The site rt.live is a website for tracking the Rt value for COVID-19 for states in the U.S. The Rt value is the average number of people infected by someone who is infected with COVID-19. In almost all cases, the Rt for each state starts high, generally around ~2-3, and then quickly declines to a number closer to 1.
The image below shows the described trend for a handful of states, but the same trend holds for almost every state.
My question is why does the Rt value start so high?
My a priori guess would be that the initial value of 2-3 is the "normal" transmission rate and it went down when states enacted quarantines/restrictions. However that does not seem to hold water as the drop off, in almost every case, occurs very quickly, and often before states first enacted quarantines. My other guess is that the model, early on is just noisy and uncertain due to the limited data, but it shows error bars which aren't that large, and doesn't explain why it's consistently high (and not more evenly distributed between, say, 0-3). I was hoping someone with more knowledge about how these models work or the data fed into them could provide better insight.