Interpretation of interaction terms

In the post about dummy variables the reader Barnabé commented that multiplicative terms in econometric regressions are frequently misinterpreted as well. He suggested this paper on the issue. Thanks!