Archive
- Taking social choice seriously: An alternative approach to reward modeling in RLHF
-
- Unconditional conditioning: Removing sleeper agent behavior in a toy model
-
- The Typed Transformer: Intro and architecture
-
- Assorted links XVI
-
- Progress and preservation in IDA
-
- A corrected model suggests climate change interventions may be within a factor of two of direct cash transfers
-
- The Economic Lives of the Poor
-
- Assorted links XV
-
- The reliability of moral judgments: A survey and systematic(ish) review
-
- Uncertainty and sensitivity analysis of GiveWell's top charity rankings
-
- Sensitivity analysis of GiveWell's cost-effectiveness analysis
-
- Uncertainty analysis of GiveWell's cost-effectiveness analysis
-
- Conditioning in causal graphs
-
- Instrumental variables on causal graphs
-
- Flip it and reverse it
-
- Assorted links XIV
-
- Baby's first graphical causal models
-
- Critiques and claims regarding Evidence-based Policy
-
- Optimism, regret and indifference
-
- YAAS evidence-based policy
-
- Assorted links XIII
-
- Assorted links XII
-
- Priority decisions
-
- Dominated decisions
-
- Assorted links XI
-
- Anthropological enumerations
-
- Pigouvian compendium
-
- Assorted Links X
-
- Gas taxes for thee, but not for me
-
- Assorted links IX
-
- Anthropological clickbait
-
- YAAS social norms
-
- The curse of the altruistic voter
-
- Assorted links VIII
-
- Clickbait from Norms in the Wild
-
- YAAS
-
- YAAS epistemology
-
- Shilling for Anki
-
- Assorted Links VII
-
- The moral imperative and mortal peril of maximizing
-
- A non-exhaustive list of putative problems with ignorant priors
-
- Ideal theory and decision theory
-
- YAAS human cognitive architecture and learning
-
- Assorted links VI
-
- An example of the lazy approach to AI safety
-
- False dichotomies and the ideal theory debate
-
- Lee Kuan Yew—Deity or dude?
-
- Exemplar's curse—Now with 80% more math!
-
- The Exemplar's Curse and Singapore
-
- Is development easy? The Singapore Story
-
- Assorted links V
-
- Ideal theory in the shadow realm
-
- Ideal theory as calibration
-
- Utopia and an infinitude of secretaries
-
- How not to write a book
-
- Deposition schemes
-
- Constructive
-
- Mo money, mo problems—autocrat remix
-
- Standalone–Value of information calculator
-
- Value of information calculator
-
- Inclusive and extractive societies each have structural advantages
-
- Preference utilitarianism—psychological or metaphysical? III
-
- Preference utilitarianism—psychological or metaphysical? II
-
- Preference utilitarianism—psychological or metaphysical? I
-
- Assorted links IV
-
- In praise of principal-agent problems
-
- Autocrats can accelerate growth through cooperation
-
- Minor blessings of a god emperor
-
- Hot takes—The Moral Limits of Markets
-
- Summary—The Moral Limits of Markets
-
- Doning with the devil
-
- Assorted links III
-
- Putting-out with smartphones II
-
- Ethics as well-order
-
- Standalone–Build your own von Neumann–Morgenstern utility function
-
- Build your own von Neumann–Morgenstern utility function
-
- Pareto improvement as partial order
-
- Putting-out with smartphones I
-
- YAAS Reading: Most of us read with our eyes and our brains
-
- A visual intuition for the instrumental argument for equality
-
- Notebook–A visual intuition for the instrumental argument for equality
-
- Notebooks
-
- Meta Monday
-
- Assorted links II
-
- Some things which are and aren't Pareto optimal
-
- Innocuous and invidious majoritarian tyrannies
-
- Self-flagellation Friday
-
- Assorted links I
-
- A lazy approach to AI safety
-
- The scarcity of cooperatives
-
- Graph of contents
-
- Toward an alternative bibliometric
-
- On casual futurism
-
- Choose your own exposition
-
- A quorum alternative
-
- About
-
Collectively Exhaustive
A weblog