What really caught my eye was the subtraction tests the author performed on DNG in his sections on the archival
The first test is strange; it would be interesting to have the native raw file in order to reconstruct the test.
The second test is invalid; it does not prove anything related to the image data format, it proves only that the rendering software (ACR) changed in the meantime.
Throughout this forum and in all the Adobe literature we are told and therefore assume, that the DNG standard has exactly the same information as the raw file
I don't think you have been told that by any competent source. The native raw files contain masses of information, which is undocumented and of unknown purpose, and which will not be converted in DNG format.
What claimed is, that the image data is the same in the native raw and DNG. This is correct in most cases. There are some exceptions, which are all right. There were some exceptions due to conversion error; the program has been mended. However, such cases demonstrate clearly, that one has to be very naive to throw away the native raw file after conversion.
The real question is:
can all the knowledge about a camera, which is useful for the raw conversion be expressed in DNG terms? (No matter if that knowledge is stored in the native raw file or is implicite.)
and the real answer is:
Wow. Yeah, that article misses on just about every analysis it attempts.
The subtraction tests are bogus. He's testing the rendering of the data from 2 different versions of the software. At no point does he actually look at the data itself.
Bad tests, lots of misleading conclusions.
I'd rate that article as slightly below useless.