I think not only for books, but for published papers etc. also there should be errata lists----these could really save many a hair-pulling moment! Moreover, this will make the errata-fixing process public, and because of that probably faster and more transparent.
In fact, it would be great if such a database were created on the stackexchange.com framework, because that works so admirably for MO.
I created a proposal on stackexchange.com, if you are interested please follow it here:
http://area51.stackexchange.com/proposals/23788/math-errata-database
Additionally, if someone has a better proposal, I am totally willing to support that too.