Pessimistic Meta-Induction and Its Limits: Lessons from Frequentist Statistics and Machine Learning Theory

Machine Demons

This essay argues that enthusiasm for ML as a turning point rests on a category confusion between two distinct scientific imaginaries: Laplace’s deterministic universe and Comte’s social physics, which sought descriptive regularities rather than causal mechanisms.

Induction and the rule of succession through a possibilistic inferential model lens

Induction is the process by which empirical evidence is transformed to knowledge. Hume famously argued---and Popper and others agree---that there can be no logical justification for induction. A weaker form of induction, due to Bayes, expresses the aforementioned knowledge in terms of probabilities, and we review some...

Understanding Generalization Requires Universal Induction

Solomonoff induction is relativized to an information vantage point, biasing toward short programs with access to all preexisting information, which reframes the inductive bias: instead of seeking some absolute notion of simplicity, it favor accessibility with respect to the authors' vantage point.

G¨odel’s Incompleteness Theorems: The Plaintext Distillation

Enduring Strengths of Knowledge-Based Methods in the Era of Machine Learning and Large Language Models: A Task-and-Architecture Perspective

Large language models have shifted AI toward statistical learning, but knowledge-based methods remain essential for tasks governed by combinatorial structure, declarative correctness, strong domain priors, and auditable reasoning. This paper treats the issue as one of task–architecture fit rather than paradigm competit...

Reopening the Dutch Book

Probabilism is the thesis that the degrees of belief of rational agents behave like Kolmogorovian probabilities. There are quite different lines of argument for probabilism in the literature, including Dutch book arguments, arguments based on representation theorems, and epistemic utility arguments. This paper focuss...