Picture a swinging pendulum.
As it moves, the direction and speed of its motion and the direction and magnitude of its acceleration are continuously changing.
In the real world, unless another force is applied, the pendulum won’t swing forever. Air resistance and friction will slow down the pendulum until it stops.
But you could compensate for this damping. You could add a timed driver that counteracts the frictions and causes the pendulum’s amplitude to increase instead of decrease over time.
For example, someone could push at regular intervals (or you could add a motor).
Or you could pump your arms and legs on a swing, adding your own timed force to increase your amplitude.
Pumping your legs requires timing- if you reverse the motions you damp the oscillation instead of driving it.
The driven pendulum- the playground swing with someone occasionally pushing behind- is I think a useful metaphor for social forces and their results. Even in the simplest model, we are observing the result of multiple interacting vectors. More than that, timing is everything. One of the strange presumptions of social science is that causal factors, in order to be considered important and real, should should work more-or-less the same across time and space; but the push only works to get you higher on the swing when it happens at the right time.
I’m quite sure that, as things actually occurred, the Obamacare Medicaid expansion worked to accelerate the opioid crisis. Given the actual track record of overdoses from 2010 and on, you would have to imagine an absurdly terrible counterfactual in places like Ohio, West Virginia, and New Hampshire in order to believe that the Medicaid expansion substantially reduced overdoses instead of increasing them. But the push on the swing came at the right time and at the right place- or more accurately, at the wrong time and at the wrong place- in states where the economy, culture, medical and immigration system, and other existing insurance and welfare programs all worked to magnify the effect on overdoses of expanding insurance to people immediately above the poverty line. Plainly, outside the US, many countries have put in place publicly funded insurance systems without worsening their drug addiction problems or causing tens of thousands of deaths. It may well be that, eventually, when Kansas and Nebraska and the rest of the Holdout states expand Medicaid, as I expect they eventually will, the push on the swing won’t have remotely the same effect on drug abuse or overdoses; I certainly hope it will not.
This is also true, though, of social “solutions” instead of social problems. A friend of mine who has been involved in studying Head Start and related programs for many years often points out that some of the pilot trials in the early 60’s were done in rural counties in Mississippi with endemic tuberculosis. Under those circumstances, at that time, the push on the swing made a big difference; I can believe that pre-school does a lot of good when the alternative is spitting up blood. But James Heckman is still using Perry Preschool and Abecedarian- tiny studies from over four decades ago, almost as old as the Head Start pilots- to make arguments about the effects of expanding pre-school and early childhood programs now, without any kind of real acknowledgement that the circumstances for low-income kids in the US are radically different now from what they were then.
Social scientists are almost inevitably in the position of trying to establish generalizable principles across time and space, rather than merely viewing their work as data fossilized immobile in amber. But contingency is the rule, not the exception, and the push on the swing makes you go faster or slower depending on when it comes.