Are they cooking the COVID-19 books?

I like obtaining information from primary sources.  Instead of watching an interpretation of what someone says on any of the news outlets, I’d rather get my information directly from the source.  This is why I will listen to the actual press briefings from the White House, and press conferences by the president first, and then if I have time later, enjoy the entertaining (horrifying?) biased interpretations of what someone apparently said (from both left-biased and right-biased editorialists).

Today, I was catching up with the president’s news conference from August 4th.  I’ve been very intrigued by the assortment of charts and graphs he’s been sharing recently.  Here is an image of one he shared during the news conference (you can verify using the displayed timestamp).  It shows an important metric, Case Fatality Rate, which is discussed frequently, but in this case, it is displayed with a two-week death lag, which I’ve not seen previously. 

In my own analysis of available data, the issue of death lag occurred to me when looking at case fatality rate, but I wasn’t clear how to adjust for it.  It turns out, others have recognized the issue as well.  This letter to the editor, published online June 11, 2020 in the Journal of Infection, discussed the issues with attempting to calculate a useful case fatality rate.  The authors talk about the time to death (death lag) issue, and argue that choosing an arbitrary death lag value (for example, two weeks), is no better than making no adjustment for death lag.  However, the authors suggest a data driven approach to calculate a constant (c∆t) to represent the optimal time delay to death (death lag).  This approach also allows one to account for population differences and other variables impacting death lag, country to country.  They present a graph showing several countries, including the United States, showing the c∆t = 5 days for the U.S.

(Source:  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7286834/)

They make an important statement in their discussion, “closer to the truth for much longer”, which is exactly what scientists ideally attempt to do - construct models that are closer to the truth, and provide reasonable approximations of reality, since we most often cannot fully capture the truth due to all sorts of constraints (which one learns about in a philosophy of science course).

So my question is, what approach was used by the White House data analyst(s) to arrive at a c∆t = 14 days?   It is clear that the White House’s adjusted CFR with a death lag of 14 days is quite a bit lower than the CFR with a death lag of 5 days, as calculated by the authors of this letter.  Curious onlookers would love to know!