The appropiate answer to this is to say comparing an id from a document retrieved from one URI to an id retrieved from a second URI is not reliable. This makes feed:ids largely useless. The primary use, of comparing with a previous version of a document retrieved from the same URI, is fine.
Graham
