This makes no sense to me, starting with the assertion that DL does not provide any upgrade inventory within two weeks of departure (but only at certain airports) not having anything to do with the *algorithm* that is used to prioritize upgrades. These are distinct parts of the upgrade process; the algorithm is different from policy rules about when upgrade inventory is necessarily zeroed out in that the algorithm orders passengers for whatever upgrade seats are available.

The author claims that ATL, BOS, NYC, DTW, MSP, and LAX departures are included while those from PHL are not, but neglects to mention what happens according to his "theory" at SEA, PDX, SLC, PIT, etc. In fact, the article only mentions two data points: upgrade inventory at a single point in time on July 8 [the article states June 8 in one place but this seems to be an error] for a LAX-LHR flight on two consecutive dates in July.

I also see no evidence that whatever change is claimed to have occurred in fact happened on July 1.

We all know that seat maps are bad indicators of the number of tickets that have been sold on a given flight. Moreover, DL has data (beyond the number of unsold seats to date) that can predict whether business class is likely to be full on particular flights, so it's silly to expect upgrade inventory to track the number of empty seats.