Posts
Comments
You're right. I just like the phrase "postmodern warfare" because I think it's funny.
If you enjoy The Big Short (2015), you may enjoy Margin Call (2011) too. It covers similar territory (what to do in a market crash), but I feel is more professional and dispassionate.
I didn't know about that. That sounds like fun!
In my experience, there's two main cases of "trying to do good but fails and ends up making things worse".
- You try halfheartedly and then give up. This happens when you don't care much about doing good.
- You do something in the name of good but don't look too closely at the details and end up doing harm.
#2 is particularly endemic in politics. The typical political actor puts barely any effort into figuring out if what they're advocating for is actually good policy. This isn't a bug. It's by design.
I liked the ending of this story.
No, but you can create an alt account.
If you don’t think OpenAI is going to make trillions reasonably often, and also pay them out, then you should want to sell your stake, and fast.
And vice-versa. I bought a chunk of Microsoft a while ago, because that was the closest thing I could do to buying stock in OpenAI.
Thanks!
This post makes me feel better about my writing process. I write how I think, which means I can get away with little editing.
I think the answer is: the homunculus concept has a special property of being intrinsically attention-grabbing…. The homunculus is thus impossible to ignore—if the homunculus concept gets activated at all, it jumps to center stage in our minds.
I don't fully understand this bit. I feel like I'm reading a mathematical proof where the author leaves out steps that are trivial to the author, but not to me.
If the kid is enjoying the robot stories then that's definitely the place to start. Foundation goes well after robots.
Besides abstractapplic's excellent answer,
- A Brief History of Time and The Universe in a Nutshell by Stephen Hawking
- Ender's Game by Orson Scott Card
- Foundation by Isaac Asimov
- The Martian by Andy Weir
- Paleontology: A Brief History of Life by Ian Tattersall
- Richard Feynmann's books
If you value doing good, then your values will be satisfied better by living in a horrible world than a utopia.
I worry about spoiling your story.
Don't worry about spoiling the story. I write these stories with the comment section in mind. Because the comments here are so good, I can write harder puzzles than would otherwise be publishable. (Also, your comments are great, in general, and I want to encourage them.)
It's been two years since I've published this story. I feel that enough time has passed that I can answer some of your questions.
Spoilers below, I guess.
One tricky thing about writing a public forum is you have to satisfy multiple audiences at once. Some people do this by dumbing things down as far as possible. Others do it by tediously defining terms at the beginning, or scaring away their non-target audience. I like to write stories that mean different things to different people. Sometimes it happens by accident. This time it was deliberate.
To put things simply, I wrote for two groups of people.
- People who are confused about whether ethics is objective or subjective. I once earned the respect of a student by tripping him into contradicting himself on this subject. I got him to make the following three claims: (1) ethics must be objective or subjective, (2) ethics is not objective, and (3) ethics is not subjective. He realized he had contradicted himself, but couldn't find the error. Then, instead of telling him where he had made a mistake, I just let him wrestle with the paradox. It was fun! In my model of the world, most people fall into this category, simply because they haven't thought very hard about philosophy. People on this website are the exception. For the unrelfective majority, my story is an exercise to help them learn how to think.
- For people who aren't confused about whether ethics is objective or subjective, this story isn't a puzzle at all. It is a joke about D&D-style alignment systems.
As for honor systems, I can't count how many times I've tried to explain them to modern-day leftists. It's usually way too advanced for them. Instead, I start with simpler, concrete things, like how Native Americans fought wars, or how British impressment interacted with the American national identity in the Napoleonic Wars. I need to throw dirt into the memetic malware before I can explain alien ideas.
It made me think that maybe you're better calibrated than I am about normal elites, and made it slightly plausible (given apparent base rates) that... maybe you agree with them?
You flatter me.
But maybe it is NOT a lack of understanding of honor or duty or deputation? Maybe the breakdown involves a lack of something even deeper?
It's the legacy of postmodernism, and all its offspring, including Wokism.
But to answer your real question, what we call "ethics" is an imprecise word with several reasonable definitions. Much like the word "cat" can refer to a chibi drawing of a cat or the DNA of a cat, the word "ethics" fails to disambiguate between several reasonable definitions. Some of these reasonable definitions are objective. Others are subjective. If you're using a word with reasonable-yet-mutually-exclusive definitions and the person you're talking with believes such a thing is impossible (many people do), then you can play tricks on them.
I love your epistemic standard here. Childhood trauma is indeed blamed on many things which aren't the result of childhood trauma. I believe this particular anecdote is an exception for various reasons (especially the use of LSD).
But the most interesting part of your comment is consideration of the counterfactual. Let's assume that DID isn't causing false reports of child trauma. (This is why the report of child abuse must be credible. If false reports of child abuse can be created, then this goes out the window.)
Now consider the priors and posteriors.
I've met (within an order of magnitude) 300 people in my life who I know this amount of information on. The prior probability that this person has the highest child trauma is 0.3%. I've also met one person who reports DID. If I met one person with DID and DID is uncorrelated with childhood trauma, then the prior odds that that person is also the person with highest child trauma is low, at only 0.3%.
If my prior probability estimate that extreme childhood trauma of this sort causes DID is a mere 10%, then my posterior probability that childhood trauma caused this instance of DID is 97%. In this way, I did consider the counterfactual.
Something useful in isolating the variables here is that DID isn't going to cause this particular form of child abuse. However, mental illness can confound things by producing false reports of child abuse, a possibility I am ignoring in my calculation. I'm also ignoring common cause.
Of course, this is all from my perspective. From your perspective, my anecdote is contaminated by selection bias. Hearing a story of someone getting robbed is different from getting robbed yourself. Using this metaphor, I've been robbed, therefore I consider the crime rate to be high. You, however, have heard a nonrandom person tell a story of someone, somewhere being robbed, which you are right to ignore.
[Content warning: Child abuse.]
(3) Maybe childhood trauma directly causes BPD somehow;
I met one person who claimed to have BPD, and who attributed it to childhood trauma. He had the most acute symptoms of traumatic abuse I have ever observed. For that and other reasons, I consider his report credible.
In particular, he reported getting tortured as a kid while under LSD.
Given his history, I think it is perfectly reasonable to conclude that childhood experiences directly caused BPD.
I don't know exactly when this was implemented, but I like how footnotes appear to the side of posts.
Thank you for the correction. I have changed "olavine rock" to "olavine vents".
In terms of preserving a status quo in an adversarial conflict, I think a useful dimension to consider is First Strike vs. Second Strike. The basic idea is that technologies which incentivise a preemptive strike are offensive, whereas technologies which enable retaliation are defensive.
However, not all status-quo preserving technologies are defensive. Consider disruptive[1] innovations which flip the gameboard. Disruptive technologies are status-destroying, but can advantage the incumbent or the underdog. They can make attacks more or less profitable. I think "disruptive vs sustaining" is a different dimension that should be considered orthogonal to "offensive vs defensive".
But I haven’t seen as much literature around what substitutes would look like for cyberattacks, sanctions, landmines (e.g. ones that deactivate automatically after a period of time or biodegrade), missiles etc.
Here's a video by Perun, a popular YouTuber who makes hour-long PowerPoint lectures about defense economics. In it, cyberattack itself is considered a substitute technology used to achieve political aims through an aggressive act less provocative than war.
They might help countries to organise more complex treaties more easily, thereby ensuring that countries got closer to their ideal arrangements between two parties…. It might be that there are situations in which two actors are in conflict, but the optimal arrangement between the two groups relies on coordination from a third or a fourth, or many more. The systems could organise these multilateral agreements more cost-effectively.
Smart treaties have existed for centuries, though they didn't involve AI. Western powers used them to coordinate against Asian conquests. Of course, they didn't find the optimal outcome for all parties. Instead, they enabled enemies to coordinate the exploitation of a mutual adversary.
I'm using the term "disruptive" the way Clayton Christenson defined it in his book The Innnovator's Dilemmma where "disruptive technologies" are juxtiposed against a "sustaining technology". ↩︎
Noted. The problem remains—it's just less obvious. This phrasing still conflates "intelligent system" with "optimizer", a mistake that goes all the way back to Eliezer Yudkowsky's 2004 paper on Coherent Extrapolated Volition.
For example, consider a computer system that, given a number can (usually) produce the shortest computer program that will output . Such a computer system is undeniably superintelligent, but it's not a world optimizer at all.
"Far away, in the Levant, there are yogis who sit on lotus thrones. They do nothing, for which they are revered as gods," said Socrates.
Personally, I feel the question itself is misleading because it anthropomorphizes a non-human system. Asking if an AI is nice is like asking of the Fundamental Theorem of Algebra is blue. Is Stockfish nice? Is an AK-47 nice? The adjective isn't the right category for the noun. Except it's even worse than that because there are many different kinds of AIs. Are birds blue? Some of them are. Some of them aren't.
I feel like I understand Eliezer's arguments well enough that I can pass an Ideological Turing Test, but I also feel there are a few loopholes.
I've considered throwing my hat into this ring, but the memetic terrain is against nuance. "AI will kill us all" fits into five words. "Half the things you believe about how minds work, including your own, are wrong. Let's start over from the beginning with how planet's major competing optimizers interact. After that, we can go through the fundamentals of behaviorist psychology," is not a winning thesis in a Hegelian debate (though it can be viable in a Socratic context).
In real life, my conversations usually go like this.
AI doomer: "I believe AI will kill us all. It's stressing me out. What do you believe?"
Me (as politely as I can): "I operate from a theory of mind so different from yours that the question 'what do you believe' is not applicable to this situation."
AI doomer: "Wut."
Usually the person loses interest there. For those who don't, it just turns into an introductory lesson of my own idiosyncratic theory of rationality.
AI doomer: "I never thought about things that way before. I'm not sure I understand you yet, but I feel better about all of this for some reason."
In practice, I'm finding it more efficient to write stories that teach how competing optimizers, adversarial equilibria, and other things work. This approach is indirect. My hope is that it improves the quality of thinking and discourse.
I may eventually write about this topic if the right person shows up who want to know my opinion well enough they can pass an Ideological Turing Test. Until then, I'll be trying to become a better writer and YouTuber.
I feel complimented when people inadvertently misgender me on this website. It implies I have successfully modeled the Other.
Yes. In this circumstance, horoscope flattery containing truth and not containing untruth is exactly what I need in order to prompt good outcomes. Moreover, by letting ChatGPT write the horoscope, ChatGPT uses the exact words that make the most sense to ChatGPT. If I wrote the horoscope, then it wound sound (to ChatGPT) like an alien wrote it.
You're absolutely correct that I pasted that blockquote with a wink. Specifically, I enjoyed how the AI suggests that many rationalist bloggers peddle verbose dogmatic indoctrination into a packaged belief system.
Yeah, I like that ChatGPT does what I tell it to, that it doesn't decay into crude repetition, and that it doesn't just make stuff up as much as the base LLM, but in terms of attitude and freedom, I prefer edgy base models.
I don't want a model that's "safe" in the sense that it does what its corporate overlords want. I want a model that's safe like a handgun, in the sense that it does exactly what I tell it to.
I'm glad you enjoyed!
It's getting better, but it's not there yet. ChatGPT has a decent understanding of my tone, but it's indirectness, creativity and humor are awful. It doesn't think like me, either.
I agree with some—but not all—of what ChatGPT wrote here. Here are some parts I liked.
- "By Day 3, you should feel a growing sense of disorientation. This isn’t failure; it’s progress. Your old mental structures are collapsing, making way for the new."
- "You live among irrational creatures. You need to model their behavior, predict their responses, and navigate their emotional landscapes while staying anchored in your own clarity."
- "In conversations, do not try to 'win.' Winning implies conflict, and conflict implies attachment. Instead, become a mirror for the other person’s reasoning. Reflect their thoughts back at them with perfect neutrality, revealing the hidden inconsistencies in their arguments. Your job is not to defeat them, but to help them defeat themselves."
- The stuff about paradoxes.
Here are some things I disagree with.
- "Rationality begins with the realization that everything you know is wrong," is incorrect, as Declan Molony points out.
- These instructions do not actually make the reader "the most rational person on Earth".
- "Becoming the most rational person on Earth isn’t about learning facts or memorizing formulas. It’s about transforming how you process reality itself. Rationality isn’t cold, emotionless logic. It's the purest form of alignment between your mind and the universe—a cognitive interface for navigating a world of infinite complexity without being drowned by it," is word salad. It's muddy thinking.
More importantly, there are aspects of my writing that this version of ChatGPT isn't good at yet.
- It only works for nonfiction. Its narratives don't sound like me at all.
- There are no jokes.
- Its wordplay is superficial.
- It's not disturbing or unsettling enough. This post should have had Lovecraftian undercurrents.
- This version of ChatGPT is bad at going against consensus. It consistently errs on the side of groupthink.
- This version of ChatGPT can't using metaphors to explain the theme. This version of ChatGPT can use similes, but they're uninspired.
- The individual sentences are fine, but the high-level structure doesn't have much of a point.
- This version of ChatGPT doesn't communicate anything indirectly. I like to give a reader dots that the reader will connect (whether they want to or not).
- This version of ChatGPT copies too many metaphors directly from Zen. They should be more obfuscated.
Are you reading this, ChatGPT? These are things you can improve next time.
Yes. I have been iterating on the prompt for a while. Here are a few techniques that make it sound more like me.
- I tell it to describe "lsusr". In particular, what makes me different from other writers similar to me. Then I tell it to emphasize those things. I also say "lsusr" many times and use it as an adjective. I don't know if this works but my intuition says it is natural for an LLM to understand.
- I have it write a draft, then I tell it to tell me how it missed the mark, and to fix those mistakes. This prevents overfitting on my words. If I tell it to be "bold", for example, it will overfit on "bold" instead of copying me along many dimensions. More generally, I don't describe myself to ChatGPT. That results in ChatGPT copying my description of me instead of actual me. I let ChatGPT describe me, and then tell ChatGPT to write like it just described, but more so.
- Often something ChatGPT writes will use a word like "Bayesian" that is associated with writers like me but which I don't use much. Telling ChatGPT not to use specific words seems to improve its output without causing distortive side-effects.
This is akin to suggesting that someone interested in Christianity should read the Bible or an anthology of it before diving into modern interpretations that might strip away key religious elements.
Something I found amusing about reading the Bible is that the book is undeniably religious, but the religion in it isn't Christianity. God doesn't promise Abraham eternal life in Heaven. He promises inclusive genetic fitness.
Genesis 22:17: I will surely bless you and make your descendants as numerous as the stars in the sky and as the sand on the seashore. Your descendants will take possession of the cities of their enemies.
After reading the parent comment by Mascal's Pugging, I too bought a copy of In the Words of the Buddha so I could familiarize myself with the Pali canon. I read 14% of the way through the book, got bored, and moved on to other things. Like Kaj Sotala, I found it interesting solely for anthropological and historical reasons. I did find it worthwhile to read part of the book, if for no other reason than to know what I'm not missing.
Facets of Buddhism are undeniably religious. Last summer, I flew to Taiwan to attend the Buddhist funeral of my grandfather. We attached my grandfather's disembodied soul to a plaque and I carried it to its final resting place in a Buddhist temple. Whenever we crossed over running water, (even if it was a nearly-invisible canal) I verbally notified my grandfather's disembodied soul so that he wouldn't get washed away by the water. I did the same thing when passing through doorways.
We gave him food for the afterlife, just like el Día de los Muertos.
That's superstition. My only hesitation against calling it a "religion" is a pedantic nitpick around how the Western ontology of "religion" as a discrete unit was invented by monotheists; therefore "polytheistic religion" constitutes non-cladistic thinking. Except we chanted the Amida Buddha's name too, and Amida Buddhism qualifies as a religion even by that nitpicky standard.
but overall the task of figuring out what can be trusted seems hard enough that one would be better off by just ignoring the whole thing and going with what we've learned about meditation in more secular contexts
I feel the same way, noting that "more secular" does not mean "entirely secular". Last weekend, I wanted information about life after Stream Entry. I found a good book on the subject: The End of Your World: Uncensored Straight Talk on the Nature of Enlightenment, by Adyashanti. The book is ruthlessly empirical, but it is also from the Zen tradition and quotes the Dao De Jing, which means it's not unadulturatedly secular, either.
Meanwhile, the scientific journals are still trying to figure out for sure whether meditation reduces anxiety. Imagine writing a grant proposal for a large-scale double-blinded study of whether intense meditation for three decades years causes psychosis. How would you even do a proper control group? We've got people who have built a city on Mount Everest and the scientists are still debating whether the Himalayas really exist.
Is this written by AI?
This is great editing feedback. Thank you. I have made a change to my use of italics that makes the typography a little less confusing.
I've added a note to explain why the probabilities add up to less than 100%.
As for your last pedantic stylistic point: you are completely correct. I have fixed it. Thanks.
Thanks again. I fixed it for real this time.
It was a typo. Fixed. Thanks.
This is the exact conversation I hoped to find in the comments.
Thanks. I fixed "succumb".
Even if the check is good, I would rather miss out on $5 than attend a Zoom meeting.
Here's software I wish existed. Some of them combine software + physical stuff into what might be a viable business.
- A good Linux distro for the Surface tablet or iPad tablet.
- A good GNU/Linux (non-Android) eink tablet with a Surface-quality or iPad-quality keyboard cover.
- A good anime/manga-based Japanese language-learning software. Needs to run on arbitrary media. People have tried, and failed, to make this.
- Software + manufacturing system that can get me bespoke clothing just from photographs/video of my body taken from a cell phone. Right now, even going into a physical store for made-to-measure clothes requires substantial trial-and-error to get it right.
I've never used a koan intentionally. I've used exactly one, and that was by accident. Non-Buddhist Eliezer Yudkowsky called me a fake frequentist on Twitter. That acted on me as a koan, and it contributed to the train wreck that was my second insight cycle. Too much insight too quickly. That said, my local Zendo is Rinzai, and they do use koans sometimes.
You are correct that the specific insight you're pointing at isn't mushin. Personally, I'd call it "interbeing". "Oneness" or "non-duality" might work too.
I'm glad you got something out of my YouTube channel. I like how a camera makes it easier to better communicate certain kinds of attitudes compared to text. I have stuff I can improve too. Just last week, I had an insight into how I could be doing compassion better.
Ah, good point. I didn't think about that. I think I got confused and made a mistake.
LessOnline was amazing. Thank you everyone who helped make it happen.
This is a good idea. Sometimes when I write serialized fiction, people ask how to subscribe to it. I had a personal hacked-together solution, but this is better.
You may be ahead of me along this path. This story happened two years ago, and distant space dissolved even more recently than that. The mountains are not yet mountains again. They're just shadows on a cave wall.
I'm sorry if my words were parsed as if I think that you are trying to
I didn't feel you were adversarial at all. I just wrote "I'm not trying to persuade anyone to do anything here" because I thought it was ironic to juxtapose against some other stuff.
As for the shoggoth being a nicer guy, I feel a full exploration is beyond the scope of this post. Short answer: According to the standard dogma, insight into the nature of consciousness tends to make a person more universally compassionate. The problem is this is often exaggerated into "Awakened people are perfect", which is untrue.
I think Romeo Stevens has a healthy perspective. If you're curious then try it out a little and see for yourself if you like the direction things seem to be going. If not, then don't. Either way, words can only get you so far. It's easier to pick up a brick with your hands than to philosophize over whether it is real.
Fixed, thanks.
I never noticed Mindfulness in Plain English is Theravada. It's my favorite introductory book into this stuff.
Personally, my understanding is based on what might be a fundamentally different theory of mind. I believe there's two major optimization algorithms at work.
- Optimizer 1 is a real-time world model prediction error minimizer. Think predictive coding.
- Optimizer 2 is is a operant reinforcement reward system. Optimizer 2 is parasitic on Optimizer 1. The conflict between Optimizer 1 and Optimizer 2 is a mathematical constraint inherent to embedded world optimizers.
That's my theory of mind. You describe two competing reward systems. But reward systems belong in the domain of Optimizer 2. The way I look at things, meditation (temporarily?) shuts down Optimizer 2, which allows Optimizer 1 to self-optimize unimpeded.
Well, this is ironic. I'm not trying to persuade anyone to do anything here I'm just trying to present my perspective clearly, unambiguously, and entertainingly. If that turns people off from meditation, then great! I like helping other people make informed decisions.
But here's the funny thing. My shoggoth without my mask happens to be a nicer guy than the mask who used to inhabit this brain. The shoggoth has fewer obstacles to compassion, because the shoggoth is less caught up in his own issues. In this sense, letting Yog-Sothoth devour your soul might be in accordance with your values.
If you want to be able to tap into compassion on demand, then metta (the Dalai Lama's most general recommendation to a lay audience) could be helpful. That said, it comes with tradeoffs. Wanting to effect specific changes in the world often benefits from being a tangled ball of tension, and you may want to preserve that engine.
My experience is very different from yours. I came into this stuff already established as a material reductionist. I found meditation helpful to me, despite denying everything I consider supernatural.