For forecasting related research, see forecasting
Current projects
I’m currently doing private consulting, and writting up my disillusionment with EA:
- Some melancholy about the value of my work depending on decisions by others beyond my control
- Why are we not harder, better, faster, stronger?
- Brief thoughts on CEA’s stewardship of the EA Forum
- Hurdles of using forecasting as a tool for making sense of AI progress
Past projects
Estimation of values
I spent a few years of my life grappling with EA (effective altruism) being nominally about doing the most good, but it not having good tools to identify and prioritize across possible interventions. Eventually I gave up when I got it through my thick head that despite my earlier hopes, there wasn’t much demand for the real version of this—as opposed to the fake version of pretending to evaluate stuff, and pretending to be “impact oriented”. Still, I think it’s an interesting body of research.
- Five steps for quantifying speculative interventions
- A Critical Review of Open Philanthropy’s Bet On Criminal Justice Reform
- An experiment eliciting relative estimates for Open Philanthropy’s 2018 AI safety grants
- A flaw in a simple version of worldview diversification
- Shallow evaluations of longtermist organizations
- 2018-2019 Long Term Future Fund Grantees: How did they do?
- Find a beta distribution that fits your desired confidence interval
- Valuing research works by eliciting comparisons from EA researchers
- External Evaluation of the EA Wiki
- Use a less coarse analysis of AMF beneficiary age and consider counterfactual deaths
- Simple comparison polling to create utility functions
- A Bayesian Adjustment to Rethink Priorities' Welfare Range Estimates
- Relative Impact of the First 10 EA Forum Prize Winners
- An estimate of the value of Metaculus questions
- Relative values for animal suffering and ACE Top Charities
- Some estimation work in the horizon
- Estimation for sanity checks
- Pathways to impact for forecasting and evaluation
- Building Blocks of Utility Maximization
- Brief evaluations of top-10 billionnaires
- Use of “I’d bet” on the EA Forum is mostly metaphorical
- Things you should buy, quantified
- People’s choices determine a partial ordering over people’s desirability
- Impact markets as a mechanism for not loosing your edge
- Updating in the face of anthropic effects is possible
Red teaming
If you are trying to be as effective as you can, wouldn’t you like someone to point out where you might be going wrong? Not so!
- The NegativeNuno EA Forum account covers negative criticism that I’m uncertain about.
- A past piece on this topic is Frank Feedback Given To Very Junior Researchers
- A Bayesian Adjustment to Rethink Priorities' Welfare Range Estimates
- A flaw in a simple version of worldview diversification
- Review of Epoch’s Scaling transformative autoregressive models
Shapley values:
If you care about doing good together, you should care about how to coordinate on who does which projects. Shapley values solve this, as does adjusting the counterfactual value to be more like Shapley values. I named my consultancy after this concept.
- Shapley Values: Better Than Counterfactuals
- Shapley Values and Philanthropic Coordination Theory
- A Shapley Value Calculator
Economic models of social movement growth:
You can make progress on gnarly economics questions by throwing compute at it. Unfortunately, reality is complicated enough that these models won’t capture all the assumptions you care about, and so might not be all that informative in real life.
- A Model of Patient Spending and Movement Building
- Labor, Capital, and the Optimal Growth of Social Movements
Categorization of new causes:
Let’s just create some obvious infrastructure to track suggestions that people make!
- Big List of Cause Candidates
- A Funnel for Cause Candidates
- International Supply Chain Accountability
- I am paying someone to update this list with causes proposed in 2021.
Technological discontinuities:
People are talking about “technological discontinuities”. How often do they happen?
AI-related
Is this AI thing going to doom us all?
- Hurdles of using forecasting as a tool for making sense of AI progress
- My highly personal skepticism braindump on existential risk from artificial intelligence
- A concern about the “evolutionary anchor” of Ajeya Cotra’s report
- There will always be a Voigt-Kampff test
- A basic argument for AI risk
- Can GPT-3 produce new ideas? Partially automating Robin Hanson and others
- Review of Epoch’s Scaling transformative autoregressive models
- Straightforwardly eliciting probabilities from GPT-3
- Military Global Information Dominance Experiments
- AI race considerations in a report by the U.S. House Committee on Armed Services
The fate of social movements:
- Why do social movements fail: Two concrete examples
- Why did the Spanish Enlightenment movement fail? (1750-1850)
- Why did the General Semantics Movement Fail?
Global health and development, and as part of that, survey-making
- A review of two books on survey-making
- A glowing review of two free online MIT Global Poverty courses
- EA Mental Health Survey: Results and Analysis. (and a more recent update
I also have some minor pieces which I still want to index: