Epistemology: The Problem of Induction
Why a Million Sunrises Still Can't Promise Tomorrow's
Every scientific law you've ever trusted is built on a logical gap that no one has ever managed to close.
The Idea
Here is the quiet crack running through the foundation of everything we think we know. We learn from experience — that's the obvious bit. Touch a hot stove, learn about heat. Watch the sun rise ten thousand times, assume it will rise again tomorrow. This process, reasoning from repeated observation to a general rule, is called induction. It underpins science, medicine, engineering, and most of the judgements you'll make before lunch. The Scottish philosopher David Hume noticed, in the mid-18th century, that induction can never actually be logically justified. To say 'the sun will rise tomorrow because it always has' is to assume that the future will resemble the past. But what justifies that assumption? Only past experience — which means you're using induction to justify induction. The argument eats itself. This isn't a technicality. It means that no amount of confirming evidence can ever fully prove a universal claim. A thousand white swans don't prove all swans are white; one black swan in Australia demolished that assumption in an afternoon. Karl Popper took this seriously enough to redesign the scientific method around it — science doesn't confirm theories, it tries to falsify them, because that's the only logical move available. What Hume exposed isn't a flaw in human thinking so much as a structural feature of knowledge itself: we are always, at some level, making a bet. The past is the only evidence we have, and the future owes us nothing.
In the World
In the years before the 2008 financial crisis, the world's most sophisticated risk models were running on historical data going back a few decades. The models saw no precedent for a simultaneous collapse across multiple housing markets, so they assigned that scenario a probability close to zero. The banks, the ratings agencies, the regulators — all of them were, in a sense, the inductivist who had never seen a black swan. Nassim Nicholas Taleb had been making exactly this argument for years. In his 2007 book 'The Black Swan,' written in the months before the crash, he argued that financial models are catastrophically overconfident because they mistake 'hasn't happened in our data set' for 'won't happen.' The models weren't bad statistics — they were a near-perfect example of the problem of induction dressed up in spreadsheets. But the most vivid version of this story belongs to Bertrand Russell, who gave it to us decades earlier in the form of a chicken. Every morning of its life, the chicken has been fed by the farmer. Each morning confirms the rule: farmer equals food, farmer equals safety. Then one day, the farmer comes — and wrings its neck. Russell's point was not to be cruel to chickens. It was that the most dangerous moment can be the one when your inductive confidence is at its peak, when the pattern feels most solid, most earned, most obviously true.
Why It Matters
Most of us walk through the day treating our assumptions as if they were facts. The project will probably go well — it usually does. My health is probably fine — nothing has gone wrong yet. This person is probably trustworthy — they've been reliable before. These aren't delusions; they're reasonable bets. But Hume's insight invites a subtle, genuinely useful shift: holding your expectations a little more lightly, not out of anxiety, but out of intellectual honesty. There is something almost meditative about this. Recognising that you cannot fully justify your beliefs about the future doesn't mean abandoning those beliefs — it means carrying them differently. You act on the best evidence you have while staying genuinely open to being wrong. This is, incidentally, what good scientists do, and what thoughtful people have always done: not paralysis, but calibrated humility. The question the problem of induction quietly asks is whether you can tell the difference between a pattern that will hold and one that is about to break — and whether you've been honest with yourself about which kind you're currently betting on.
A Question to Ponder
Which belief of yours feels most solidly proven by experience — and what would it take for that belief to be the chicken's?
Get a new one of these every morning.
Start learning with Thinkable