Interestingly, the way to overcome this problem is to develop and use combination therapies. I learned (in a virology class, but the principle applies to anything with a replicating genome) that a very good estimate of the combined resistance rate is simply the product of the separate resistance rates associated with each drug in isolation. I don't think this is commonly done with antibiotics yet, but perhaps someday this will become necessary.
I should also mention that when the selective pressure imposed by an antibiotic is removed, a population of resistant bacteria will invariably revert to a wild-type phenotype because of random drift (or loss of its resistance plasmid where applicable). So old antibiotics do become useful again after enough time passes.
Antibiotic drug development essentially ceased in by the 1980s for a bunch of complex reasons - breaches of capitalism as well as scientific naivety - I think that we'd be just fine right now if the pipeline had not stalled.