More generally, sticking with an approach that "works" by some shallow criterion, even though on deeper analysis it doesn't work, because the alternatives are not as good by the shallow criterion.
The example I point to is tokamaks for fusion energy.
The example I point to is tokamaks for fusion energy.