See also this post for brief notes on:
  • Ross Anderson | Security Engineering
  • David Christian | Origin Story: A Big History of Everything
  • Richard Feynman | The Character of Physical Law  
  • Richard Feynman | The Feynman Lectures on Physics
  • Jessica Livingston | Founders at Work
  • Walter Scheidel | The Great Leveler: Violence and the History of Inequality
  • Thomas Schelling | The Strategy of Conflict



  • Greg Egan | Diaspora
    • A wild and very well thought-out ride through a far-future world, featuring fictional physics theories plausible enough that they may one day win a Nobel. 
  • Greg Egan | Permutation City
    • This book is mostly a sustained thought experiment that extrapolates crazy but internally consistent conclusions from its philosophical premises.
  • Neal Stephenson | The Baroque Cycle (consists of Quicksilver, The Confusion, and The System of the World) 
    • An expansive story about science, finance, politics, and alchemy in the late 1600s and early 1700s. Almost every natural philosopher of the time makes an appearance, with Newton, Leibniz, and Hooke each being significant characters. It's published as a trilogy, but reads more like a single gargantuan novel.

See also this post for brief notes on:
  • Scott Alexander | Unsong
  • Lois McMaster Bujold | The Curse of Chalion
  • Ted Chiang | Exhalation
  • Hannu Rajaniemi | Summerland 
  • Neal Stephenson | Cryptonomicon


  • Growth and civilisation
    • A constantly expanding economy is one reason why societal norms and moral values have shifted for the better. If growth ceases, a reversion from norms of value-creation and tolerance back to norms of capturing value from others and defending your own in-group is possible in the long run.
  • EA ideas 1: rigour and opportunity in charity
    • Effective altruism is about carefully reasoning how to do the most good. A focus on impartial welfarist good, the effectiveness of our efforts, and open-minded uncertainty-acknowledging reasoning leads to a different picture of charity than the usual, but also one that is more likely to do good.
  • EA ideas 2: expected value and risk neutrality
    • A rational agent maximises the expected value of what it cares about. Expected value reasoning is not free of problems, but, outside extreme thought experiments and applied carefully, it clears most of them, including "Pascal's mugging" (high-stakes, low-probability situations). Expected value reasoning implies risk neutrality. The most effective charity may often be a risky one, and gains from giving may be dominated by a few risky bets.
  • EA ideas 3: uncertainty
    • We are uncertain about both what is right and what is true (being mindful of the difference is often important). Moral uncertainty raises the question of how we should act when we have credence in more than one moral theory. Uncertainty about truth has many sources, including ones broader than uncertainty about specific facts, such as our biases or the difficulty of confirming some facts. These uncertainties suggest we are unaware of huge problems and opportunities.
  • EA ideas 4: utilitarianism
    • While not a necessary part of EA thinking, utilitarianism is the most successful description of the core of human ethics so far. In principle (if not practice, due to the complexity of defining utility), it is capable of deciding every moral question, an important property for a moral system. Our moral progress over the past few centuries can be summarised as a transition to more utilitarian morality.
  • Technological progress
    • There are several reasons to think that it's easy to care too little about technological progress. There are many ways we can try to model what technological progress should be like, and hard to figure out which model is right. Which model is right impacts the question of whether technological progress is stagnating, which is also a confusing question. Despite all this confusion, to bring about a good future probably requires that our civilisation gets a lot better at making choices about technology.
  • Nuclear power is good (alternative title: burning things considered harmful)
    • Nuclear power is seen as dangerous and unclean. In fact, statistics reveal it is equal to wind and solar in both safety and environmental impact, while the health and environmental impacts of fossil fuels (in particular coal) are massive; as bad in lost human life as a Chernobyl a week, and a significant driver of climate change. We should build more nuclear power as fast as we can.
  • Death is bad
    • Surprisingly many people have arguments against making people immortal, ranging from environmentalism to effects on social progress to equality. I argue that technology that can remove the mandatory death sentence that everyone is born with would be good. 
  • Effective Altruism in practice
    • EA ideas are discussed in a previous post series, but what does the EA movement/community/whatever actually look like in practice, where did it come from, and whose idea was it to give those philosophy nerds all that money?
  • EA as a Schelling point
    • A significant way in which the Effective Altruism community creates value is by acting as a "focal" or "Schelling point" where talented, ambitious, and altruistic people tend to gather and can meet each other. It might be useful to think about what optimising for being a Schelling point looks like, and I list some vague thoughts on that.
  • AI risk intro 1: advanced AI might be very bad (collaboration with Callum McDougall)
    • The most likely way for human civilisation to be destroyed this century is through advanced AI systems, and in particular misaligned AI systems (rather than just humans using advanced AI for bad ends). This post is meant as an accessible introduction to the arguments.
  • AI risk intro 2: solving the problem (collaboration with Callum McDougall)
    • The field of AI alignment is growing but does not yet have a central paradigm. This post surveys the types of work that people are concretely doing to try to reduce risks from advanced AI systems. If you're confused about what this work actually looks like, this post is for you, regardless of your background.
  • A model of research skill
    • Doing research means answering questions no one yet knows the answer to. This post tries to find and describe the core parts of being good at this

Project reports

  • Deciding not to found a human-data-for-alignment startup (collaboration with Matt Putz)
    • Matt Putz and I worked together for the first half of summer 2022 to figure out if we should found a startup with the purpose of helping AI alignment researchers get the datasets they need to train their ML models. This post is a summary of our findings, and why we decided to not do it. 



  • Two proofs
    • Two accessible, visual, and (dare I say it?) fun proofs of simple mathematical results.
  • Classical physics
    • A summary of the form, gist, and "character" of all fundamental laws of classical physics.
  • Data science 1
    • Notes on fundamental data science concepts (notation; some probability laws; maximum likelihood estimation; supervised and unsupervised learning; fitting, interpreting, and visualising linear models; empirical distributions; KL divergence).
  • Data science 2
    • More data science notes (Monte Carlo methods; Bayesianism and frequentism; randomised computational methods for Bayesian and frequentist calculations; Markov's, Chebyshev's, and Jensen's inequalities; causal diagrams; Markov chains). 
  • Lambda calculus
    • The lambda calculus is a simple model of computation (like Turing machines). This post introduces it, shows how to do useful things in it, and works up to a lambda calculus interpreter written in lambda calculus. 
  • Information theory 1
    • Notes on what information and entropy are (both intuitively and axiomatically) , basic concepts and results in information theory (mutual information, relative entropy AKA KL divergence, the data processing inequality, etc.), and what the point of source and channel coding is.
  • Information theory 2: source coding
    • Notes on source coding, a big branch of information theory that deals with compressing information, including an overview of Huffman, arithmetic, and Lempel-Ziv coding, and a proof of the source coding theorem.
  • Information theory 3: channel coding
    • Notes on channel coding, a big branch of information theory that deals with storing information in error-resistant ways, including an overview of Hamming codes, and a proof of the channel coding theorem.



    No comments:

    Post a Comment