As far as I saw from the start, predictions for Georgia were that they'd start to see an uptick in about 3-5 weeks. I think that window starts this week.

You can read the article I linked, the predictions were for the week ending May 16.

It has largely been within that consensus range at the low end: "579 and 2,292"; but that's splitting hairs, since the low end is not an increasing count.

With that said, my own reading prior to this point has suggested we'd see the uptick starting in this coming week, through the next two weeks.

Georgia is now backfilling based on date of first symptoms, rather than date of diagnosis. Numbers as far back as a week or two week could still increase.

That's not completely true.

The panel of experts estimated about 1k/day during the week ending May 16, ranging 579/day to 2292/day.

And while so far there aren't that many cases reported today, the data isn't complete due to the lag in testing results. In fact the lag is for 14 days. So we're still in a wait and see point.

The 579/day estimate is the low end of the 90% confidence interval. You can use the 7-day rolling average the NYT provides to eliminate the swings from reporting variance. Of course, the apocalypse will always be two weeks away.

90% confidence means anything within that range would validate their model. And even if their model was perfect, 10% chance the real value would fall outside that range. So as long as the real number (which we will will have in 2 weeks once the rolling average and final tallies are in) then, well, they're still right.

Rolling averages introduce a lag.