В Кембридже трансформировали теорию вероятности из изучения случайности в инструмент рационального выбора

Commercial Feature Cambridge’s great argument about probability For sixty years, a question about chance passed from one Cambridge thinker to another, until John Venn, John Maynard Keynes and Frank Ramsey had transformed probability into a theory of evidence, belief and rational choice. In the dining hall at Gonville and Caius there is a stained glass window of three overlapping circles in blue, orange and purple. It commemorates John Venn, who is known worldwide for a diagram he did not invent, and hardly at all for the book he wrote that actually mattered. Venn came up to Caius in 1853, graduated sixth Wrangler in the Mathematical Tripos of 1857, and never really left. He was elected a Fellow that same year and held the fellowship for the next sixty-six years, latterly as President of the college. He was a restless and fascinating sort. Among other things he built a machine for bowling cricket balls, good enough that when the touring Australians tried it out in Cambridge it clean bowled one of their best batsmen several times over. In his last years, working with his son, he set out to write a biographical entry for every student who had ever passed through the University. The first volume came out in 1922 carrying 76,000 names and got no further than 1751. He died the following spring. Venn located probability in repeated events But the work for which he is best remembered is The Logic of Chance. The book asked a simple question: does probability describe a pattern in the world, or merely the strength of our belief about what will happen? Venn’s answer was that probability was fundamentally a matter of frequency. To say that a coin has a one-half chance of landing heads is not, in his account, to describe some hidden property of this particular toss. It is to say that, over an indefinitely long series of tosses, heads will occur about half the time. Probability belongs to the pattern across repeated trials and not to one-off events. The mathematics behind that intuition was already old by Venn’s time. Jacob Bernoulli had proved an early form of what we now call the Law of Large Numbers, published posthumously in 1713, and Siméon-Denis Poisson gave the principle its name and a broader form in 1837. The Law of Large Numbers tells us that small samples can stray far from expectation, while large ones tend to converge toward it. Seven heads in ten tosses of a fair coin would hardly raise an eyebrow. Seven hundred thousand heads in a million tosses would be extraordinarily unlikely. The same distinction appears in games at an online casino. A player can win or lose heavily over a handful of spins, while measures such as return to player (RTP) describe what is expected across a very large number of plays, not what will happen on the next one. Venn’s approach was philosophical rather than mathematical. By then he was teaching moral science at Cambridge, where probability was treated alongside questions of logic and evidence. His concern was not primarily how to calculate a probability, but what a statement of probability actually meant. More precisely, Venn was asking what made such a statement true. If we say that the probability of heads is one half, what in the real world does that number refer to? His answer was concrete: not our confidence about the next toss, nor some invisible quantity attached to it, but the proportion of heads in the relevant series of tosses. The individual event could remain completely unpredictable. Probability emerged only when it was placed among a sufficiently large class of similar events. That gave probability a clear footing wherever such a series existed. But it also exposed the limit of Venn’s account. A single coin toss could be called fifty-fifty only because it belonged to a class of tosses in which heads and tails appeared in equal proportions. The probability did not belong to the individual toss itself. What, then, about an event that cannot meaningfully be repeated? What is the probability that this horse wins this race, that a war breaks out next year, or that a particular scientific hypothesis is true? There may be evidence pointing one way or another, but no endless series of identical cases from which a frequency can emerge. John Mayard Keynes moved probability from repeated events to evidence The next answer also came from Cambridge. In 1921, John Maynard Keynes, a Fellow of King’s College, published A Treatise on Probability. The book had been taking shape for years: an early version, Principles of Probability, had been submitted to King’s as a fellowship dissertation in 1907. Keynes shifted the question. Probability, he argued, was not fundamentally a frequency in the world but a logical relation between evidence and conclusion. Given what we know, how strongly does the evidence support what we are being asked to believe? On this account, probability was a form of inductive logic. Two people may reach different conclusions if they have different evidence. But given the same evidence, Keynes thought there was an objectively correct degree of support for a conclusion. If both know the same facts about a horse’s form, the track, and the weather, the evidence supports “this horse will win” to the same extent for both of them, even if one is more optimistic than the other. This solved a problem that Venn’s approach struggled with. A unique event did not need an endless series of repetitions before we could reason probabilistically about it. We could ask how strongly the evidence supported the proposition that this horse would win this race, even though this particular race would happen only once. Modern sports betting works in much the same setting: bookmakers and bettors assign odds to a match that will happen only once, drawing on evidence such as form, injuries, conditions, and past performance. The probability is a judgment about what the available evidence supports, not a frequency observed across endless repetitions of that exact match. But Keynes’s theory became less comfortable when one asked for a number. He did not believe that every relationship between evidence and conclusion could be expressed as a precise probability. The evidence might clearly support one outcome more strongly than another without telling us whether the correct figure was 60 percent, 70 percent, or anything else. That was not a flaw Keynes had overlooked; It followed from the theory itself. Probability, for him, was a logical relation, and logical relations did not have to come neatly numbered. Some could be compared; others might not even be rankable at all. Philosophically, that gave probability a much wider domain than Venn had allowed. Practically, however, it created a problem. A theory meant to explain rational uncertainty could often tell you that one conclusion was better supported than another while refusing to say by how much. And that left the next question almost unavoidable: if probability is supposed to guide decisions under uncertainty, how can we measure it? Frank Ramsey found a way to put a number on belief Keynes did not have to look far for his most formidable critic. Frank Ramsey was a young Fellow of King’s, barely into his twenties, who had already translated Wittgenstein’s notoriously difficult Tractatus into English, distinguished himself in mathematics, and acquired a reputation around Cambridge for seeing through difficult problems. In 1926, at the age of 23, Ramsey presented a paper to the Cambridge Moral Sciences Club called “Truth and Probability”. He began with the problem Keynes had left behind. If probability was supposed to measure how strongly the evidence supported a conclusion, what exactly was being measured? Keynes had treated those relations as objective features of logic. Ramsey was unconvinced. Ramsey proposed starting somewhere more tangible: with what people actually do when they are uncertain. Suppose you say there is a 75 percent chance that a horse will win. What does that number mean? Ramsey’s answer was to look at the bets you would be prepared to make. If you are willing to risk enough money on the horse at certain odds but refuse at others, your choices reveal something about how strongly you believe it will win. The belief need not be a feeling you can inspect directly. It can be inferred from the decisions you are willing to make. That was the crucial move. Probability could be understood as a degree of belief, and degrees of belief could, at least in principle, be measured through behaviour. This was very different from Keynes. For Keynes, two people with the same evidence faced the same objective probability relation, whether or not either recognized it correctly. For Ramsey, probability belonged to the person doing the reasoning. Two people presented with the same evidence could assign different probabilities because they brought different beliefs, experiences, and judgments to it. Ramsey’s next step was to ask what it would mean for a set of beliefs to be internally coherent. Betting again provided the test. Imagine that your beliefs assign probabilities in a way that violates the ordinary rules of probability. Perhaps your judgments about several related events do not quite fit together. Taken one at a time, each may seem harmless. But Ramsey showed that inconsistency has a price. A sufficiently clever opponent could offer you a collection of wagers, each acceptable according to your own stated beliefs, whose combined result guaranteed that you lost money whatever happened. In other words, if your degrees of belief do not fit together according to the mathematics of probability, someone can make a book against you. The idea later became known as the Dutch Book Argument. Its force came from turning an abstract mathematical requirement into a practical test of rationality. The rules of probability were no longer simply conventions for manipulating numbers. Violating them meant holding beliefs that could be exploited against one another. Ramsey had therefore supplied something Keynes had struggled to provide. Probability could be numerical even when the event would happen only once. There did not need to be an infinite series of identical horse races, elections, wars, or business decisions. What mattered was the strength of a person’s belief and whether that system of beliefs was coherent. Then Ramsey connected belief to action. That connection is what made the idea so powerful. Once probabilities could be inferred from choices, they could be combined with the value a person placed on different outcomes. A 50 percent chance of winning £10 is not the same decision as a 50 percent chance of losing your house, even though the probability is identical. Rational choice therefore required two things: beliefs about what might happen and preferences over what those outcomes would be worth. From that pairing grew much of modern decision theory. The same basic structure appears whenever someone must choose under uncertainty: an insurer pricing a policy, an investor comparing risks, a doctor weighing treatments, or a machine-learning system ranking possible predictions. Assign probabilities to uncertain outcomes, assign values to what those outcomes mean, and choose accordingly. Ramsey did not live to see how far the idea would travel. He died in January 1930 at the age of 26. “Truth and Probability” had never been published during his lifetime; it appeared the following year in a posthumous collection of his papers. Its influence grew slowly. Other thinkers, most notably Bruno de Finetti and later Leonard Savage, developed subjective probability and decision theory into much more systematic forms. But Ramsey had already supplied many of the essential pieces: probability as degree of belief, betting as a way of measuring it, and coherence as a requirement of rational thought. In the space of little more than half a century, Cambridge had moved probability through three different worlds. For Venn, it belonged to the pattern that emerged across repeated events. For Keynes, it described the strength of the evidence for a conclusion. For Ramsey, it became something a person carried into a decision and revealed through the choices he was prepared to make. What began as a question about chance had become a theory of rational belief — and, with Ramsey, a way of putting a number on what a belief is worth. - News / University to wind down Cambridge Farm Animal Veterinary Services26 August 2026 - Comment / ‘Man of the people’: is a Cambridge educated prime minister really that surprising?  26 August 2026 - News / Eddington water system remains unused despite government pledge to switch it on this year27 August 2026 - News / John’s granted permission for new student accommodation despite climate concerns24 August 2026 - Music / Unemployed in summertime? Here’s your soundtrack18 August 2026