Actively tracing contacts and testing them would be the key thing. When one person is found to be infected, you figure out where they've been and who they've interacted with, so you can test those people. Any of those people test positive? They go into quarantine and you trace their contacts. That's how you slow and eventually break the spread. Note that contacts don't need to be symptomatic to be tested.
Instead we're effectively doing the opposite, which is testing IF AND ONLY IF you BOTH have symptoms AND can prove contact with a confirmed case. Which doesn't do much to stop transmission (it's too late at this point; you've already spread it), and certainly doesn't give an accurate picture of the prevalence or behavior of the disease.
Have any nations found success with that strategy? Could it be that we simply couldn't conduct that level of investigation fast enough at scale to be effective? And are the tests accurate enough for that approach to work?
It's my understanding that this is what South Korea, Hong Kong, and possibly Taiwan did (and are still doing). As well as China once they acknowledged the outbreak. (Which is not to condone China's methods of enforcing the quarantine.)
I'm just a lay observer of all this, not an authority. But my impression is that the U.S. could have done it effectively if a) the CDC test situation had not been borked and b) the federal govt had acknowledged the risk of epidemic (because I expect it's the CDC who has expertise in this tracing exercise, much more than state health departments).
As to whether it's too late, I honestly don't know. But it seems like it would be useful in controlling the epidemic regardless of its state.
Instead we're effectively doing the opposite, which is testing IF AND ONLY IF you BOTH have symptoms AND can prove contact with a confirmed case. Which doesn't do much to stop transmission (it's too late at this point; you've already spread it), and certainly doesn't give an accurate picture of the prevalence or behavior of the disease.