There is the popular argument that it takes many more glasses to
wash the glass than it contains, but the people quoting this statistic never talk about bread plates or all the cloth on the table.
These aren't impossible things to study, to determine the impact and optimize customs more appropriately, even though it seems likely that a minuscule capital outlay in the California water system would take care of drinking water and restaurant laundry and dish washing indefinitely.
The same goes for the psychological impact of the regulation. In popular treatments, any social science related statement is valid simply by being said. "Maybe" forgoing water at the table makes people feel that they've "done their part" and they are then more likely to do no more, and "maybe" they use up all of the energy they have to devote toward water issues. Given that dysphagia is notoriously unexpected, "maybe" the health impact of no water at the table far outweighs the zero-effect it has on water consumption.
All of these maybes are terrible ways to determine public policy. Instead, take a look at the relevant literature, and do some valid experiments.
I think these bad things get entrenched because people likely to get passionate about the issue aren't likely to explore the math, and people who have an intuitive sense of the numbers, don't want to sound crazy by getting worked up enough about the issues to balance out the people who have good intentions. Sure, it doesn't hurt that much to fine restaurants or make people ask for water, but if doesn't do anything productive then the net effect is still negative.