Philosophy of Science
The Flaw at the Heart of Every Experiment Ever Run
Every scientific conclusion ever drawn — every law of physics, every medical breakthrough, every climate model — rests on a logical move that cannot actually be justified.
The Idea
Here is the move science makes, millions of times a day: something happened repeatedly in the past, therefore it will happen again. The sun rose yesterday, and the day before, and for four billion years before that — so it will rise tomorrow. Drop a ball, it falls. Drop it again, it falls again. After enough repetitions, we call it gravity and move on. This inference from observed cases to a general rule is called induction, and the Scottish philosopher David Hume noticed, in the 1740s, that it has no logical foundation whatsoever. To justify induction, you'd have to argue: 'induction has worked in the past, therefore it will work in the future.' But that is itself an inductive argument. The justification swallows itself. This isn't a technicality. It means science's entire method of building knowledge — observe, generalise, predict — cannot be proven valid without circular reasoning. Karl Popper, in the twentieth century, tried to sidestep this by reframing science as falsification rather than confirmation: we can never prove a theory true, only fail to prove it false. But this doesn't dissolve the problem so much as relocate it. Even choosing which experiments to run, and trusting that past failures of a theory are evidence against it, still leans on inductive assumptions. The problem of induction isn't a crack in the foundations of science. It might be the foundations.
In the World
In 1998, two independent teams of astronomers were measuring distant supernovae to settle a fairly straightforward question: at what rate is the universe's expansion slowing down? Everyone assumed it was slowing — gravity pulls things together, after all, and gravity had always pulled things together. The teams' data, however, showed the expansion was accelerating. The universe is not just expanding; it is expanding faster and faster, driven by something now called dark energy, which nobody can explain and which constitutes roughly 68 percent of everything that exists. What makes this story philosophically interesting isn't just the surprise — it's what the surprise reveals. Every prior observation of gravity had confirmed the same rule. Billions of data points, centuries of successful prediction, the entire architecture of Newtonian and Einsteinian physics — all of it built on the reasonable assumption that what had always been true would keep being true. And then, at cosmological scales, it wasn't. The scientists had not made an error. Their instruments were not faulty. The universe had simply declined to behave as the accumulated past suggested it should. Hume, dead for two centuries by then, would not have been surprised. He would have said: the past never had any obligation to resemble the future. We just assumed it did, because assuming otherwise is nearly impossible to live with — and because, most of the time, it works.
Why It Matters
You don't need to be doing cosmology for this to be useful. The problem of induction is live every time you trust that the bridge will hold, that the medication will work the way the trial suggested, that your experience of a person tells you who they are. These are not irrational bets — they are sensible ones. But Hume's insight is a healthy corrective to the confidence with which we mistake pattern for certainty. Science's strength is not that it has solved the problem of induction. It hasn't. Its strength is that it has built a self-correcting practice around the fact that any of its conclusions might be wrong — and that the right response to that is not paralysis, but rigorous, humble updating. The next time someone tells you that science has 'proven' something, the philosophically honest answer is that science has provided compelling, so-far-undefeated evidence for something. That is not a weakness. It is a different and more honest kind of strength than proof.
A Question to Ponder
Is there anything you believe to be true — not just probably true, but certain — that doesn't ultimately rest on the assumption that the past resembles the future?
Get a new one of these every morning.
Start learning with Thinkable