Because variation in testing makes it hard to know *exactly* what is happening, it's important to account for one's priors, and IMO those priors should be that there has been a gradual slowdown in new cases as more places have implemented social distancing.
To be more precise, my priors are that as of a week or two ago (that's what we're seeing given the lag in the data) there would have been enough social distancing in place to slow but not reverse the spread. (Say, an R0 of ~1.1-1.6). The data seems consistent with that.
Of course, that's a national average. Maybe, as of 10 days ago, R0 was 2.1 in Florida, 0.9 in NYC and 0.7 in the SF Bay. The data seems consistent with that. But it's hard to anchor to anything when so many cases are undetected, so it could also be consistent with other theories.
