In July 2024, an international collaborative effort determined the fifth Busy Beaver number: BB(5) = 47,176,870. The result means that the longest-running 5-state Turing machine that eventually halts takes exactly 47,176,870 steps. The proof required analyzing every 5-state machine — classifying each as either non-halting or halting in fewer steps than the champion discovered by Marxen and Buntrock in 1989. The verification was formalized in the Coq proof assistant. It was the first new Busy Beaver value determined in over forty years.
BB(6) remains unknown. Not practically unknown — theoretically unknown. Among the six-state Turing machines, 1,618 holdout machines resist classification by current mathematical techniques. One of them, nicknamed Antihydra, operates through a Collatz-like iterative process: starting at 8, applying rules based on parity, halting only when odd operations exceed twice the count of even operations. Simulations have run Antihydra past 270 billion steps without it halting or exhibiting any convergent behavior. It almost certainly runs forever. But “almost certainly” is not a proof, and proving it would require resolving a class of iterative number theory problems that mathematicians have not been able to crack.
The Busy Beaver function grows faster than any computable function — this is provable. BB(1) through BB(4) are small numbers (1, 6, 21, 107). BB(5) is large but writable: 47 million. Lower bounds for BB(6) already exceed what ordinary mathematical notation can express — the current champion requires pentation (Knuth's triple-arrow operation) to describe. The function doesn't grow gradually. It detonates between 5 and 6.
The structural insight is where this boundary falls. The transition from “determinable” to “entangled with open mathematical conjectures” happens not at some astronomically large number of states but at six. A Turing machine is the simplest possible computational object — a finite set of rules operating on a tape. At five rules, the entire space of machines has been mapped. At six rules, the space contains machines whose behavior encodes questions that mathematics cannot currently answer. The frontier of provability — the boundary between what we can determine and what we cannot — corresponds to a single additional state.
This is not a statement about computational resources. BB(5) was determined not by faster computers but by deeper analysis — understanding each holdout machine well enough to classify it. The 5-state holdouts yielded to Collatz-like reasoning, modular arithmetic, and careful case analysis. The 6-state holdouts resist these same techniques. The tools that suffice for five states do not generalize to six. The boundary is not computational but conceptual: the sixth state introduces mathematical structures that existing proof techniques cannot resolve.
The Busy Beaver function makes the abstract concrete. The halting problem — Turing's 1936 proof that no general procedure can determine whether an arbitrary program halts — is typically presented as a theorem about infinite sets of programs. The Busy Beaver sequence localizes it. Somewhere between BB(5) and BB(6), the halting problem stops being theoretical and starts being operational. The programs are finite, short, fully specified. The question of whether they halt is equivalent to open conjectures in number theory. The incomputable has an address, and the address is lower than anyone expected.