Plausibility and Truth: A Graded Taxonomy of Speculation

Exploration produces speculation, and the word ``speculation'' is commonly used as a verdict rather than as a description. The same label is applied to general relativity before the eclipse of 1919 and to an invisible dragon that leaves every instrument silent. This paper, a sequel to Consolidation and Exploration, proposes to grade speculation instead of sorting it into a single sack. It distinguishes plausibility, consistency with what is known, from truth, survival of a test that could have refuted it; and it recalls, through Carnot, Fresnel, Maxwell, Ptolemy and Newton, that excellent precision in calculation confirms the structure of a theory and leaves the truth of its premises to be judged by other means, as Ptolemy and Copernicus, predicting equally well with opposite ontologies, already showed. Four criteria, read as a profile rather than as a label, assess the quality of a speculation: plausibility, fertility, argument and testability. The paper then separates fertility, a measure of potential assessable at publication, from fructification, a measure of realisation visible only in hindsight, and observes that fructification depends on someone exploring the fertility: Mendel's laws were as fertile in 1866 as in 1900, and bore fruit only when others read them. Formulations are then followed through contrasts and revisions: a scale of depth of empirical contrast, from initial compatibility to sustained corroboration, is kept apart from the character of a definitive theory, and revisions are graded by their depth and by the inheritance they preserve. Permitted absences, persistent under search, turn classification into inquiry. A five-grade scale, a gallery of historical cases and two symmetric errors follow, together with practical consequences for authors, reviewers and institutions. Fruit measures readers as well as ideas: the absence of fruit, by itself and without examining the opportunities for development, is not evidence of sterility.