Critiquing poor ML papers, recommending classics of science philosophy
fleetwood___ · x · 2026-08-25
After reading too many low-quality ML papers, the author turned to the philosophy of science for guidance and recommended three key texts: 1. Platt's 'Strong Inference' on why some fields advance faster; 2. Jonas & Kording's 'Could a neuroscientist understand a microprocessor', critiquing correlation analysis via microprocessors; 3. Tarantola's 'Popper, Bayes and the Inverse Problem'. These offer profound insights into causal inference and empirical methodology in ML research.
More from Research
- Open Source AI Eval Skills Guide Coding Agents — petergyang · 2026-08-25
- Critique of LoCoMo and LongMemEval: Seeking better memory benchmarks — LowDistribution3995 · 2026-08-25
- Standard Machines launches to teach AI advanced chip design via RL — ycombinator · 2026-08-25
- SWE-Bench ProMax Benchmark Targets Large-Scale Code Refactoring — rseroter · 2026-08-25
- Optima now supports benchmarking Qwen3.8 27B — ArtificialAnlys · 2026-08-25
- PhysCaP: Physics-Informed Agent for Robotic Manipulation — Chen-Yu Lin · 2026-08-25