Michael Nielsen – Why aliens will have a different tech stack than us

Summary

Dwarkesh Patel interviews Michael Nielsen about how science recognizes progress despite long, ambiguous, and sometimes hostile verification loops. The discussion covers historical cases such as Michelson-Morley, Copernicus, Darwin, Prout, and relativity, and explores what this means for AI-driven scientific discovery. Nielsen argues that science cannot be reduced to a simple method, that many bottlenecks are non-verifiable, and that alien civilizations may develop very different technology stacks.

• Explores why scientific progress is hard to reduce to a simple falsification or verification method.
• Uses Michelson-Morley, Copernicus, Darwin, and Prout to show how verification loops can be long or hostile.
• Argues AI may help with certain science bottlenecks but not automatically with all of them.
• Frames AlphaFold as primarily a data acquisition success rather than a deep explanation.
• Speculates alien civilizations may follow very different science and technology trees.
• Discusses potential gains from trade between different civilizations and limits of comparative advantage.
• Covers open science, collective discovery, personal learning methods, and quantum computing speculation.