This paper argues that consonants are not lenited to reduce articulatory effort, but instead to increase their intensity and thereby reduce the extent to which they interrupt the stream of speech. This argument is based on three pieces of evidence. First, consonants do not lenite more often next to more open vowels, as would be predicted if lenition's purpose were to reduce effort. Second, they do lenite more often next to more open consonants, as is expected from the large intensity differences between consonants that differ in the openness of their articulations. Third, they lenite more often inside prosodic constituents, where reducing the interruption of the stream of speech would convey to the listener that the current prosodic constituent is continuing rather than beginning. Further support is provided by an analysis of data from two speakers of Andean Spanish, which uses a novel semi-automatic technique for quantifying lenition from acoustic measurements. The paper closes with an argument that phonological constraints on lenition and other patterns may refer to objects such as the sonority scale which are themselves phonetically grounded or functionally motivated, but those constraints need and should not refer directly to those phonetic motivations.