I'm hitting the target from behind with these stats

crit: 51.40% (cruelty, axe spec, 1037 rating, 789 agi, berz stance 3%)
hit: 0.76% (25 rating)
exp: 9 + 1/2 WM (Soa, orc racial) 3.25%
2x1.5s axes

So I should be left with 46.51% of the table that should be 4.8% hits and rest crits.
Actually I'm at 2400 attempts and 0 hits and 46% crits.
I'll get a bigger sample and way to a hit if even possible and I'm also logging the fight so I'll update this post later.

EDIT: I'm using this gear and specs with 40 crit food.
Had to use both arp proc trinkets for the crit so I'll clear up the log.

I'm not getting any hits anymore either.

2000 swings from behind

+3% hit rate
~55% crit
+0.5% expertise

24% glance + 24% miss + 6% dodge gives 46% crit room.

0 hits.

What this basically means is that this was fixed and the old mechanics were unwanted. We should probably retest all hit table assumptions again.

12,500 swings with 262 hit rating and 131 expertise rating, both being the highest integer values possible without reaching either assumed cap (ferals gain 10 expertise from talents). This gives a roughly 0.01% chance of a miss and 0.005% chance of a dodge. Ended with 1 miss and no dodges (ignore the 2 parries, those happened while moving my character to keep from going AFK). Pretty strong evidence that hit and glancing are still exactly 8% and 24%. Later this weekend hopefully I'll be able to test some more and get a dodge.

The second log is truncated to where I brilliantly decided to remove DC:Greatness so I could test crit reduction. My character sheet crit chance was 47.96%, hit and expertise unchanged. 43.2% actual crits observed, so 4.8% reduction looks pretty good there too.

Overall report - World of Logs - Real Time Raid Analysis

Crit reduction report - World of Logs - Real Time Raid Analysis

Edit: Added screenshots of my stats and Recount.

Got a dodge after another ~8500 swings at the target dummy, so 6.5% dodge chance looks unchanged as well.

World of Logs - Real Time Raid Analysis

So, I just tested on live this evening, attacking from the front with (tooltip) 47.23 crit, 21.09 miss, 3.09 dodge, 10.59 parry, and 24 glance (for a total of 105.00%). Thus, after crit reduction, I should still be crit capped. And in 1001 swings, I saw 52 regular hits. Which would imply that the behavior has *not* changed. So, um... yeah. Something strange is happening.

Are you looking at WMO/WOL logs or Recount? Recount doesn't separate hits from blocks, which confused Landsoul and I for a bit when we tested. Attacking from the front while crit capped gave me no hits but 4-5% blocks. Attacking from the back while crit capped gave me 0 hits (0 blocks as well, of course).

Edit: To clarify, Recount does count full blocks as a separate swing outcome, just not partial blocks. The target dummies only block 76 damage so you obviously don't see any unless you're not wearing any gear. Also should be noted that my version of Recount is very old and newer versions may recognize partial blocks.

Either I wasn't clear or I'm confused. When I read hellord's post saying that this may not be an issue anymore I tested by attacking from the front, and Recount showed about 5% hits. However, when I uploaded the log to WOL and checked it I found that I actually had 0 hits and those swings that Recount labeled as hits were actually blocks. The only way to remove blocks from the hit table is to attack from the back, so as far as I'm aware this has no bearing on crit depression, right?

Oh, I see what you're saying - it's not a parse bug, but a "there's more crap on the hit table" issue. Upon retesting from behind, I am in fact able to push all hits off the table and get nothing but crits, misses, dodges, and glances. Which confirms that the crit cap is again 4.8% higher than we've been recently thinking.

Honestly (and I'd have to review the testing carefully to be sure) I almost wonder if this whole crit conversion discussion was always just an artifact of testing from in front and forgetting about blocks. I have a hard time believing that they changed something like this without posting about about it at all.

From a quick glance at the OP and follow-up posts, there's no mention of blocks in the original parse, so I think that's probably the case. It just demonstrates how much we depend on our tools to be accurate, and what happens when they aren't!

I'm ashamed I completely forgot block. Recount even in the newest version (updated after 3.2) doesn't show blocks unless fully blocked. At the time of the tests I didn't have the gear to double check the depression from behind the target, however now that is more or less proved we can safely say the value you must obtain to cap crit is .76-misses - dodges + .048.

The developer of Recount wrote a comment on that topic in the WowAce forums:
Recount - Page 328 - WowAce Forums

edit: