Observable negation and double negation

We now study an explicitly different interpretation. Classical formulas use ordinary complement of truth regions. In the open-set interpretation, the values of formulas must remain open. On a nondiscrete space, a classical complement need not satisfy that requirement. For this example there are just two points, zero and one. The only opens are the empty set, the singleton one, and the whole carrier. The singleton zero is not open. Interior means the largest open subset of the set being considered. Use the labelled two-point space, with X interpreted by the open singleton containing one. Its classical complement is the singleton containing zero. That set is not open in this topology, so it cannot be the open-set value of negation. Take the largest open subset inside the complement. This is Heyting negation. It includes every open event disjoint from U, while remaining disjoint from U itself. It need not preserve every classical falsifying point. The complement of the singleton one is the singleton zero. Its only open subset is empty. Therefore open-set negation of our chosen X is empty, even though its classical complement was nonempty. Apply the same open-set rule again to the result just obtained. The complement of empty is the whole carrier, already open. Double open-set negation therefore gives the whole two-point carrier. The intermediate empty set matters. Replacing this two-step evaluation with a classical simplification would erase it and give the wrong open-set result. A recursive semantics evaluates each connective using the specified operation. Classically, double negation returns the original truth region, the singleton one. Taking its interior afterward still gives the singleton one because it is open. That is different from the recursive open-set result, the whole carrier. We have displayed two procedures with different answers. This is not an inconsistency within either procedure. It shows that classical truth evaluation followed by one interior operation does not generally reproduce recursive Heyting evaluation. An open event and its Heyting negation are disjoint by construction. Every open event disjoint from U is included in this largest open exclusion. These statements follow from the interior definition, not from assuming that all classical complement identities survive. In a discrete topology, every complement is already open, so interior removes nothing. Heyting negation then agrees with classical complement. The difference is created by the observability requirement in a nondiscrete topology. On the same two-point topology, start with the empty open event instead. Compute its Heyting negation and double negation. Keep the result of the first operation visible before applying the second. The first complement is the whole carrier, whose interior is itself. Negating that whole carrier gives the interior of empty, so the double negation is empty. The procedure is the same even though this input happens to return to itself. Open-set negation means interior of complement at each step. It should not be substituted silently into the earlier classical inference videos. Next we will find the corresponding open implication by asking for its defining universal property.