Another way to look at the problem:
How does the server generate the attack table?
Simplified example:
5% miss, 10% dodge, 10% parry, 0% block, 30% crit (45% hit)
(No resilience or extra defense)
Code:
001-050 miss
051-150 dodge
151-250 parry
-------- block
251-550 crit
551-1000 hit
With 5% crit reduction (from any source), there are two basic ways the attack table could look:
Code:
001-050 miss 001-050 miss
051-150 dodge 051-150 dodge
151-250 parry 151-250 parry
-------- block -------- block
251-500 crit 251-300 hit (mitigated)
501-1000 hit 301-550 crit
551-1000 hit
Now, how do I justify the possibility of table #2? From what I hear, there is no entry in the attack table for "resilience-mitigated crit". But is there an entry for "defense-mitigated crit"? If so, resilience-mitigated crits were probably rolled into it [insert reference to Blood Craze/Enrage here].
If table #1 is correct, then in my previous data, a) would be the right answer (RESULT: 14% crit). If table #2 is correct, then d) would be right (RESULT: 6% crit, 8% hit).
It is possible to test this* to some degree against a tank with high passive avoidance--but not enough to be 'uncrushable' vs. the person attacking. (Assuming there's nothing serious I don't know about the PvP attack table: ) Find someone who would be crit capped** against your high-avoidance tank. Have that someone autoattack the tank (no specials), and watch the combat log.
If I am correct, all attacks should be misses, blocks, dodges, parries, and crits. There should be no regular hits at all. If there are any regular hits, then table #1 can't be right.***
*Again, I lack resources to test this myself at the moment. I'm not in a raid guild, and really don't know any tanks who are raiding at a high level right now. I'll still try to test it if I get a chance, though. Maybe I'll spam trade for a test subject next time I'm on...
**It is imperative that your tank have enough avoidance to push regular hits from your attacker off the attack table--don't forget to take crit reduction, hit rating, etc. into account here.
***This test can be repeated with zero resilience on the tank and with some resilience on the tank in order to determine whether resilience and defense mitigate crits differently.
Bookmarks