This is probably the most informative and best researched information to appear in this sorry matter.
Apart from one small confusing paragraph about the AoA on the instrument panel it is a valuable contribution - possibly more so than the accident reports will prove to be.
Perhaps the most significant lesson to be taken from the article is the assumption that within three seconds, pilots will recognise and deal with a trim runaway. That may be the case taken as a single failure but in the context of the cockpit environment in this case of MCAS activation, where the crew was already trying to deal with a rogue stall warning, in an after takeoff phase etc., it was just one more item on a loaded list and three seconds was not enough for recognition and action. Meantime the controls get heavier and MCAS repeats...
The cause of all this chaos was the same - the AoA failure. It didn’t just set off MCAS - it set off multiple spurious warnings. It did this on three known occasions and only when a third pilot was available could the situation be partially defused. That took more than three seconds too...
So when designing on probabilities of failure, it isn’t enough to examine a single system or to presume that everything else in the cockpit is normal. Some basic design rules need looking at here.
As for the company culture, others are already commenting on that aspect.