Convergent evolution or chance look-alikes

Why should caddis fly larvae and a species of aquatic snail look alike? Mimicry is rather common in nature for it often confers some sort of advantage to one or both of the species in the turmoil of evolutionary pressures. Or so the theory goes. Most examples of convergent evolution involve closely related species. In the present case, though, the species are in different phyla. The caddis fly larva builds its snail-like shell by cementing grains of sand together with a silk-like secretion, while the snail's shell is a calcareous excretion. One would expect a strong advantage to be conferred on one or the other species, especially in the matter of predation. Using brook trout as predators, how ever, proved perplexing, for the trout would eat only the snails, avoiding the carbon-copy larvae.

(Berger, Joel, and Kaster, Jerry; "Convergent Evolution between Phyla...." Evolution, 33:511, 1979.)

Comment. This is a remarkable case of mimicry. One wonders how the caddisfly larvae know exactly what snails look like, and how the unique shell constructing methods were coded into its genes by evolution. Or did the snail emulate the larvae?