00.00 12k melee (unblocked)
01.00 9k arcane storm
02.30 20k arcane breath
04.50 12k melee
He can't swing at the same time as casting his breath by the looks of things, but can cast arcane storm.
Now, with 3% short of block cap, quite frequently in any given string of hits one will get through that isn't blocked, increasing the total damage of the string by ~1.6k over what would be possible if block capped. Occasionally you'll get two unblocked hits in a row, making the string 3.2k more damage than was possible if you were block capped. It's easy to contrive examples where that 1.6k or 3.2k might be the difference between living or dying. e.g. say for whatever reason, you received a total of 52k effective healing over this 4.5 second period. If not block capped, one of the melee hits slips through without being blocked and the string hits you for 53k and you die horribly. If block capped, it hits you for 51.4k and you live long enough for the next heals to land. Now the question is, how likely is it that you'll take ~1k less effective healing than you need to to survive? Usually when one dies it's not because the healing deficit was of the order of 1-2k, but rather because a 10k+ heal that you were expecting never arrived for whatever reason.
Equally, it's easy to see how an extra % of pure avoidance will make it less likely that you'll take the full string in the first place, increasing the odds of avoiding at least one. I don't think this can be resolved purely through rhetoric, but would have to be simulated in some form to get a definitive answer. As a general principle, and it's nothing new or exciting:
The smaller the proportion of incoming damage that can be absorbed by block value, the less valuable being block capped is, especially when it comes at the expensve of pure avoidance.
_________________
