They both are injuries? The same as punching someone in the face is injurious whether in self defense or as an attack. The correct thing to do when someone starts getting violent is to stop them from continuing. This sometimes requires violence, hence often injuries. Sometimes the violent person hasn’t yet actually punched anyone, but is doing things that strongly imply that they are planning on hurting someone. Such situations also call for stopping the violent person from continuing, even though no actual injury has happened yet.
mruwnik
Karma: 748
I’ve done some baselining on MirrorCode, which is similar. I reckon I am (or at least was) a decent software engineer. Turns out it’s really hard to re-implement a byte-for-byte copy of a program which isn’t trivial or really well specified. And even those have quirks. My approach was usually to set up hypothesis as a test fuzzer in order to find as many edge cases as possible. This helps a lot. Still, I don’t think I ever got a full 100% correct on the final score, and a large portion of time spent on creating MirrorCode was a vetting process to filter out any programs that are “unfair”.
Most programs are buggy. A faithful copy of such a program requires both understanding the intent behind the program, and also working out what exactly is causing a given bug, and then reproducing it. A couple of the less polished programs I baselined caused me a lot of frustration when it turned out I had to break my beautiful abstractions and flows to make a monstrosity that spits out the same values as the program being tested. Another fun source of differences are system differences. Floating point differences are one source. But also e.g. i18n of the underlying system. All in all, loads of opportunity for fun.
Also worth mentioning is that there aren’t that many people around (the community is small) and the same people keep popping up everywhere. @Myles H is a fun case, who started volunteering for https://aisafety.info/, which in turn led to other opportunities and he’s now at Redwood. Volunteering is a good way to get yourself on people’s radar. Being in general useful is a very good skill to have, and underrated.
Wait, I’m confused. What is the difference between not admitting to themselves and not knowing? Do you mean something like subconsciously knowing? Or maybe cognitive dissonance?
Check Yudkovsky’s other writings (especially fiction) for multiple detailed discussions of these topics.
The simplest way to get rid of social injustice is to get rid of society. Most people would think that not an acceptable cost?
Nit pick of your nit pick is that he doesn’t propose that it would include using nuclear weapons, he proposes that it would include the will to use nuclear weapons. The point is not “lets start throwing nukes around” it’s “this threat is larger than nuclear war—so nukes should also be an option if nothing else will work”.
The thing is that in her world model, she did see the Holy Spirit. That’s how she understands “seeing the Holy Spirit”. It very much is a valid representation of her beliefs. Whether her model is correct (or even good) is a different matter. Christianity has a bunch of jargon, as any such group with a history of thought does (I yesterday had a bit of a confusing discussion before I realised that the meaning of “world model” is not obvious). It seems fair to say that it’s worth first establishing that all phrases being used are understood in the same way by everyone, but that also requires noticing when a given phrase is non obvious or even counter-intuitive.
The entity with the intention to deceive doesn’t have to be the same as the entity misrepresenting its (which?) beliefs, true. But in that case it’s better to say that “Christianity is acting in bad faith”, rather than “this Christian over here is acting in bad faith”. Saying that someone is acting in bad faith is a statement about their intentions, not their actions.
You seem to have a somewhat idiosyncratic interpretation of “lying’ (or I do—might be a cultural thing). My understanding of lying is “saying something you know to be untrue”. Which sort of is intentionally misrepresenting what they believe? Whether they do it instrumentally or terminally is besides the point?
“Accidentally bad faithing” doesn’t make sense (hence the “I like people who obey the law (here meaning never committing a social faux pas)” example). If you misrepresent the truth—or even outright lie—but but didn’t intend to, then that’s not bad faith. Bad faith is when you intentionally set out to deceive someone. The intentions here are important. It’s like the difference between manslaughter and cold blooded planned murder. Both result in a corpse, but the later set out to intentionally kill someone.
The Christian talking about the witness of the Holy Spirit is not setting out to deceive you. They truly believe in their position. It might not be true, it might leave you with a totally different understanding than they’re thinking of, they might even be aware that they’re using “witness” in a somewhat unusual manner (“open your heart to Jesus”...) etc., but it’s not explicitly intended to deceive you, and so is not in bad faith.
“Cynical” seems to often be used as an incorrect synonym for what bad faith is pointing at, e.g. “cynically mention the witness of the Holy Spirit as evidence”.
Why assume this is only words? AI systems can already generate images, video and sound, which gives you a lot more subconscious bandwidth (scents and tactile feedback are admittedly harder). There’s also the shortcut of finding a charismatic/convincing person and convincing them to help, which may or may not make things a lot easier.
It seems like being persuasive is mainly working out what the entity being persuaded wants to get and wants to avoid. Once you can work that out (basically—something like cognitive empathy, including how psychopaths work), you then just have to select arguments that suggest that they are more likely to get what they want (and less likely to get what they don’t want) if they agree with you.
If you can simulate another entity with high fidelity, then you can just run a bunch of different arguments by them and see which ones tend to work better. This transforms it into an optimisation problem. Or you can even do it the other way round and map which arguments work on which bins of people.
You might need to add sources for this. I roll to disbelieve, and Claude seems to also think that while you can construct a narrative that might be technically true, it would be at best misleading.
“Here are 5 bad interactions with this person” can be interpreted as “we don’t get along well”. “Here are 5 different people who had bad interactions with this person” is “that person doesn’t get along well with others”. This post is showing that certain people have patterns of harming others, as opposed to just having harmed the author.
where almost everyone seemed like a sexual abuser
There is something to the whole “my boyfriend will punch you in the face if you try anything funny” threat. I’m glad I don’t have to worry about violence nowadays, but it’s much harder to introduce credible deterrence mechanisms which aren’t backed by imminent physical pain.
I would like the world to be saved and think being good and not being evil
There’s also the whole thing about having something worthwhile saving. Winning all battles but losing the war is a very sad way to end.
Which is simply not true, according to my experience. You can. And even do it immediately
This is quite dependent on the person. I know people who can do so easily, and I know people who can’t at all. It might be something you can learn with practice, but until you can do it, you can’t.
Beware the typical mind fallacy.
I guess it’s mainly too long, and therefore unclear? There are at least 4 main points that you are addressing in one large comment, along with a bunch of smaller issues. Splitting it into multiple, targeted ones would make it easier to react to them—it would also make it easier for you to work out what people don’t like about it.
I think you’re making a good point, but it could be boiled down to 2-3 sentences
I’m not sure that modeling people as rational agents in this kind of situation is correct. I’d assume that for every 5 people who know, there are 5 who are certain they know but are incorrect, 5 who have no idea but sound authoritative and another 20 who heard something from someone and are pretty sure it was over there, maybe? It should sort itself out after a while, but depending on the circumstances the sooner you have accurate information, the better.
The ideal approach, of course, is to just ask Claude to come up with some example situations and then research where to go (with backups) - spend 10min on it once every now and then, to make sure you’re up to date, and just have the places marked somewhere.
I notice I’m confused now. Manifest Destiny makes sense in the context of this post—there’s something of value to be achieved, and there will be costs. I’m not sure if I agree with this, but it’s coherent. What I don’t understand is how egregores using people via their personal incentives (for lack of a better description) fits in? It would seem that people just being people and things happening is sort of the opposite (or at least orthogonal) to actively trying to make things better? Do you mean something about shaping incentives being the method of conquest? This seems obviously true (capitalism vs communism being an good example), but if so, then using colonialism as an example might be a bad choice, or at least would need more inference steps explained.
I’d guess you have a different understanding of what danger is. “has a history of attacking children” and “will have a corrosive effect on morale” are both dangers, just of different things. Companies kick out people that seem to be introducing a danger of being net negative for financial reasons (very widely understood). A church will kick out someone who is a danger of leading others into sin. A well managed forum will ban someone who is a danger of making the experience worse for everyone else.
Nowhere does the OP say anything about kicking people out for not going along with the group dogma. There are multiple groups where that is treated as a real danger, but I don’t think that’s a fair criticism here? Unless you mean by dogma something about open discussion, truth seeking etc., in which case acting against them really is a danger which needs to be removed. A garden without a gardener won’t last long.