Considerable effort has been devoted over the years to fighting uncertainty in geographic information in its different manifestations. Thus far, research on handling inaccuracy, fuzziness, error and related issues has focused for the most part on problems with spatial data and their direct products, typically representations of spatial objects or fields. This paper seeks to broaden the discussion of uncertainty in the geospatial domain by shifting the focus from information to knowledge. It turns out that there is a surprising number of things that we cannot know (or questions we cannot answer) that are not the result of imperfect information. Forms of not knowing are pervasive in domains as diverse as mathematics, logic, physics, and linguistics, and are apparently irreducible. This being the case it may help to explore how these realms of ignorance may affect our efforts. The paper distinguishes three different modes or forms of geospatial knowledge production, and argues that each of them has built–in imperfections, for reasons of logical principle and not just empirical fact. While much can and needs to be done to manage and resolve uncertainties where possible, I argue for accepting that uncertainty is an intrinsic property of complex knowledge and not just a flaw that needs to be excised.