The paper introduces a new arithmetic system called Reflective Grounded Arithmetic (RGA), designed to challenge a common oversimplification about Gödel's incompleteness theorems. The popular version of those theorems says that no consistent arithmetic system can prove all true statements about numbers, but this framing quietly assumes classical logic, where every statement is either true or false from the start. RGA instead builds truth from the ground up: a statement counts as true only when a computational process actually certifies it, and false only when a specific numerical counterexample is found. Statements that neither get certified nor refuted are simply left without a truth value, rather than being forced into one category or the other.
Within this framework, RGA achieves things that classical incompleteness theorems might seem to rule out. The system is both consistent and "openly complete," meaning that provability lines up exactly with grounded truth for well-formed statements. It can define recursive functions freely, prove basic arithmetic facts like the totality of addition and multiplication, and its expressive power matches precisely the class of problems a standard computer can recognize, in the sense of the Church-Turing thesis. All of these results were formally verified using the Isabelle/HOL proof assistant, giving them a high level of mathematical rigor.
The tradeoff is a phenomenon called omega-incompleteness. You can prove, for every specific number, that some property holds of that number, yet the system cannot prove the universal statement that the property holds for all numbers at once. The universal closure is not just unprovable but genuinely lacks a truth value within the system. This places RGA in a logical territory that is neither classical nor intuitionistic: it accepts double-negation elimination, rejects the unrestricted law of excluded middle for quantified statements, and whenever a universal claim is refuted, the system produces an explicit numerical witness to the failure. The result is a coherent and computationally grounded alternative for thinking about arithmetic and its limits.