I can already see a lot of folks are planning covid-19 causal inference strategies that aren't going to work at all

First and foremost, policy responses have been highly endogenous, so you can mostly stick your identification strategies that use policy timing as an instrument in the garbage

Timing is also tricky! Covid-19 isn't capable of hitting everywhere at the same time, so you have to make a judgement call of how to define when the epidemic "began" in an area. This degree of freedom makes it possible to show whatever you want with the data

Another issue is that social distancing and policies are only loosely correlated. It's not going to happen that a government lets the epidemic rage without implementing lockdown, but even if they did, people would still choose to stay home