Exemplar's spreadsheet does show haste as being way more valueable right now, but both simulationcraft and redcape's spreadsheet still show it as the lowest.

In fact, after messing around with Redcape's for a bit, I was able to make haste jump in value incredibly by setting latency to 0.2s, anything between 0 and 0.1 and 0.3 onward had haste as the lowest stat. On simulationcraft I was able to increase haste's value by setting the latency to the highest and reducing the player skill to average (although, in this case, haste got closer to crit, but didn't pass mastery).
After looking at your spreadsheet Exemplar, I saw that latency is set to 0.2s and changing it has no impact on the dps or stat weights.
So, my guess is that around 0.2s of latency, haste seems to get a jump, maybe because of some relation to reaction time and CS cooldown.

I think we need to have a closer look into haste's value right now, as it is a known stat to have ups and downs.

Under my old model (the one where I hadn't yet broken autoattack and Censure), even reduced to 0.1 second empty time it was a DPS loss to not shove in an additional filler. However I didn't go as detailed as specific ability breakdown. My gut intuition (which could easily be wrong, which is why I'm listing it as such) is that with sufficient Haste it may be worthwhile to skip HW and Cons. So any 2 of TV, HoW, Exo between CS. If we start approaching this point we will probably want to math out Zealotry to determine whether a pause or a non-TV is most advantageous. This much haste should still be at least a tier, if not two, away.

Naididae - yeah, Delay's become somewhat obsolete in the spreadsheet since Blizzard fully implemented the latency fix. It's presently not impacting anything at any value. It's only important if there's significant gaps between abilities where we're waiting for a button to press (such as under a 1-filler system). Something that is presently few and far between, but could become meaningful if we closely approach a 3 second CS CD and may voluntarily skip something. If AoW procs and you still hit something else first, the GCD is always shorter then melee swing timer, so you still have time to press it before you possibly lose a new proc (unless Exo is low on your priority, in which case things displace it anyway and a slow reaction time is even more moot). Even if it's taking half a second for DivPurp to appear as available to your UI, that gives you a whole second before the GCD even completes to decide to use it over something else which could be blocked from its own DivPurp proc.

Delay's just not useful at present, unless you want to model someone who is not pressing buttons during GCDs - and that's a GIGO model.

I agree that we should investigate things further. Redcape's retired and no longer updating his spreadsheet. Rawr is Beta (from what I can tell the only reason it's not Alpha is because the software functions for other specs). I'm under the impression Simcraft is still heavily a work in progress for Ret, as they've had no Ret contributor. Even so, I've had several PM over the last month with folks stating Simcraft was showing Haste > Mastery > Crit - perhaps it's setting or gear related. I'd feel more comfortable if there was a fully functional competitor to my spreadsheet in order to directly support or refute conclusions.

Edit: Indeed, looking at my logs my Censure is about half what my spreadsheet predicts. Wowhead is showing a different coefficient for Censure then what I'm using. Unsure when this may have changed, as I check coefficients compulsively when a spell is listed as having changed (checked everything a dozen times after PTR shut and Cata launched). Plugging in Wowhead's coefficients drops below what my log provided. In short, time to go back to the training dummy. At least it's an easy one since its value will only vary within 1 damage.

We've established that breaking the t11 4pc bonus early without a t12 2pc bonus is a dps loss,
I was wondering then, when to break our t11 2pc(372) bonus and go t12 3pc+ or equivalent(378) in order to minimize our dps loss from losing a 10% bonus to TV.

TV, being our #1 damaging skill overall in a single target situation, would take a big hit when we lose that 10% bonus, which I do not believe going from ilvl 372 to 378 on a single item slot will make up for.

Would it be a better choice to stick with our t11 2pc bonus until we can at least gain a t12 4pc bonus?
Or breaking a 2pc bonus would still be better if replaced with a heroic (391) piece?

PS. I eagerly await Examplar's update on his new findings on haste, as it will indeed have a titanic impact on how we select and reforge our gear at this juncture of progression into Firelands.

I don't understand how the fix could have changed haste's value so much.
What I mean is, of course it had an impact on all fight lengths, except for the standard setting (which I think was 300?), and we could see that from the dps breakdown and overall excepted that were very near to what we see in reality.
So, what would be expected wouldn't it be for the dps to stay the same with the standard setting and just have it not take a gigantic dive with higher lengths, thus having the rating values stay around the same range on all fight lengths?
To summarize: why did this fix increase haste's value so much even on the standard setting which was working properly even before the fix? Edit: or does the graph rating value consider some kind of average for each fight length, instead of just the length you have selected.

Autoattack and Censure were flat wrong, even at 300 seconds, unless you had exactly the weaponspeed and Haste rating used when the # of autoattacks and Censure were calculated. And by exact I mean precisely the same Haste Rating + buffs.

If you added Haste you did not receive more autoattacks or Censure. Removed Haste and did not lose autoattacks or Censure. Autoattacks and Censure are still a reasonable portion of our overall damage. If Haste does not impact them at all, then Haste is undervalued. Ditto for the reverse, if you removed Haste you did not receive less autoattacks/Censure. Thus the rating for Haste was linear and not obviously wrong on a graph or with most setting changes.

Exo procs were actually being calculated properly, so that at least wasn't impacted.

I hope to have time today to test and then update Censure's coefficients in my spreadsheet and see it this causes alignment to match Simcraft (although since I previously matched Redcape, I expect Haste to at least be a bit above his values). If I'm still getting results of Haste > Mastery, then someone would need to point out another spreadsheet error to explain the issue.

Tests of what, Simcraft? Also 1.7, 1.4, 0.8 are both higher and lower than my previous values. I had closer to 1.25M, 1.05C, 0.9H in my spreadsheet. The variance between the stats was tighter - still significant and clearly obvious which was desirable, but not by huge margins and Mastery was not more than double Haste.

Although only slightly, but Seals of Command seems to be a bit off too. It's consistently around 4.2% of my damage dealt on the spreadsheet, while on logs, simulationcraft and redcape's it shows more around 3.4%.

My working formula for Censure is
or 0.05*SP + 0.0965*AP total damage done over 15 seconds per stack, before modifiers.

However, I'm not sure when this was last tested/confirmed; certainly 4.0 beta, likely reconfirmed early on in 4.0 release. It also agrees with the wowhead tooltip.

I went ahead and re-ran both tests using the built in Normal gear and Heroic gear profiles. I'm using SimulationCraft 420-2 for World of Warcraft 4.2.0 Live (build level 14333)

Normal - Simulationcraft Results

Str 2.6045
Agi 1.1363
AP 1.1493
Exp 2.4490
Hit 3.2405
Crit 1.3093
Haste 0.9697
Mastery 1.5670
Wdps 10.4649
Wspeed 326.7772


Heroic - Simulationcraft Results

Str 2.6311
Agi 1.2527
AP 1.1225
Exp 3.0213
Hit 3.7046
Crit 1.4113
Haste 1.0051
Mastery 1.7485
Wdps 10.8172
Wspeed 369.0123


So of the other reports I ran with different trinkets for testing have some values for haste reach all the way down to .500 but those are run under a older version of simcraft and the values changed after I reported an issue with it and started looking for other resolutions.

There are other sources from which you may find information on Ret's BiS trinkets, this isn't really the place for it unless mechanics of the trinkets are in question. [Ret] Trinket ratings estimation [Firelands 4.2+] has some discussion on relative trinket weights.


moderator edit: Just a note to avoid confusion -- this post was a response to a discussion which I removed. I'm leaving this post as it has some merits on its own.

Are the haste breakpoints for additional Censure ticks known? If there's obvious break points people might be seeing different weights.

Might have missed it but at what point does the haste pull ahead? Ive done some personal live tests and come about 1-2k Dps short with haste. Though raid buffs could alter that quite a bit. Also after looking at sets of gear I was unable to stack enough haste for apparatus to pro it. If apparatus price critic we may want to look at using a different second trinket.