Although network effects can make predation more likely to succeed, we find that the leading anti-predation rules may lower or raise efficiency and consumer welfare in network markets. We find that: (a) the extensive debates about the ‘correct’ measure of cost on which to base price floors are unlikely to be productive; (b) the Ordover-Willig rule that is widely thought to be correct in theory but difficult to apply in practice is, in fact, incorrect in theory; and (c) efficient price floors would have to depend on consumer expectations and coordination processes that are unlikely to be observable in practice.