Completely | Science

In a completely science framework, a result must be robust. However, researchers can torture data until it confesses. By running 20 statistical tests on random noise, one will appear "significant" (p < 0.05) by pure chance. If a study does not correct for multiple comparisons, it is statistically fraudulent—not complete science.

By the 1970s, the hypothesis that CFCs were destroying stratospheric ozone was mounting. By 1985, the Antarctic ozone hole was directly observed. The science was complete enough that 197 nations banned CFCs. Today, the ozone layer is healing. That is completely science at work. completely science

Philosophers of science, from Hume to Kuhn, have noted a problem: induction. No number of white swans proves “all swans are white.” Similarly, gravitational theory has been confirmed billions of times, but we cannot prove it will hold tomorrow with 100% certainty. That is the problem of induction. In a completely science framework, a result must be robust

Thus, “completely science” does not mean “absolutely certain forever.” Instead, it means: Given all available evidence, using the most rigorous

Given all available evidence, using the most rigorous methods we currently possess, no competing explanation fits better, and the theory has survived every serious attempt to falsify it.

This is sometimes called “scientific realism.” It is the most honest kind of certainty humans can achieve. And it works—spectacularly well. Computers, vaccines, rockets, and GPS all depend on knowledge that is completely science.