You’ve probably been told that probability is the science of uncertainty. You’ve probably used it to make massive business decisions, trade stocks, or build AI models. But here is the dirty secret: the math of probability doesn’t care about reality, and true “randomness” doesn’t actually exist.
Probability isn’t a window into the future; it’s a mirror reflecting your own assumptions.
For centuries, probability was a messy playground for gamblers and statisticians. Then, in 1933, a Soviet mathematician named Andrey Kolmogorov wrote a thin little book that ended the chaos. He chained probability to three brutal, unbreakable rules. He didn’t just organize the math; he trapped it in a cage.
Kolmogorov proved that probability is just a trio of concepts: a sample space, a list of events, and a measure (P). It’s an axiomatic system, exactly like Euclidean geometry. As David Hilbert famously said about geometry, you could replace the words “points, lines, planes” with “tables, chairs, beer mugs,” and as long as they follow the rules, the math stays exactly the same.
But here is the twist that breaks your brain.
What is a “random variable”? It isn’t random. It is a completely deterministic function. If you input the exact same state of the universe, it spits out the exact same number. The “randomness” isn’t in the math. It’s shoved entirely into P—the probability measure. And where does P come from? Not from the math.
Mathematics doesn’t do randomness. It just builds the cage for your uncertainty to rattle around in.
The math is perfectly strict, but it refuses to tell you what P actually *is* in the real world. Because the math stays silent, humans rushed in with their own beliefs. Today, you have warring factions: Frequentists who swear P is the limit of infinite repeated trials. Bayesians who insist P is a rational degree of belief. Propensity theorists who claim P is a physical tendency.
They use the exact same formulas. They prove the exact same theorems. But they are living in completely different universes.
The formula doesn’t change, but the worldview does. That’s why a statistician and a gambler can look at the same 30% and mean entirely different things.
If you are consuming data, trading algorithms, or building AI, you need to wake up. The math won’t save you. The math is just a vehicle. You are the one supplying the fuel. Stop treating probability models as objective truth, and start questioning where the P actually came from. Symmetry? Historical data? Or just someone’s gut feeling?
The universe doesn’t care about your need for certainty. It will remain stubbornly uncertain. The only question is whether you’ll be a blind consumer of models, or a clear-eyed builder who knows exactly where the math ends and the belief begins.
FAQ
Q: If the math is just axioms, how can we trust any data model?
A: You can trust the math to be internally consistent, but you cannot trust it to reflect reality. A model is only as good as the assumptions you inject into P. Garbage in, absolute certainty out.
Q: What's the practical implication of knowing this?
A: Stop treating probabilistic outputs as objective facts. When an AI gives you a 90% confidence score, that 90% is heavily dependent on whether the creator was a Bayesian or a Frequentist. Question the measure, not just the output.
Q: What's the contrarian take?
A: Kolmogorov didn't solve probability; he just quarantined the philosophy so mathematicians wouldn't have to deal with it. He swept the human element under the rug so the math could look clean.