Although I mostly agree with you (e.g. I consider the lack of forthrightness as big a concern as the potential liver tox - or bigger (as I alluded to in the original post)), but some potential, minor, disagreements:
While I agree that the mapping of ALT >3xULN to Hy's Law is poorly understood, I think the other things make it a bigger risk than they would if the only thing we saw was the 2.8% ALT >3xULN. The bilirubin, the spike in ALT on first dose, and even the quick liver tox in the alcoholic. Further there is data that Fibrogen hasn't actually published, so far as I can tell: they appear to have run a special trial just to determine how the drug worked in those with moderate liver impairment ( https://clinicaltrials.gov/ct2/show/NCT02161224?term=roxadustat&rank=11&submit_fld_opt= ). Perhaps nothing of importance happened, but given their behavior it is reasonable, IMO, to wonder.
BTW - I'll reiterate what I have said before (on the subject of IONS): all companies have character flaws - e.g. hiding safety data is a big one, and another is massive post hocing - so, by itself it can't be a disqualifier. But it does go into the decision mix for me.
BTW2 - Interesting note: the only company for which I have never been able to find a systemic flaw has been BMRN. It actually kinda bugs me -g-.