Noting that every week when the data are released, there are 0 cases reported (so, for 8/17-8/23, the data are released at noon on 8/19 and have 0 cases for the week). Unless the resolution criteria change, it seems the data as of 10/1 will be based on a data release containing 0 new cases for the week of 9/28-10/4. Therefore, my read of this situation is there is actually one less week than I have been using for my calculations.
The observation is correct, but I don't think this means you can use it to drop one week from the remaining ones (6): when we say that we have a recent average of, say, ~10 cases/week, this number (10) comes from the revision of the previous weeks' numbers (the latest week is usually zero, as you correctly say), which will be there in the last week of reporting, too - it will not go away.
Please notice that I, and I guess most of us, when we refer to a week here, we mean the 7-day period from Wednesday to Tuesday; it is true that CDC reports on Aug 20 (Wed) the cases up to Aug 19 (Tue), while the most recent entry in their weekly table (usually zero) indicates "Week Start Date - Aug 17" (Sun) - so there seems to be some ambiguity here about what exactly is the latest week reported. But this does not have any practical implication for us - we measure the weeks as the days between reports, and, starting from Aug 20 (last report), there are 6 weeks (i.e. 6 weekly reports) remaining until resolution (Oct 1), and not 5.
The observation is correct, but I don't think this means you can use it to drop one week from the remaining ones (6): when we say that we have a recent average of, say, ~10 cases/week, this number (10) comes from the revision of the previous weeks' numbers (the latest week is usually zero, as you correctly say), which will be there in the last week of reporting, too - it will not go away.
Please notice that I, and I guess most of us, when we refer to a week here, we mean the 7-day period from Wednesday to Tuesday; it is true that CDC reports on Aug 20 (Wed) the cases up to Aug 19 (Tue), while the most recent entry in their weekly table (usually zero) indicates "Week Start Date - Aug 17" (Sun) - so there seems to be some ambiguity here about what exactly is the latest week reported. But this does not have any practical implication for us - we measure the weeks as the days between reports, and, starting from Aug 20 (last report), there are 6 weeks (i.e. 6 weekly reports) remaining until resolution (Oct 1), and not 5.