I have no idea what you mean about my test assuming crit conversion and crit reduction both exist. I don't see how I assume anything of the kind. I may use both terms, but I don't believe I in any way suggest that there's two separate reductions occurring - I'm fairly sure what I said, and what I was testing, was that what we have previously called crit reduction is really implemented as a forced conversion of 4.8% crits to hits. Really, I think the salient facts are as follows:

1) Against boss level mobs, our observed crit rate is 4.8 lower than our "tooltip" crit rate - that is, lower than we might "expect".

2) No matter how much crit we may have, 4.8% of our attack table will still be regular hits.

Vulajin's testing shows the former pretty conclusively; my testing is reasonably strong evidence for the second. So, what does this mean in terms of the mechanic? Well, there's two different ways of phrasing it - please note that they are completely identical in practice. There is no functional difference. They are two different ways of saying the same thing.

Crit Reduction
Crit is reduced by 4.8%, but the hit table had 4.8% hit on it which cannot be removed. Thus, the amount of room in the hit table for crits is 100-G-D-M-4.8, with the 4.8% representing the hits that we cannot get rid of. Thus, we our capped with our tooltip crit rate, C, after being reduced by crit reduction (4.8%), matches this value - that is, when C-4.8 = 100 - D - M - G - 4.8 - that is, when C = 100 - D - M - G.

Crit Conversion
4.8% of crits are converted to hits, after the crit cap is imposed. That is, we our crit capped when 100-D-M-G = C, and our hit table consists of 4.8% hit, C-4.8% crit, and the expected D + M + G of other stuff.

Note that either way, the crit cap occurs when C = 100 - D - M - G; note that our observed crit rate in this case is 100-D-M-G-4.8. This is different from the prior theory, when we believed ourselves capped when 100-D-M-G=C-4.8, by 4.8%.

So, yes, our *actual* crit is 100-D-M-G-4.8 when we're capped. But this is when our *tooltip* crit satisfies C=100-D-M-G. And this is only 4.8% lower than it was before.

Also note that either way, the theory is consistant with the observed data - positing that the crit cap is otherwise does not - we *observed* 100-D-M-G-4.8 crits, meaning that we would be capped with 100-D-M-G crit. And I really don't see how any other interpretation can be valid if it doesn't give the same crit cap, given that the data is an excellent fit for this one.

Like, consider your test; if we have 35% left for crit after dodges, misses, and glances, and we test at 30% crit, what do we expect to see? Well, we expect to see our crit reduced by 5% to 25%, and thus 10% misses. There's the 5% forced reduction/conversion/whatever-the-heck-you-want-to-call-it, and the 5% we didn't even try to remove because we didn't have enough crit. I'm not seeing any reasonable theory that proposes otherwise, so I'm not sure what testing in this case would actually demonstrate.

[edit] Whoops, you can feel free to delete this. My apologies.

I've been thinking about this crit conversion effect for the last day or two, and thought up what might be a posible explanation for the 4.8% crit to hit conversion we are seeing.

What if the effect is caused by bosses having resilience?

A lvl 80 player would take about 470 Resilience to get 5% crit converted to hit. It wouldn't be unrealistic to think then that a lvl 83 mob with 500 resilience would have about 4.8% crit converted to hit.

As I understand it, This mechanic is in game already and could easily have been used to reduce dps from excessive crit scaling. The fact that resilience as a mechanic was added after the core combat table was built could explain this crit reduction interacting with the crit cap in an weird way.

For example, if the process was:
Generate combat table
Generate random number
Check result
If Crit was it countered by resilience? if yes then result = Hit

Then we would see pretty much exactally what has been observed.

The only thing I am not clear on is what happens to say a player with resilience getting hit by someone thats crit capped. Wowwiki only uses the example of a player with 100% crit hitting someone with 5% resilience and beign reduced to 95%, but what abotu a player with 105%?

I would think that a rogue or DW warrior attacking a Prot pally or warrior would do it. Unfortunately As a resto druid I don't have access to either to test with.

I also did some brief testing with OP at 103.5% crit on the most recent PTR testing the "warrior's only have ~3% crit depression" theory. I stopped after I saw ~2-3 hits, proving it incorrect, and just assumed it was most likely the same 4.8% depression rate. Hellord's data with OP testing also shows a test with 104.33% crit and has 1 hit in 127 attacks.