It was you who pointed at the day-to-day discrepancies and asked "what's the point then". The point is to have a metric that you can work with in realistic timeframes, e.g. every week or every other week. The day-to-day discrepancy is not a big deal, unless you work in the media and you have to push a daily bulletin to ensure "engagement".
I think there is a problem with the method though, if you’re selectively testing and changing the number of tests per day, then you’re going to have trouble establishing a useful positivity rate, or am I wrong? There should be some consistency.
If the results presented are hourly, daily or weekly, that’s just the resolution of said metric.
Maybe you’re arguing there isn’t much point in having higher resolution metrics than weekly but I don’t see how taking more frequent measurements matter, it doesn’t change the outcomes.
