Jacobs’s paper is not for neophytes of natural language generation or knowledge representation. It is written for an audience that can appreciate the difficulties of providing a plausible mechanism for a knowledge representation that can be easily and straightforwardly extended into generation applications. Herein lies the crux of the paper: is the plausibility of a mechanism enough to attack a problem as significant as that of language generation? Jacobs makes it clear that conceptual and linguistic knowledge should be separated, yet linked as needed. Through a series of processing steps, he then aims at a reduction of the problem of language generation to a problem of search. The paper describes a smart way of generating language, at least on the practical (software) level. There is little theoretical contribution to the problem of language generation in particular and language processing in general. However, Jacobs’s methodology will contribute to theories in the long run.