Ok so the testing part.
first of all finding a damage source that always hit for the same amount. I've found it in the arcane missile spell from mage. Just went duel a mage equiping different sets of stuff. The conciderated mage was lvl 80 so there were no partial resists from any difference of levels. He doesn't got any spell penetration, just hit / crit. I've noted down only the normal tick, so no crit, no resist at all. I've tryed some different stuff so my def was getting from 400 to 572 (max I can get today) and just let him burn me down while on IBF.
So the results :
Def => w/o / with
400 => 727 / 559
426 => 727 / 595
470 => 727 / 562
496 => 727 / 539
541 => 727 / 429
572 => 727 / 391
The first thing I found about that was the odd 400 - 426 def mechanism. I need to conduct more test on that part before we can say anything. This part was done in frost presence. I've done some tests while in blood presence and found similar results :
541 => 856 / 505
572 => 856 / 462
Then I speced for Frost Aura, and there are the most surprising thing. I'm not really up to date with the partial resists mechanic, and this may be where I fail finding an explication to theses numbers :
541 def, Frost presence up, IBF up, same mage : 386 (86 resisted) => so what is that resist at 18.2% ? And more important, why this hit was hitting me at 472 instead of 429 ?
Some more results listed as :
Def / Presence (F/B) / up or down for ISB :
541 / B / up : 454 + 86
572 / F / up : 651 + 86
572 / F / up : 506 + 255
572 / B / up : 416 + 86
As I wrote this down, I may have found a clue I need to check about, the small resist is alway 86, so the partial resist may be apllyed before anything, but then the 86 are to be linked to the 856, so near 10%, then the 255 resist are ~30% of full damage. So if someone may explain this to me I'll be thanksfull.
As for my conclusion I've found that at 541 def we are at ~41% damage reduction, and then at 572 def we are at ~46%.
Further information on this later, I need to convince my mage that I really need him blasting me for hours and wait for my 1 minute CD ^^
I did some testing a little while ago on PTR, and concluded:
each point of Defence above 400 gives +0.15% additional (eg. incremental) IBF damage reduction.
I tested up to 596 defence and it was linear.
So at 600 you are 50% IBF, and 400 you're at baseline learnt 20%.
Test screenshots and so on here, and here. (Two posts within one thread).
Tarrke, your test results are consistent with my findings, so I think we can say it's proven fact now.
Only remaining question is whether it remains linear above 600 defence.
Again, Gravity, testing method is meaningful.
Tarrke, something to remember is that depending on what sort of damage you are taking, and how consistent it is, there are other modifiers that may come and go. For starters, Frost presence currently reduces all spell damage (read: non-physical) by 15%, though with much of this the system is not fully explicated as for the order of calculations. Like you found with the partial resist. I expect they've changed the magic resist system some to go along with all the other tank changes they've made.
IBF, I'm fairly certain from observation, is applied after the fundamental passives, in other words, if you're taking a physical swing, and your armor reduction is 50%, and your IBF is a 40% reduction, you will not take only 10% of the hit, you'll take 40% of 50% of the hit, or from a 20k hit, you'll take 4k damage.
There's also racial passives that become very easy to forget because they're not represented anywhere but in your spell book, though that shouldn't be an issue with arcane damage since no one has a natural "resistance" to that.
We need to keep a close eye on gear/talents and the like when doing these tests. It's easy to think of them as baseline, but they may have varying implications depending on where and how they're applied in the final calculation. For example, the effulgent meta gem offers 2% magic damage reduction, Frost presence provides 15%, Magic Suppression gives 5%. I'm not entirely sure yet if they're applied multiplicatively (2% x 5% x 15% = 23.2%) or are added in a factor later applied (2% + 5% + 15% = 22%). I don't know the formulae so rather than assuming baseline behavior (something that will be consistant as you change other numbers), we need to remove them as factors and make sure we know all of the variables in play.
One thing I do like is that your results point towards one part of the calculations, being that partial resists may be (incorrectly?) calculating before some part of our damage reduction mechanics.
I'd love to believe that the scale is linear and basic, though I'd be a little surprised it it were. At any rate, let's get some more controlled data collection, and be sure to note every detail of the spec and gear you're using, along with race and any other bits you can pick out.
For consistant damage, the general thought has been spell damage instead of physical for consistancy. I like the arcane missle idea, I've also seen people use lava, and the like.
I'm pleased to find an other human beeing that IS as nuts as me on tests protocols, and for that I may never thank you enougth. Ok that said, I've found a friend to play with. Here is the entire protocole for that test :
I'm a human DK w/o Frost presence (at the beginning of the test). I was spec into frost, and the spec used was this one : World of Warcraft Europe -> Info -> Chevalier de la mort -> Talents
I was fighting a human mage and his spec was this one : World of Warcraft Europe -> Info -> Calculateurs de talents
He wasn't wearing any Meta and I was wearing (or not depending on equiped slot) the meta with stam and armor (the other one is bound to my resist magic stuff). The only button he pushed during the whole test was Arcane missile, not any other one. No stun, no frost something, nothing else. So I wasn't ever caught in a snearing effect so nothing was ever there to amplify his dmg on me, or from me to reduce his damage (except obviously from IBF).
I've conduct some testing in frost presence and some others in blood presence. The blood tests were to find out what is the scaling of IBF with def. The tests in frost presence were to find how the magique dmg reduction and IBF were cumulating each others.
I'm pretty sure I've missed nothing, but, hey I'm human as stated above. If you find something in that protocole then please tell me. Nothing is perfect, and controlling all conditions is hard sometime, but this time, I think I've got everything under control.
And on the how they code their magic resistance, no even a clue at this time...
Always good to be taking note of every condition, often enough you can find the root cause of deviations from expected patterns. Thanks again for the testing. =)