I think the most important idea of the Sequences has aged well: the idea that you should have multiple working hypotheses, instead of falling in love with only one. It’s summarized in T.C. Chamberlin’s paper and Abram’s post.
But Eliezer’s sweet coating around it, the reporting on scientific controversies, hasn’t aged so well:
The replication crisis has hit many studies that Kahneman relied on
Neural networks have outpaced neat AI
Statisticians no longer spend much time arguing Bayesianism vs frequentism
Physicists no longer spend much time arguing many-worlds vs collapse
Due to that, anyone trying to learn from the Sequences today will develop a certain archaic slant that isn’t shared by experts. Getting rid of it took me some effort and embarrassment. So if we want an updated textbook, let’s start by reporting on today’s actual scientific controversies. (We should keep the CDT vs EDT controversy though, it has worked out well for us. And of course past controversies like religion or phlogiston should stay too.)
Statisticians no longer spend much time arguing Bayesianism vs frequentism
Did they, when the Sequences were written? My impression was that the camps were well-established then, and well-established now, and the main difference has been that the Bayesians have had their tools improved by additional compute more than the Frequentists have; currently the question seems to be ‘Bayesianism’ vs. ‘pragmatism’ in much the same way that the debate in physics seems to be ‘MWI’ vs. ‘shut up and compute.’
Like, out of those four areas, I was trained in three of them before I found Less Wrong, and maybe I just had good professors / got lucky but I came in predisposed to think Eliezer’s view was sensible for all of them, but also lots of people were pragmatists because it worked out better in a social context. (The few exceptions, like a decision analysis professor who flatly insisted on Bayesian probability, were because you really couldn’t make sense of the class if you were interpreting everything as a frequentist. But all the analyses were simple enough that you didn’t really use what a statistician would call ‘Bayesian methods’ rather than just the Bayesian interpretation of probability.)
I think Eliezer’s presentation of the Bayesianism vs frequentism arguments in science came from E. T. Jaynes’ posthumous book Probability Theory: The Logic of Science, which was written about arguments that took place over Jaynes’ lifetime, well before the Sequences were written.
My recent post, Public Positions and Private Guts, is a conceptual descendant of the claim that there’s a better way of doing science than the scientific method, in that it sees the scientific method as one member of a class of methods used for discovering and communicating knowledge of different types.
But “conceptual descendant” seems important; the typology in my post is perhaps something Eliezer saw then but isn’t something he discusses in that post. (To be clear, that post is my take on Anna’s concept, and where Anna got that typology from is unknown to me; it might even be Eliezer!)
One of the reasons why academia has all those strict norms around plagiarism and citing sources is that it makes the “conceptual family tree” legible. Otherwise it just kind of becomes soupy and difficult to discern.
I think the most important idea of the Sequences has aged well: the idea that you should have multiple working hypotheses, instead of falling in love with only one. It’s summarized in T.C. Chamberlin’s paper and Abram’s post.
But Eliezer’s sweet coating around it, the reporting on scientific controversies, hasn’t aged so well:
The replication crisis has hit many studies that Kahneman relied on
Neural networks have outpaced neat AI
Statisticians no longer spend much time arguing Bayesianism vs frequentism
Physicists no longer spend much time arguing many-worlds vs collapse
Due to that, anyone trying to learn from the Sequences today will develop a certain archaic slant that isn’t shared by experts. Getting rid of it took me some effort and embarrassment. So if we want an updated textbook, let’s start by reporting on today’s actual scientific controversies. (We should keep the CDT vs EDT controversy though, it has worked out well for us. And of course past controversies like religion or phlogiston should stay too.)
Did they, when the Sequences were written? My impression was that the camps were well-established then, and well-established now, and the main difference has been that the Bayesians have had their tools improved by additional compute more than the Frequentists have; currently the question seems to be ‘Bayesianism’ vs. ‘pragmatism’ in much the same way that the debate in physics seems to be ‘MWI’ vs. ‘shut up and compute.’
Like, out of those four areas, I was trained in three of them before I found Less Wrong, and maybe I just had good professors / got lucky but I came in predisposed to think Eliezer’s view was sensible for all of them, but also lots of people were pragmatists because it worked out better in a social context. (The few exceptions, like a decision analysis professor who flatly insisted on Bayesian probability, were because you really couldn’t make sense of the class if you were interpreting everything as a frequentist. But all the analyses were simple enough that you didn’t really use what a statistician would call ‘Bayesian methods’ rather than just the Bayesian interpretation of probability.)
I think Eliezer’s presentation of the Bayesianism vs frequentism arguments in science came from E. T. Jaynes’ posthumous book Probability Theory: The Logic of Science, which was written about arguments that took place over Jaynes’ lifetime, well before the Sequences were written.
The “payoff” of of MWI, that there is a better way of doing science than the scientific method seems to have been dropped as well.
My recent post, Public Positions and Private Guts, is a conceptual descendant of the claim that there’s a better way of doing science than the scientific method, in that it sees the scientific method as one member of a class of methods used for discovering and communicating knowledge of different types.
But “conceptual descendant” seems important; the typology in my post is perhaps something Eliezer saw then but isn’t something he discusses in that post. (To be clear, that post is my take on Anna’s concept, and where Anna got that typology from is unknown to me; it might even be Eliezer!)
One of the reasons why academia has all those strict norms around plagiarism and citing sources is that it makes the “conceptual family tree” legible. Otherwise it just kind of becomes soupy and difficult to discern.