An interesting question suggested by the post is what true statements about itself can an agent produce in that setting, assuming it can control which statements to withhold (for how long or under what conditions), even if it knows them to be true.
If the agent decides to print any statement at all, that’s probably because it follows from some formal system, right? Then there seems to be no harm in printing all statements that follow from that formal system.
If the agent decides to print any statement at all, that’s probably because it follows from some formal system, right? Then there seems to be no harm in printing all statements that follow from that formal system.