I think itâs a bit misleading to say EA philosophy âlacks rigorâ, because it could be taken to imply it falls below some sort of known disciplinary standard of reasoning/âevidence that at least some other philosophy reaches. I donât think this is even close to being true. EA philosophy to me means mostly âBostrom, Ord and MacAskillâs academic papers, and stuff that came out of the Global Priorities Instituteâ. And that stuff has been published in very good journals over and over again. Even MIRIâs unorthodox ideas about decision theory have been written-up and published in a very good philosophy journal! EA philosophy is about as academically mainstream as philosophy gets. Itâs true a large majority of academic philosophers disagree with at least some of it, but that is also true of any comparable rival body of philosophical work.
Great comment David! It made me focus in on the heart of the question here. It is simply thisâWhat is the right counterfactual to EA, the philosophy/âacademic discipline? OP is comparing EA philosophy/âdiscipline to Econ. But is that fair? When I read your comment this morning, I noticed how I utterly failed to clarify that EA is less rigorous⌠than what?! It got me thinking empirically and I quickly whipped up some stuff using OpenAlex which is a open-source repository of publications used by bibliometricians. Now this preliminary analysis I show doesnât resolve the question of what is counterfactual to EA, but it begins to describe the problem better.
If youâre interested, see my GitHub repo with more details on the research design that I imagined and the Python/âR code. I wonder if OP will like this because Iâm doing stuff that a (design-based) econometrician would do :-)
First up, lets validate what David said makes senseâAre EA publications in top journals?
Yep. David is not wrong. But of course that is not the question! The claim I want to make is that EA publications are more insular than other stuff and the obstacle to making this claim is what the hell is âother stuffâ?! This is where OpenAlex topics come in. OpenAlex uses a clsutering+classification pipeline to classify papers as belonging to some topic. Here is a plot showing that:
Now, the next step would be to ask ourselves, âEA has such a spread of topics. But field X has a much wider spread. This is why EA is insularâ But what is that field X? EA compared to Deontology? Utilitarianism? These have around for decadesâhow is that a fair comparison group? What exactly is the benchmark to weigh EA, the discipline, against?
Now maybe the way to do this, is to pull out all the papers in these topics I have plotted above from OpenAlex and compare against those. But I guess a better way to do this would be to pull out abstracts of all publications, clean up, tokenize, cluster and see whats close by and compare against that. Can someone else make this pipeline more concrete?
I think itâs a bit misleading to say EA philosophy âlacks rigorâ, because it could be taken to imply it falls below some sort of known disciplinary standard of reasoning/âevidence that at least some other philosophy reaches. I donât think this is even close to being true. EA philosophy to me means mostly âBostrom, Ord and MacAskillâs academic papers, and stuff that came out of the Global Priorities Instituteâ. And that stuff has been published in very good journals over and over again. Even MIRIâs unorthodox ideas about decision theory have been written-up and published in a very good philosophy journal! EA philosophy is about as academically mainstream as philosophy gets. Itâs true a large majority of academic philosophers disagree with at least some of it, but that is also true of any comparable rival body of philosophical work.
Great comment David! It made me focus in on the heart of the question here. It is simply thisâWhat is the right counterfactual to EA, the philosophy/âacademic discipline? OP is comparing EA philosophy/âdiscipline to Econ. But is that fair? When I read your comment this morning, I noticed how I utterly failed to clarify that EA is less rigorous⌠than what?! It got me thinking empirically and I quickly whipped up some stuff using OpenAlex which is a open-source repository of publications used by bibliometricians. Now this preliminary analysis I show doesnât resolve the question of what is counterfactual to EA, but it begins to describe the problem better.
If youâre interested, see my GitHub repo with more details on the research design that I imagined and the Python/âR code. I wonder if OP will like this because Iâm doing stuff that a (design-based) econometrician would do :-)
First up, lets validate what David said makes senseâAre EA publications in top journals?
Yep. David is not wrong. But of course that is not the question! The claim I want to make is that EA publications are more insular than other stuff and the obstacle to making this claim is what the hell is âother stuffâ?! This is where OpenAlex topics come in. OpenAlex uses a clsutering+classification pipeline to classify papers as belonging to some topic. Here is a plot showing that:
Now, the next step would be to ask ourselves, âEA has such a spread of topics. But field X has a much wider spread. This is why EA is insularâ But what is that field X? EA compared to Deontology? Utilitarianism? These have around for decadesâhow is that a fair comparison group? What exactly is the benchmark to weigh EA, the discipline, against?
Now maybe the way to do this, is to pull out all the papers in these topics I have plotted above from OpenAlex and compare against those. But I guess a better way to do this would be to pull out abstracts of all publications, clean up, tokenize, cluster and see whats close by and compare against that. Can someone else make this pipeline more concrete?
EDIT: Fixed broken links