His thesis still irritates me. Lukeprog claims philosophers are doing shoddy work, and he can e.g. solve meta-ethics all by himself. He starts writing his meta-ethics sequence and it has just the basic intro stuff, but nonetheless since he claimed he could solve it, it gets promoted to one of the few curated sequences on Less Wrong. And then he just...stops, he never gets even close to solving meta-ethics and it remains in the Less Wrong curated sequences. It’s been 6 years since the last post Lukeprog, where is the solution to meta-ethics?
My early intellectual influences were rationalists or free-market leaning economists, such as Scott Alexander and Robin Hanson. When I take a sincere look at the evidence today and try my very hardest to discern what is actually true from false, I conclude they mostly are getting things right.
I think the difference with star-craft is that with altruistic interventions there is less of a blatant self interest to believe a certain thing, which means it’s easier to shift over time. I started out also reading Scott Alexander etc (and still do) but over time also added academic texts and my beliefs shifted to the left, even though I don’t have a strong self interested reason to believe in either of them. If academia was also split into equal factions like the star-craft community we would have to be more worried that interventions from each faction would cancel each other out (like your pro-charter city anti-charter city example), instead we see that academia mostly converges on leftwing ideas, even across different countries, different demographics and different generations. If everyone switches to one ‘concession’ intervention the smaller group (in this case the rightwing) benefits more.
Maybe instead we can match people up, so the large general group has an equal number of people donating to ‘concession’ interventions as there are in the non-general group, who all give to ‘concession’. If there is a lot of convergence there is little concession, if academia is almost perfectly split there’s almost only concessions.
The real problem is that people like you, me, and academics (researchers) are not really the people with the power to decide which interventions get money. That’s mostly decided by non-researchers who spend their lives focussing on accruing political or economic capital, rather than knowledge about altruistic interventions. We can make cooperation mechanisms to match our donations all we want, but a random billionaire can just decide to donate billions of dollars, more money than we will ever see, to the museum next to his house. It may be wise to do this kind of monetary coordination at the federal level aka taxes and subsidies.
To minimize the chance of statistical noise or incorrect inference polluting our conclusions, we create experiments with randomly chosen intervention and control groups, so we are sure the intervention is causally connected to the outcome.
The problem with RCT’s is that they are expensive and measure narrow, direct, continuous effects, while they’re unpractical for broad, indirect, or discontinuous effects. Which means those interventions privilege the status-quo. I don’t see why we should limit ourselves to only randomized controlled trials. Case-control studies are cheaper and can be done by individual researchers or small teams. The fact that they have less statistical power is irrelevant when we have literally zero studies about so many things. The same thing can be said for cohort studies. If we find an intervention with orders of magnitude more impact than others through a cohort study, I’m not going to wait for an RCT (If that would even happen because in practice not everything is measurable with RCTs) before I start donating.
Good post.
His thesis still irritates me. Lukeprog claims philosophers are doing shoddy work, and he can e.g. solve meta-ethics all by himself. He starts writing his meta-ethics sequence and it has just the basic intro stuff, but nonetheless since he claimed he could solve it, it gets promoted to one of the few curated sequences on Less Wrong. And then he just...stops, he never gets even close to solving meta-ethics and it remains in the Less Wrong curated sequences. It’s been 6 years since the last post Lukeprog, where is the solution to meta-ethics?
I think the difference with star-craft is that with altruistic interventions there is less of a blatant self interest to believe a certain thing, which means it’s easier to shift over time. I started out also reading Scott Alexander etc (and still do) but over time also added academic texts and my beliefs shifted to the left, even though I don’t have a strong self interested reason to believe in either of them. If academia was also split into equal factions like the star-craft community we would have to be more worried that interventions from each faction would cancel each other out (like your pro-charter city anti-charter city example), instead we see that academia mostly converges on leftwing ideas, even across different countries, different demographics and different generations. If everyone switches to one ‘concession’ intervention the smaller group (in this case the rightwing) benefits more.
Maybe instead we can match people up, so the large general group has an equal number of people donating to ‘concession’ interventions as there are in the non-general group, who all give to ‘concession’. If there is a lot of convergence there is little concession, if academia is almost perfectly split there’s almost only concessions.
The real problem is that people like you, me, and academics (researchers) are not really the people with the power to decide which interventions get money. That’s mostly decided by non-researchers who spend their lives focussing on accruing political or economic capital, rather than knowledge about altruistic interventions. We can make cooperation mechanisms to match our donations all we want, but a random billionaire can just decide to donate billions of dollars, more money than we will ever see, to the museum next to his house. It may be wise to do this kind of monetary coordination at the federal level aka taxes and subsidies.
The problem with RCT’s is that they are expensive and measure narrow, direct, continuous effects, while they’re unpractical for broad, indirect, or discontinuous effects. Which means those interventions privilege the status-quo. I don’t see why we should limit ourselves to only randomized controlled trials. Case-control studies are cheaper and can be done by individual researchers or small teams. The fact that they have less statistical power is irrelevant when we have literally zero studies about so many things. The same thing can be said for cohort studies. If we find an intervention with orders of magnitude more impact than others through a cohort study, I’m not going to wait for an RCT (If that would even happen because in practice not everything is measurable with RCTs) before I start donating.