I have been reading about the new data guard features in 9i and something does not make sense to me.

I understand the concept behind data guard. I understand the different protection modes (Guarantee, Instant, Rapid, Delayed), which by the way have changed names between Oracle 9.0.1 and 9.2. I also understand that you can specify "dalay" as an option when configuring the log_archive_dest parameter. This will allow for a delay in applying the redo logs to the standby database. I also understand that this delay can be used for the prevention of "bad" sql (such as an accidental Drop Table command) on the production DB from propagating into the standby database.

This is what I don't understand:

If my database is setup with guaranteed protection and a delay of 45 minutes, this is what I read "In the event that some user error occurs.. and if the error has not yet propagated to the standby database, you could initiate a graceful switchover to that standby database, thus avoiding the need to perform an expensive imcomplete recovery on the primary database." If you do this, have you not just violated the reasoning behind setting up your db environment with guaranteed protection? In effect, your standby db is no longer in complete sync with the production db and you will have data loss.