The $P > 0.5$ Condition: Why Condorcet's Jury Theorem Proves That a Larger, Less-Informed Group Is More Likely to Be Correct Than a Smaller, Expert One

Paris, 1785. The Marquis de Condorcet, a mathematician and philosopher of the French Enlightenment, published an essay on the probability of decisions made by majority vote. He was trying to solve a structural problem of the impending revolution: how could a nation trust the uneducated masses to govern themselves? Condorcet did not answer with philosophy; he answered with probability. He demonstrated that if each voter in a group is even slightly more likely to be right than wrong, adding more voters pushes the group's overall chance of making the correct decision toward absolute certainty.[1][9]

The math behind Condorcet's Jury Theorem is brutally elegant. As the Stanford Encyclopedia of Philosophy defines the core premise, the theorem requires that "each voter is more likely to be correct than incorrect"—the foundational $p > 0.5$ condition. Imagine asking a single expert to diagnose a complex systemic issue; they might be highly accurate, perhaps getting it right 90 percent of the time ($p=0.9$). But if you ask 10,000 laypeople who are only marginally better than a coin flip—say, right 52 percent of the time ($p=0.52$)—the law of large numbers takes over.[1][8]

To understand the scale of this effect, consider the numerical divergence between the two groups. In a 9-member panel of experts where each has a 90 percent accuracy rate, the group's collective accuracy peaks at roughly 99.1 percent. However, for the crowd of 10,000 laypeople operating at a mere 52 percent individual accuracy, the sheer volume of independent judgments causes the errors on the high side to cancel out the errors on the low side. The probability of the majority being correct in that crowd reaches 99.9 percent. Scale, Condorcet proved, systematically dominates individual expertise once the threshold is cleared.[8][9]

But the theorem contains a devastating trapdoor. The exact same mathematical engine runs in reverse if $p$ drops below the 0.5 threshold. If the crowd is systematically misinformed, biased, or panicked, such that their individual chance of being right is 49 percent, then adding more voters mathematically guarantees failure. A large crowd with $p < 0.5$ will asymptotically approach a zero percent chance of making the correct decision. This is why the theorem serves simultaneously as the ultimate mathematical defense of democracy and a stark warning about the necessity of public education.[2][5]

The most vulnerable load-bearing pillar in Condorcet's model is the assumption of independence. The theorem requires that each voter makes up their own mind without simply copying their neighbor. As researchers in the European Political Science Review note, the epistemic value of a representative group relies heavily on cognitive diversity. If 10,000 people all watch the same flawed television broadcast and vote based on its singular conclusion, the system does not possess 10,000 independent variables; it effectively has an $N$ of 1.[2][6]

When voters correlate their errors, the mathematical advantage of the crowd collapses instantly. This vulnerability explains why peer review systems and legal juries go to such extraordinary lengths to isolate participants. The British Journal for the Philosophy of Science highlights that applying jury theorems to academic peer review only works if the reviewers evaluate the material independently rather than forming a consensus before voting. The moment independence is breached, the $p > 0.5$ condition is compromised.[3][4]

Today, this 18th-century theorem governs the architecture of modern decentralized systems. It is the reason why ensemble machine learning models, which aggregate the predictions of hundreds of weak algorithms, routinely outperform single, highly tuned neural networks. It explains why prediction markets can forecast election results more accurately than isolated political pundits. In every case, the system designer's primary job is not to make the individual nodes brilliant, but simply to ensure they remain independent and clear the 50 percent threshold.[7][9]

The tension between the crowd and the expert is not a philosophical debate; it is an optimization problem. A small panel of experts is highly efficient and resistant to mass hysteria, making it ideal for situations where $p$ for the general public is genuinely below 0.5—such as diagnosing a rare disease or engineering a bridge. But for complex, multi-variable societal problems where no single expert can process all the information, a large, moderately informed crowd with independent viewpoints will mathematically arrive at the correct answer more reliably than any genius.[5][9]

The enduring brilliance of Condorcet's Jury Theorem is that it strips the romance out of collective intelligence and replaces it with a formula. It proves that a functioning decentralized system does not require a population of polymaths. It only requires a framework that protects the independence of its participants and ensures they are just a fraction of a percent better than random chance. The mathematics of probability handle the rest.[1][9]