Charles Mills has argued against ideal theory in political philosophy on the basis that it contains idealizations. He calls for political philosophers to do more nonideal theory, namely political theory that pays more attention to the most visible oppressions in society, such as those based on race, gender, and class. Mills's argument relies on a distinction between idealization and abstraction. Idealizations involve adding false assumptions to one's model, which is unacceptable, whereas abstractions merely leave out details without undermining descriptive power. (...) By studying formal models of injustice, I argue that the idealization/abstraction distinction is unhelpful. Either the distinction exists only relative to one's modeling purposes, or all models in political theory contain idealizations. Either way, the distinction does not help Mills's cause. Furthermore, there are arguments from philosophy of science for the epistemic benefits of idealizations. However, Mills's call for greater emphasis on the most visible mechanisms of oppression can be supported without relying on an idealization/abstraction distinction. I provide three alternative reasons for why we should prefer political theories that place more emphasis on race-, class-, and gender-based oppression. (shrink)

Constructive field theory aims to rigorously construct concrete, nontrivial solutions to Lagrangians used in particle physics. I examine the relationship of solutions in constructive field theory to both axiomatic and Lagrangian quantum field theory. I argue that Lagrangian QFT provides conditions for what counts as a successful constructive solution and other information that guides constructive field theorists to solutions. Solutions matter because they describe the behavior of QFT systems and thus what QFT says the world is like. Constructive field theory (...) clarifies existing disputes about which parts of QFT are philosophically relevant and how rigor relates to these disputes. (shrink)

The ML community recognizes the importance of anticipating and mitigating the potential negative impacts of benchmark research. In this position paper, we argue that more attention needs to be paid to areas of ethical risk that lie at the technical and scientific core of ML benchmarks. We identify overlooked structural similarities between human IQ and ML benchmarks. Human intelligence and ML benchmarks share similarities in setting standards for describing, evaluating and comparing performance on tasks relevant to intelligence. This enables us (...) to unlock lessons from feminist philosophy of science scholarship that need to be considered by the ML benchmark community. Finally, we outline practical recommendations for benchmark research ethics and ethics review. (shrink)