Home | Search

2026-02-09

Update

Disclaimer

• Quick Note

Main

• I have noticed myself violating all of my own rules listed here, basically all the time.
• In the last one year after I first wrote this post, I have made a commitment to work on ASI risk, published a lot of content on my website, spoken to lots of people on this topic, and worked on multiple high quality projects that aren't just "quick notes".
• First, I just want a register an emotional reaction here: "holy fuck, past Samuel saw this trainwreck coming in one year in advance"
• can, possible, because
  • It is true that most relationships between X and Y, for most X and Y in the real world, are multicasual and non-linear. However, trying to consistently use the phrases "causually downstream of" and "causally upstream of" has made my writings less comprehensible not just to others, but even to myself.
  • The word "because" comes very naturally to me, in my thinking, not just my writing.
  • I actually think I am making a lot of mistakes in thinking because I use the word "because" so frequently. I notice that there is some causal relationship between X and Y and immediately jump to say "X happened because Y".
  • Did you notice it? I literally just made the same mistake again.
  • To be more precise, the fact that I make a lot of mistakes in my thinking is casually downstream of the fact that when I notice some causal relationship between X and Y, I often immediately jump to saying "X happened because Y"
• less, more
  • So much of discussion, especially political discussion, seems to really be simulations all the way down. There is an objective reality to politics, and that objective reality is what are masses, charges, positions, velocities, accelerations, etc of all the atoms in your brain.
  • Most people (myself included) are so far disconnected from this objective reality when it comes to political discussions. We don't even deeply understand what is going on inside our own brains, let alone inside anyone else's.
  • The words you are uttering are not the same as the words you are thinking. The words you are thinking right now are not the same as the thoughts and feelings you would have if you either self-reflected more, or tested them out more in the world. And even all these thoughts and feelings seem further disconnected from what is actually going on neurochemically, which are further disconnected from the physics.
  • So yes, when I use words like "atleast one", "sometimes", "often", "most", "almost all", "usually", etc, usually this is not a word that even I myself can define more precisely.
  • Again, to make another meta joke, I just used the world usually in the sentence above. And I can't define that more precisely either
  • In a casual discussion, suppose I say "usually I do X". I haven't actually gone and measured what fraction of the time I do X, under what circumstances, in any precise manner. What I am actually reporting is closer to a gut reaction based on a simulation of my past self that I have running in my head.
  • (I don't use the words "less" or "more" as frequently as I use words as I use the words "atleast one", "sometimes", "often", etc. But I think this is a general point being made about all quantifier words. In my writings, I find the need to quantify time, and quantify the opinions of other people, a lot more often than I find the need to quantify actual masses, distances, and so on.)
• is, ought
  • First, a bunch of my latest philosophy on moral realism versus non-realism
    • I do now think morality is atleast somewhat solveable, but it is complicated. Given some initial state of some group of people, you can actually run a deterministic simulation of their brains and environment on a computer. After some (simulated) years, you can measure mental state of each person down to the atomic level. You can then map each mental state to some function that says which states are "good".
    • If you do this, my guess is you will notice strong commonalities between the solutions you got for each person. That would make morality atleast semi-objective.
    • One major problem you'll face trying to do this in-practice (even if you did have the tech) - is that you maybe can't run a simulation for an individual's brain without also having other individuals and an environment inside the same simulation. Morality is atleast somewhat social, and people living in complete isolation, atleast in natural biological environments, tend to literally kill themselves.
    • Another major problem you'll face is the mapping function. Is high dopamine level good? Is the person uttering the words "I feel happy 8 out of 10" on a self-reported survey good? Right now, a number of people would agree with me that measuring a few different neurotransmitter levels is the closest we can get to something that has consensus. It's not obvious to me if humanity will ever reach consensus on one common mapping function or not.
  • Now, coming back to reality
    • I do find myself judging people basically all the time, for political reasons. This may be something I need to either change or accept about myself. I am currently ready to do neither.
    • It is empirically true that persuading people to change their core values or core life strategy is very hard and a waste of time for most people most of time. What works more often is to assume their self-reported values are the right ones (no matter how much you internally dislike these values), and give them advice with the assumption in both.
    • I do still think persuasion can be cracked and we can literally have the next political or religious leader figure this out at scale. This is not a job most people seem ready for, but maybe some people are ready.
• On naming people by name
  • First, there are a lot of political pressures to not name people by name. It is harder to say that Mohammed bin Sulman's wife persuaded him to do X, where X is some action that lots of people think is bad. It is easier to say that the Saudi Arabian govt did X.
  • Lots of people are hiding behind abstractions on purpose. If MbS does X, you often actually have no idea who was the person who actually convinced him to do X, who deserves maybe even more blame than he does.
  • In my writings, I have often criticised US govt and been more specific, by naming intellgence community or executive branch or specifically their leadership and so on. I do not actually know how the exact network of relationships between say, the heads of the NSA CIA and FBI and their one-level juniors works, and it is non-trivial on purpose to figure it out.
  • I agree it is generally not ideal to write sentences of the form "society should do X". Atleast some of the time, it is possible to identify a specific person Alice who should do X, break X down into subproblems and solve many of them, actually cold email Alice and try to persuade them that 'look I've already done most of the work, why don't you finish this?' (Or if cold email fails, try to get a referral from one or two degrees of connection anyway, etc etc) I have had first-hand experience of this succeeding.
  • I still think it is not entirely useless to utter the sentence "society should do X". Either you or someone else might read this post, and then be inspired to do the N steps after that to make X actually happen.
  • Yes, changing people's core values is often still hard. But maybe they're malleable if you make it easier for them to change them.

2026-01-03

(made minor edits, mostly this post was written by 2025-03-29)

Taboo words on this website

Disclaimer

• Quick Note

I am sympathetic to the lesswrong social norm of taboo words. Basically you avoid the usage of words that are not defined in a clear way where everyone agrees on the definition. If you think somsone is using a word without clear definition agreed by everyone, you can ask them to repeat their point without using this word.

Taboo words on this websites:

• can, possible, because
  • On this website, especially on the high quality posts, I almost will never say "X happened because Y". I will almost never say "X is possible" or "X is impossible". (I hold myself to a lower standard on "quick notes".)
  • Saying "X happened because Y" means in an alternate universe where Y did not happen, X would not have happened.
  • Especially when studying history it is difficult to construct this alternate universe thought experiment in a way everyone agrees what the thought experiment should be, and what its outcome would be. (I accept that in theory this thought experiment works fine. Human brains don't violate deterministic universe. It's just in practice we don't know how to get the outcomes of these experiments.)
  • In practice often multiple pre-conditions existed when an event happened. The combination of those pre-existing conditions is not simple linear algebra, "A and B and C => Y", it's "f(A,B,C) => Y" for some poorly understood f.
  • I prefer the phrases "causally upstream" and "causally downstream" over the word "because".
  • There are different levels of "impossible". There are events that would have required people to have different ideas or imaginations in order to happen, there are events that merely required the right amount of funding or the right set of social bonds to form, there are events that required increased amounts of engineering resources, there are events that would straight up violate the laws of physics if they happened. Unless an event would violate the laws of physics, I'm hesitant to say it is impossible. And if it violates the laws of physics, I may be open to its possibility, human understanding of our current physical laws is less than a century old, it could have exceptions.
  • If nothing is impossible, then saying "X is possible" also conveys zero useful information so I avoid saying it.
• less, more
  • I will almost never say "X is less" or "X is more". Scales are relative, and where exactly the middle of the scale gets defined to be is often a political question. I will almost always say "X is less than Y" or "X is more than Y". This could be when referring to any adjective. X could be more beautiful, powerful, wealthy, empathetic, knowledgible etc. than Y, but X will never just be more beautiful, powerful, wealthy, empathetic, knowledgible etc. period.
• should, ought
  • I wrote a lengthy disclaimer on is-ought distinction elsewhere on this website. In short, I believe is-ought is real and really important. I will almost never tell someone they should value XYZ as a terminal goal in life. Instead if I am asked to offer advice, I will tell them potential consequences of various actions. Whether those consequences are good as per their values is something they'll have to decide on.
  • In general I'll be specific about who the subject of each sentence is, and make sure it points to individuals not a group. Often people skip the subject and the sentence gets interpreted with "everyone" as the subject. I will almost never say "society should do/think/say/do X" or "the government should do/think/say X" or "everyone should do/think/so", I will generally be more specific about exactly which people in society or government I am referring to.

Subscribe

Enter email or phone number to subscribe. You will receive atmost one update per month

Comment

Enter comment