The paper builds a bridge between two different ways of giving meaning to logical statements. The first approach, called game semantics, imagines logic as a kind of game between two players: one trying to prove a statement and one trying to refute it. The moves in the game correspond to steps in a computation or proof, making this framework well suited for thinking about programming languages and interactive reasoning. The second approach, called proof-theoretic semantics (specifically base-extension semantics), says that the meaning of a statement comes from the rules that allow you to derive it, rather than from some abstract notion of truth. Instead of asking "is this statement true in some mathematical model," you ask "can this statement be proved from a given collection of basic inference rules."
The central contribution is showing that these two frameworks are, in a precise technical sense, equivalent. The authors establish what is called a fully abstract correspondence, meaning that two programs or proofs are indistinguishable in the game-based picture if and only if they are indistinguishable in the rule-based picture. This is a strong and satisfying result because it means neither framework is hiding information the other misses. The proof draws on techniques previously used to connect proof-theoretic semantics with intuitionistic logic, a form of logic where "truth" requires constructive evidence rather than just the absence of a counterexample.
To make these abstract ideas concrete, the authors walk through the example of solving a 4x4 Sudoku puzzle. In this setting, the atomic rules correspond to the constraints of the puzzle, proofs correspond to solving strategies, and the game corresponds to interactively demonstrating that a solution is correct. This example neatly illustrates how the framework captures real reasoning tasks, where meaning arises from the process of working something out rather than from a pre-given answer.