I will discuss a conjecture which provides a relationship between Manin
symbols in the homology of modular curves and cup products of cyclotomic units
in Galois cohomology with restricted ramification. I hope to be able to give
something of an idea of its meaning, with an eye towards generalization. I will
also relate recent progress by Fukaya and Kato that essentially proves the
conjecture under a mild but difficult to remove hypothesis.