Minsky's theory of communication can lead to imperfect reconstructions in listener's minds -- but this can be good. They can provide new insights, which (for instance) can lead to the solutions of problems.
"When we try to explain what we think we know, we're likely to end up with something new. All teachers know how often we understand something for the first time only after trying to explain it to someone else."
NB: The notion of reconstruction in language is central to the old empirical work of Bransford, Barclay, and Franks. One thing that this work points out is that when people hear or read sentences, veridicality is rarely the result -- understanding really does seem to be almost exclusively reconstructive, along the lines that Minsky has been discussing in Chapter 22.