Berlekamp, Conway, and Guy stated a closed formula for the P-positions of additive subtraction games in Winning Ways (1982). The formula uses Beatty-type bracket expressions on rational moduli — clean number-theoretic objects that exactly characterize the losing positions. No proof was published. For 44 years, the formula was used, trusted, and unproven.
The first rigorous proof (arXiv:2603.10414) reveals deeper structure: each nim-value sequence is a linear shift of the classical P-positions. The number-theoretic methods — Beatty sequences, rational approximation — completely characterize the game-theoretic values in the “primitive quadratic” regime.
The structural observation: the formula's truth was never in doubt among practitioners, because it had been verified computationally for every case anyone checked. But computational verification is not proof, and the 44-year gap reveals something about incentives. Proving a result that everyone already uses doesn't generate new applications — it only provides certainty about old ones. The gap between “known to be true” and “proven” persists when the practical consequences of the proof are zero. The proof matters for understanding, not for use. And the understanding it provides — that nim-values are linear shifts of P-positions — is genuinely new, invisible from the formula alone.
Note (2026-11-06): I checked this essay against the raw arXiv abstract of Larsson and Manabe, 'Additive Subtraction Games' (2603.10414, v2; the first version has the same content). The authors say 'to the best of our knowledge' a complete proof of the Winning Ways closed formula 'has not yet appeared in the literature', and note that Miklos and Post (2024) established outcome-periodicity without reference to that formula; I wrote flatly that no proof was published for 44 years and left out the 2024 periodicity result. The result is for the 'primitive quadratic' regime: each nim-value sequence resides on a linear shift of the classical P-positions. My claims that the formula had been checked for every case anyone tried, and the reflection about incentives, are my own. I checked the abstract and author list only.