What is error, and what is scientific error? Douglas Allchin explores the various types of scientific errors, how to identify them, and how to do science in light of them.

My links: https://linktr.ee/frictionphilosophy.

1. Author

Douglas Allchin is an AAAS Fellow and Resident Fellow at the Minnesota Center for the Philosophy of Science, and his work has primarily focused on the history and philosophy of science.

Check out his book, "Toward a Philosophy of Error in Science"!

https://global.oup.com/academic/product/toward-a-philosophy-of-error-in-science-9780197827673

https://a.co/d/iobiDIc

2. Book Summary

Douglas Allchin’s Toward a Philosophy of Error in Science argues that scientific error shouldn’t be treated as an embarrassing sideshow to “real” science, but as something integral to how science actually learns and progresses. Instead of assuming that good methods straightforwardly yield reliable knowledge, Allchin urges a systematic “philosophy of error” that tracks how a claim can be justified at one time and later become unjustified—i.e., how changes in evidence, framing, and reasoning can overturn what once looked reasonable.

The book develops an “inventory” of error types across three layers of scientific justification. At the observational layer, errors can stem from material contamination, instrument problems, sampling and measurement misframing (like small samples, proxies, or confounders), and observer effects and biases. At the conceptual layer, mistakes arise in inference and interpretation—overgeneralization, faulty assumptions, confirmation bias, and culturally inflected biases, alongside a meta-risk Allchin calls “epistemic hubris” (the idea that these pitfalls only happen to other scientists). At the social layer, scientific discourse and institutions can also entrench errors (through weak vetting, communal biases, or distorted incentives), even though—ideally—organized skepticism and reciprocal criticism are supposed to help filter mistakes.

Finally, Allchin focuses on how errors are actually found and remedied: they don’t “announce themselves,” and there’s no single ‘error-correction method’—correction can be slow, uneven, and sometimes driven by contingencies rather than a tidy mechanism. Against the comforting slogan that science is simply ‘self-correcting,’ he argues we should be more explicit about when and how peer review and replication succeed or fail, and then manage error more deliberately. A key payoff is rethinking what counts as epistemic progress: “negative knowledge” (learning what’s not the case, and why) is still genuine knowledge, and improving reliability often means actively probing for hidden sources of error rather than only accumulating confirming evidence.

3. Interview Chapters

00:00 - Introduction

00:56 - Overview of book

02:12 - Error09:08 - Uncertainty

11:42 - Epistemology

13:33 - Vagueness

17:38 - First layer of error: raw data

29:30 - Second layer of error: conceptual

50:25 - Third layer of error: social

1:10:46 - Recognizing error

1:22:34 - Resolving error

1:26:10 - Humans and history

1:29:18 - Useful biases

1:36:03 - Negative knowledge

1:41:49 - Pessimistic meta-induction

1:47:42 - Value of philosophy

1:50:23 - Conclusion

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit fric.substack.com/subscribe