind of leans into this rather than clarifying the issue. I get that sense that people mean to address discursive strategies that in some way misleading, bad faith, “not truth-seeking”, or similar “vibes”. But in my view, whether these aspects apply to a person’s statements or a group’s rhetoric are likely to be contested, and the disagreement about those meta-questions are likely to have a high degree of overlap with the underlying object-level disagreement. I am worried that this causes a dynamic where people who disagree on the undelying issue often get into meta-debates about who is doing bad discource stuff, and that this is often unproductive.
It seems you are saying something similar; you dislike the current system, and say it should be replaced instead of reformed, but don’t have a clear argument for the details.
I strongly disagree with proponents of the “defund the police” slogan. Literally nothing I said has anything to do with my views on policing. I brought it up purely to address it in the context of your OP, as an example of the phenomenon under discussion. My personal interest in this has nothing to do with the defund the police example itself, and I think this goes to my concern about how meta-discussions often tend to meander to unproductive topics.
That said, I’m curious what you think that supporters of the slogan would feel about your argument. My suspicion is that they wouldn’t think that the issue is as clean cut as you describe. I will avoid trying to describe what arguments they might make since I don’t want violate my own advice and get into the weeds on a topic that I view as a sideshow, but hopefully you can see my point here. When people have strong disagreements, accusations of meta-level bad behavior that seems obvious to a person on one side of the underlying issue may not seem as obvious to someone on the other side.
And I wasn’t guessing about EA either; I have been in the room, repeatedly, when senior people in EA talked about shifting the Overton window on AI risk. So yes, don’t accuse others of doing this, but that doesn’t mean you can’t call them out when they say they are doing it!
Part of my interest in these discussions is that I have zero interactions with anyone in the EA/Rationalist spheres other than reading stuff online and in more recent years doing my own posting. I don’t have any inside knowledge, but I find statements like this very interesting and informative because I would like to understand what is going on to the extent I can. I find this difficult however, because I often find myself in the situation where when the statements being evaluated are public, I feel like people sometimes are overreading them. My read of many public statements that attract allegations of inappropriate discourse game playing is that they often appear to be downstream of disagreements on the object-level issues.
So yes, don’t accuse others of doing this, but that doesn’t mean you can’t call them out when they say they are doing it!
I’m not really sure how to take your statement as not being an accusation of EAs doing the thing you are criticizing. Are you saying that pronenent EAs would read what you have said here and publicly say “yes, we are doing this”? If not, I don’t think it really makes sense to say you are just calling out what they say they are doing. I understand you have non-public information here and of course you would use all the information you have access to to inform your own views. The thing that is difficult for me, as someone working off of only public info, is how to come to my own understanding of what is going on. If these people aren’t openly saying what they are doing, I can’t rely on “but they are saying it!” as something that shows that the behavior in question is over the line.
Edit: I appear to be rate-limited, so I will add my clarifications here (hopefully that isn’t breaking any rules, appologies if so), and then comment once the limit is up (this and all below was added after the subsequent comment).
I don’t think there is a bright line, there’s just a point being made about a gradient where discourse chess is on one side, and talking about object level facts in on the other. And I pointed out that on the chess side, people suck at getting what they want.
Perhaps I misunderstood you then. My perspective is something like this:
Speed of a car on a highway is a gradient, but if you are caught speeding, for practical purposes we might break down that gradient into buckets. If you are going 61 mph in a 60 zone, I have zero problem. If you are going 200 mph in a 60 zone, I have zero problem with you doing time in prison. Somewhere in there is a “messy middle” where people might reasonably disagree, but the disagreement is within acceptable parameters. Maybe I try to stay with 10 mpg of the speed limit but for you its 15. We disagree, but its a prison vs no-prison magnitude of disagreement. Even if think you should stay below 70 I don’t think 71 deserves jail time.
The impression I got from your post, which may be mistaken, is that you think people who behave in a way that is too strategic are doing something that is bad or wrong. It seemed to me like you thought that “crossing a line” in this regard is or should be a norm-violation. I’m of the view that while in any given case what a person does along these lines might not be ideal, it shouldn’t be considered a norm violation. I think it is often very messy to distinquish strategic/gameplaying/intended-overton-shifting speech from “just saying what you think”. You could write something that to you feels like you are just putting your best understanding of the truth out there, but which someone else feels like is being too strategic. I think when people speech, they should avoid the “norm-violating” versions of strategic speech but shouldn’t focus too much on whether or not their speech is strategic or not.
So I agree that imputing this type of behavior, as an accusation, is worrying, but that’s different than pointing out when the behavior was in fact intentional.
For clarity, when I say it seems like you’re making an accusation, I’m not saying this accusation is wrong. If a person says “I saw so-and-so commit this crime” that is accusing that person of a crime. Its not bad to correctly accuse people of things! My issues is I feel like its often unclear what the accusation is, or (like I in this case I think?) if there even is an accusation. To me it seemed like you were saying that some EAs did an intentially bad thing, and it was bad of them to do it, I’m trying to understand if that is correct or not basically (and if you are saying they did something bad, what that thing is).
I’m somewhat confused by what people mean by “strategic” in these discussions. It seems to me like there are aspects of communication that I would call “stategic” but are uncontroversial. I would suggest the “grown not crafted” idea from IABIED. I think this is a extremely succinct and effective way to communicate the underlying technical idea, but of course expressing the idea in more technical language would also be truthful (arguably even more so). The difference isn’t that the “grown not crafted” way of communicating the idea is more truthful, but (I would infer) that it is predicted ot be more helpful in allowing others to understand the idea, even if they aren’t already famiiar with relevant technical knowledge. In other words, it is a more strategic way of communicating the idea.
I don’t think that example is particularly controversial, but I think these discussions are meant to also apply to statements that are highly contentious and subject to adversarial dynamics. When we consider statements that are more in those domains, I think we enter a kind of “messy middle” where the truthfulness of a statement is much more contested and it isn’t necessarily as easy to seperate the core content from the communication strategy.
Let’s take the “DEFUND THE POLICE” example from the post. I definitely believe that some people engage in the “window-stretching” as you describe. But I could also imagine proponents of the slogan saying something like this:
“I genuinely believe that current policing practices are so unjust that the police deserve to be defunded. It would be better for policing to be reformed to correct these injustices while still carrying out legitimate policing functions. But saying I want to “defund the police” is genuinely true under the current system. If I only advocated for more centrist reforms that would actually be misleading about what I believe and trying to fit my beliefs into the overton window, not the other way around. Its extremely important to use slogans like “DEFUND THE POLICE” when you actually believe in them because the powerful simply ignore centrist calls for reform. They need to know how strongly we actually feel about this. Calling for measures such as defunding is the only way to have our voices heard”.
I think such a view lives in this messy middle. People who agree are likely to see it as genuine while critics will be tempted to claim inappropriate use of stategy. The issue of whether the person who holds this view is being “truthful” essentially collapses back into the underlying object-level issues. Those who agree on the object-level will also agree of the meta question of whether what the person is doing is acceptable, and the reverse for someone who disagrees.
As I allude to here I think the best approach is to simply argue over the object-level question, and leave aiside the dicussions about whether someone is being stategic, how truth-seeking they are, if they are “gaslighting” etc.
I think it would likely make a difference if you primarily had in mind the step where an AI system generates some text or video vs when a human user receives the generated content and communicates it further.
For the output generation itself my sense is there are various arguments out there and it will likely get litigated in various contexts but its unclear how it will pan out. Some arguments this is protected are arguments similar to the ones that came up in the Moody v NetChoice SC case (although Barrett explicitly calls out in a concurrence that AI might change the analysis), as well as the possibility that the rights of human “listeners” would be protected.
My take here is that this issue is likely to be important but is very much in flux and its worthwhile to think about how AI policy and legal actions can held this move in a positive direction, given that early cases on this topic could be highly influential on this area of law.
For human users who repost AI generated content, my gut instinct is that this is likely to be viewed as normal speech, with the AI part not changing things all that much necessarily. I think this would mean any regulations would need to address existing 1A exceptions (fraud, defamation etc.) or meet strict scrutiny.
Since the release of ChatGPT, at any given time, anyone on the planet with a few bucks could access the current most capable AI model, the SOTA.
I do think you are correct that what is going on with Mythos strongly suggests that AI companies will be more likely to withhold models and/or capabilities in the future, and that this is potentially very concerning.
On the other hand, I’m not sure that the quoted statement is strictly true. AI companies may have internal “helpful only” models or internal models with fewer safeguards that aren’t made publicly available. Likewise with everything that is going down between Anthropic and the current admin we know that they made available a special system (“Claude Gov”) to the military, which presumably has fewer safeguards compared to publicly available models.
If you follow through some of the links you get to this paper which seems to suggest that the mechanism might be requiring issuance of new shares?
I’m not so convinced by their arguments. It has the vibe of trying hard to avoid seeming like a taking while accomplishing the same thing as a taking, which I would expect a court to pick up on and be skeptical of. Particularly with the existing composition of the Supreme Court I think its going to be hard to rely on such arguments.