One is that the value of transactions with the new lower fee is outweighed by increased orphan costs and miners refuse to include them en-masse. Wallet authors lose the staring match and go back to setting higher fees until such a time as block propagation is optimised and the orphan costs go down. Nodes that are encountering memory pressure can increase their min relay fee locally until their usage fits inside their resources. It's annoying to do this by hand but by no means infeasible.

The other is that the total value of transactions even with the lower fee is not outweighed by orphan costs. The value of a transaction is higher than its simple monetary value - the fact that Bitcoin is useful, growing and considered cheap also has a value which is impossible to calculate, but we know it's there (because Bitcoin does not exist in a vacuum and has competitors). In this case miners stop including lots of useful transactions that represent desired economic activity and are put under pressure by the community to change their policies. If all miners do this and making small blocks is considered errant behaviour, then we're back to the same situation we're in today.

The possibility you're worried about - that someone does a DoS attack by flooding the network with small transactions - is only an issue in the first situation, and it is by no means the easiest or cheapest way to DoS Bitcoin. We all want to see more DoS resistance but basically any change to Bitcoin can be objected to on anti-DoS grounds at the moment, and this will remain the case until someone steps up to spend significant time on resource scheduling and code audits.