I've no idea with ADSL, but that is my observations and understanding with VDSL.
I would think that wwwombat would be able to give a much more technical answer.
I didn't spend enough time on ADSL/2+ to note any inconsistent behaviour. And I don't think we have enough evidence of behaviour in VDSL2 with a target different from 6dB to be sure.
I think J0hn is correct in theory. The value 6 shouldn't be hard-coded, and should be gathered validly from the sync process. It is most certainly how I would write the code ...
But, and it is a big but, we have scant evidence that this is true for every modem, every firmware revision, every DSLAM type, every linecard type, every chipset type. Corners are cut, bugs exist, testing rushed and incomplete. People get things wrong, including the observations made by ourselves and related on here.
I've spent years watching and deciphering the DLM behaviour in FTTC, and can safely say it is always a mistake to make a definitive statement. Especially based on a sample of 1.
Best answer right now? Keep monitoring (thanks for that software, R0nski), and build a pool of evidence, looking for both the norm and the oddities. Expect to start figuring things it after a couple of months of honing those observations.