Adams and Maher read C&C and Haugeland as agreeing that high-bandwidth "connections" are not really interfaces. They read the C&C "trust and glue" conditions as high-bandwidth conditions. (cf. pp. 2-4.)
A) I don't think that that is a correct reading of C&C.
B) I don't think the trust and glue conditions are high-bandwidth conditions.
Regarding A), there is some place where Clark challenges Haugeland's idea that there cannot be a high-bandwidth interface. Why, Clark asks, can there be no such thing? I don't have my copy of Supersizing here with me in Germany, but I think it's in there. (One could probably check the index of Supersizing for "Haugeland" and track it down.)
Regarding B), I think you can separate the trust and glue conditions through the example of trusting a coin toss in navigation. (This is in my Philosophical Explorations paper, "Distinguishing Virtue Epistemology and Extended Cognition".) Basically, you have Otto who trusts whatever the "lucky" coin tells him in order to navigate through NYC. (Set up the story to fit the trust and glue conditions.) But, a coin toss only carries, what, one bit of information? Low-bandwidth. So, you can trust and be glued to a low-bandwidth connection, it seems to me.