[Goose]

To complicate matters, R(eff) can't be directly observed; it can only be estimated, and usually with a lag of about 2 weeks.  So it is really easy to get into trouble without knowing it, which is what happened in June/July.

Very true. San Mateo County is "all in" for R(eff). As they have pointed out, there are also many ways to compute R(eff), leading to even more confusion. I use https://calcat.covid19.ca.gov/cacovidmodels/, which shows several "models". The current State of California approach to increased opening utilizes the "simplified" way of scoring counties performance. The State uses the number of cases per 100,000 residents and the percentage of tests results reported positive over seven days. They aren't using R(eff). Seems a bit strange to me. I know of no justification (explanation) of why these metrics were chosen. I also know of no justification for the thresholds for the various "colors". The chosen metrics also have a "moving window" evaluation period, so they suffer from the same delays as R(eff). Very curious IMHO.

[Goose]

To complicate matters, R(eff) can't be directly observed; it can only be estimated, and usually with a lag of about 2 weeks.  So it is really easy to get into trouble without knowing it, which is what happened in June/July.

Very true. San Mateo County is "all in" for R(eff). As they have pointed out, there are also many ways to compute R(eff), leading to even more confusion. I use https://calcat.covid19.ca.gov/cacovidmodels/, which shows several "models". The current State of California approach to increased opening utilizes the "simplified" way of scoring counties performance. The State uses the number of cases per 100,000 residents and the percentage of tests results reported positive over seven days. They aren't using R(eff). Seems a bit strange to me. I know of no justification (explanation) of why these metrics were chosen. I also know of no justification for the thresholds for the various "colors". The chosen metrics also have a "moving window" evaluation period, so they suffer from the same delays as R(eff). Very curious IMHO.