Claims

The impulse to use mathematics, statistics, and data more rather than less is a proper impulse in science, even though some people are overly mechanical in believing statistical results without applying judgment.

The inability to predict imminent events (like earthquake timing or hurricane location) is not a valid test of whether a discipline is in intellectual decline, because geologists and seismologists understand earthquakes better than anyone else does but still cannot predict when the next one will occur, and meteorologists understand hurricanes better than anyone else but don't know exactly when or where the next one will come.

Ironically, real physical sandpiles do NOT follow power-law avalanche distributions at all, violating Bak's model; Bak responded by saying 'self-organized criticality only applies to the systems it applies to,' showing indifference to whether his model matches actual sandpiles because he was interested in universal mechanisms, not literal sand.

Physics differs from social science in that measurement error in physics is generally smaller and better understood, whereas in social science measurement error is substantial and often not adequately accounted for, though science fundamentally requires quantifying uncertainties rather than ignoring them.

Reductionism—the methodology of explaining phenomena by analyzing smaller and smaller scales of spacetime to find more fundamental laws—has been useful but is now conceptually dead because spacetime itself is not fundamental; therefore claims about what is 'smaller' or 'more fundamental' in spacetime miss what is actually fundamental

Gödel's incompleteness theorem applies to scientific theories: any sufficiently sophisticated mathematical theory will have assumptions that cannot be proven from within the theory and true statements that cannot be derived from the axioms; therefore there will never be a final theory of everything and reality will always transcend any conceptual framework

When you study and learn the current scientific textbooks, you are not wasting time on settled questions; rather, you are learning the tools you need to discover new frameworks, because the current textbooks will look as silly in 50 years as 19th-century physics looks to us now

As a scientist, Dyson does not have much faith in predictions because science is organized unpredictability—scientists arrange experiments to be as unpredictable as possible and then observe what happens, making true predictions impossible from within the scientific method itself.

In response to question about prediction in science: Occasionally scientists make successful predictions (Dirac and others), but these are exceptions and not the rule; most predictions are wrong and disappear from literature; therefore successful predictions should be valued but not relied upon for understanding the future.

Regarding reductionist vs. synthetic science: Both reductionist and synthetic science are necessary; Rose's point is that the 21st century may have more synthetic science and less reductionist science, representing a shift in balance rather than the elimination of either approach.

Scientists studying climate sit in air-conditioned buildings running computer models rather than putting on winter clothes and measuring what is really happening in swamps and clouds, which is why climate experts end up believing their own models more than they trust empirical observation.