Welcome to my website.

I already have a Substack, and a Twitter Account.

However, I want my texts to build on each other, and these platforms are not designed for that. Even though articles build on top of each other, the motivation for each article should be self-contained.

Basics

This section outlines the core principles behind my thoughts. Armed with them, predicting my point of view about most topics should be much easier.

Death is bad. We keep getting fucked by it. We keep dying. And we have no global plan to stop it.
We are on track to human extinction from AI. As in, everyone dying. The rest pales in comparison.
We live in a physical world. This is the foundation of my worldview. It is easy to take it for granted, but can always benefit from reflecting on it.
Humanism is great, here is an introduction to it: Humanism.

Collective Failures

Collectively, we do not take notes. This makes us unable to systematically care, learn and improve. Without taking notes, there can be no accounting nor accountability.
Ideologies are slow. They are a good tool for coordination, but they are not a good tool for thinking. Unfortunately, they are the best we have for coordinating our beliefs at scale.

Intelligence

We would all benefit from reflecting about obvious stuff. This entire section is dedicated to this idea.
People who do not take notes condemn themselves to forgetting too much.
Spaced Repetition is the State of the Art of committing things to memory.
Reliable processes (missing) are the key to solving most of our problems.

My Physicalism

What is Common Knowledge? explains how coordination relies a lot on social knowledge. While the view is informed by Epistemic Modal Logic, the content of the page is legible to non-mathematicians.
Information is physical. The realm of information is often thought about in abstract terms like "memes", "beliefs", "intelligence", etc. I prefer to think of it as physical.
Coordination is physical. The realm of coordination is often thought about in abstract terms like "social structures", "institutions", "norms", etc. I prefer to think of it as physical.
We are physical (missing). We are physical beings. Our beliefs, our values, our emotions, our memories, our ability to care, all of those are physical.
Logicism vs Physicalism (missing) explains the main alternative to physicalism.

Legacy

Below are a couple of legacy posts from my Substack and Twitter.

Substack

These are my Substack Articles. I do not plan to update their content, but I wrote a bunch of them in the past and sometimes, I feel like writing longer articles.

Lying is Cowardice, not strategy was my first article. It explains how many people in the AI Safety community enforce the Overton Window through self-censorship.
On Lies and Liars is a follow-up.
For Civilisation and Against Niceness is a more general article on how niceness can be corrosive when not counterbalanced.
There is no IQ for AI dissects the implicit belief in the existence of predictable intelligence thresholds for AI.
About Feelings is a pragmatic approach to understanding feelings.
Thoughts about Civilisation foreshadows this website. It is actually 3 articles in one.
- Compromise is the Essence of Democracy explains that being constantly dissatisfied in a consequence of a healthy democracy. Conversely, if some people were fully satisfied, it would be a sign of an oligarchy.
- We are all just haggling takes a step back on polarising issues, and shows that we could be disagreeing much more than we currently do. In the grand scheme of things, we are at the level of haggling, more than at the level of grand disagreements.
- Materialist Moral Progress explains how material progress facilitates moral progress.
Preventing Extinction from Superintelligence describes my stance on Superinteligence and extinction from AI.
Formalism explains my views on meta-mathematics, and the core intution underpinning most of the progress in mathematics and computer science.
Writing Bankruptcy is where I changed my approach to writing. I had too much in the backlog and needed a better system. It also foreshadows this website.
Notes: Skills, Writing and Better Reddit is yet another article that foreshadows this website, and would have benefitted from being split.
- Some Skills lists 3 skills that are fundamental to thinking: re-reading the same thing 100 times, staring at the wall and taking notes.
- The Pain of Writing lists the motivations behind my writing style.
- A Way describes how I think and act.
- Deep Conversation Software is my usual complaining about how we need better software to have conversations than Social Media, and how Reddit is not it.
Myopic Changes and the Baseline features an intellectual technique that I often use to evaluate plans: comparing all possible plans to a conservative realistic baseline. This makes it easy to stay grounded.
Giga-Braining and the Complexity Task is an early attempt at explaining how dangerous complexity is and how consistently smart people get screwed by it.
People mastering different topics can lead to conflicts. It does what it says on the title. People in good faith and with the same values can enter conflicts merely because they have different areas of expertise. The article's content is valuable, but it is quite poorly written.
Rationalism, Authority, Intuition and their Interactions is a follow-up to the previous article. It describes major frameworks that people use to act in the world, and how easily they can talk past each other.
Thinking about AI Extinction Risks differently describes a common failure mode in discussions about AI extinction risks, where one party is confused about why not everyone is manic if extinction risks are real, and the other party just talks about orthogonality thesis and instrumental convergence instead.
On Boundaries features some musings around the concept of boundaries in game theory.
Practical Phrases is a poor explanation of how practicing common phrases can still be useful.
White Mirror is a rant against overly cynical anti-System fiction.
On Maintenance is one of my favourite articles. There, I investigate maintenance, which is both critical and underappreciated.
The Tally is a short foray into psychoanalysis. It describes the Tally: the dynamic where everyone keeps a warped emotional score of everyone else's actions.
Incoherence and Dissonance explains how Incoherence is the default state of complex beings, and how Dissonance - our immediate drive to ignore it - is a bad coping mechanism.
Belief Contradictions is a follow-up to the previous article.
The Responsibility of the Weak exposes a common moral loophole. From "With great power comes great responsibility", we get "With little power comes little responsibility". And thus everyone has an incentive to be weak.
Anger Checklist does what it says on the title.
Coordination Projects is a list of projects focused on coordination. It arose because many people asked me "Concretely, what do you think would help with coordination?". This is another article that should have been split.
- Practical Phrases Repository is an idea for building a Wiki of useful phrases that are easy to forget or not practice.
- Grounding Self Mastery describes the need for an equivalent to the IQ test for self-mastery.
- Coordination Research describes a couple of concrete research questions that a field of Coordination Study would tackle, and for which an answer would greatly improve our ability to coordinate as humanity.
- Good Neutral Policies extends the concept of Pareto Optimality to the more concrete setting of democratic policies. The goal is to draft a slamdunk political agenda made exclusively of such policies.
- Community Software lists desiderata for coordination software.
- Humanity Priorities is about the need for a 21st century Universal Declaration of Human Rights.
- No BS Methodology is an attempt at building a reasonable methodology that all members of a small group can agree on when tackling problems.
- Trusted Public Third Parties describes what new trusted public third parties we would benefit from.
The Source of Happiness is a recounting of an early introspection of mine. I just wanted to give an example of a non-religious, non-buddhist, non-psychoanalytic, approach to understanding oneself.
About the Progress Studies Movement describes how I feel about the Progress Studies movement: it is too focused on superficial problems like "building more" and not enough on making our institutions strong enough to not screw itself in the first place.
About Brain Laterilisation explains why I do not buy the claim that "logical rational thinking" differs strongly from "creative artistic thinking".
Rethinking Superintelligence describes a common failure in the discourse around superintelligence: grounding superintelligence on "smarter than a human" as opposed to "smarter than all of humanity combined".
Core Disagreements in the AGI Discourse is an attempt to unfuck the AGI discourse by identifying its upstream disagreements rather than re-hashing the same downstream points over and over again.

And that's it for the old Substack Articles!

Tweets

I sometimes write long tweets. While Substack articles were a bit longer than what I want, these ones are shorter. They are thus interesting from a different perspective.

No Interest in Human Intelligence describes how as a society, we have given up on improving human intelligence, and how we are avoiding it by instead pouring all of our hopes and money into an artificial superintelligent saviour for humanity.
Humanity should end death. This tweets states it, and deplores the fact that it is not part of any political agenda.
Weak Alignment vs Strong Alignment makes a distinction between "Making ChatGPT format its answers nicely" and "Enforcing that an AI system smarter than all of humanity combined only does what we want".
The Prisonner of Hell Dilemma should be a full article on its own. It is the main mechanism by which nice people make the world a worse place.
Patience in Disagreements is a short piece on how we should be more patient in disagreements. Genuine disagreements take time to resolve, and getting frustrated only diminishes the chance of a good resolution.
Self-fulfilling Performativity uses a stupid meme as a starting point for a reflection on how "faking it" can "make it" unbeknownst to us.
80 billion humans is a thought experiment about what an AI lab stating their intent to put a large amount of AGI instances would mean for humanity. Soon after, Dario published his essay titled machines of loving grace.
Fast Hardware makes for Slow Software is a rant about not only is software enshittifying despite hardware getting faster, but it is also enshittifying because hardware is getting faster.
As humanity, we are not Confronting Death seriously. We are mostly dodging the problem.

Site Map

Meta