(p. 157) … impossible events (see sidebar), such as the combination "Door picked=red" and "Door shown empty=red".
(p. 159 sidebar) … the standard approach (as adopted in Table 6.6) is to assign equal probability to each column entry (hence the assignment of 1/3's in the table"
(p. 158 Table 6.6) Since the combination Door picked=red" and "Door shown empty=red" is impossible, the entries in column "Switch" are set to 1/3, according to the standard approach. But the entries in column "Stick" are 1 0 0. Is this done by purpose? Or can/should the entries under "Stick" also be set to 1/3, consistent with the standard approach?
Yes, there isindeed an inconsistency here. Although it does not matter what values we assign in the NPT to impossible state combinations, we should have assigned 1/3 to all entries for the "Stick" option. I will add this to the errata.