I think he means that by going with the hit cap you lose a negligible amount of DPS. Actually the difference is quite big (approx 1k DPS for me, I play at 12%) but I understand that for some peoplpe it doesn't worth the gameplay changes.
Different gameplay for different folks.
Playing with hit cap is almost as bad as stacking mastery after 4717 haste. The RNG just isn't as high as people want it to seem to be. Maybe if you go as low as ~9% but sure not at ~12%.
Wouldn´t a valid point in going for hit being low valued in sims it being because it doesn´t take proccs really in to account. Sure they are modelled correctly but simcraft doesn´t "use them" smartly. Being reactive to that should increase the value of hit rating unless i´m off.
6637 is the last haste cap for now? (i.e. after it I should go to mastery?)
While hit cap might be needed for optimal play I'm very sure that most people will actually benefit from not going for hit cap. That sort of min/maxing with procs just doesn't work in practice and I doubt anyone is able to pull it off right now. The way SimCraft is modeled is by far the easiest way to approach this spec right now, at least with 4 piece when the times you have no procs/CDs up is very low.
Later on there will be no discussion whether we'll hit cap or not since we'll have so much stats on our gear. Right now I wouldn't recommend it though.
Why am I getting stat weight results with haste being so good?
My char is sitting around 4750 haste and 6740 mastery.
Why do I get the feeling that if I reforge into haste and run the sim again it's just going to tell me that mastery is now the strong stat and I should reforge into it.
It's also irrelevant because look at the difference in your stat weights - .09. If you converted all 6740 Mastery into Haste the difference would be 600 DPS just based on stat weights, nearly nothing in the scheme of things (in reality it would be negative DPS because stats interact with each other multiplicatively).
My current stats are 6643 haste and 5428 mastery and the graph tells me to reforge 800 haste into mastery in order to gain 207 DPS, i mean who cares about 200 dps when u're doing above 100k dps ?
After getting 6637 threshold is worthless capping haste or not?
Sim #1 (10k iterations, low lag, patchwerk, player skill elite, scaling mastery + haste)
Stats: Hit 4078, Crit 3145, Haste 4719, Mastery 8932
Scale Factors: Haste 2.91, Mastery 2.54
Sim #2 (same settings)
Stats: Hit 4096, Crit 3145, Haste 6769, Mastery 6864
Scale Factors: Haste 3.03, Mastery 2.67
Then there's fights like Will of the Empereror where the normal stat priority will do great as well as full mastery builds will.
Seeing how theorycrafting goes for Affliction and how most of optimal play is just theories and something noone is able to pull off in a raid environment AND the fact that we don't know how big the gain actually is I think we should just trust SimCraft when it comes to single target. For multiple targets it will be depending on so many things that we just can't say what's better until it has been proven with math and evaulated how hard that is to pull off as a normal player and how much better it would be versus the easier safe route.
I know it's frustrating for everyone who don't spend time theorycrafting and play on a casual level trying to come here and get tips. But believe me, it's even worse for us who try play on a high level where we are able to handle fight mechanics perfectly but need to squeeze out that extra DPS that is needed to kill the boss. As we try maximize our gear and output it's super confusing when we can't tell for sure if we should hit cap/not and go for haste thresholds then stack mastery or just keep both stats even.
We have a long way to go before it will be set in stone what is better stat wise on different fight styles. Until then all I can suggest is using SimCraft for guidelines and listen to people who play on a high level or are just known in the community as good theorycrafters when it comes to ideas for how to behave in a raid. This spec actually has that high of a skill cap. Much higher than I had anticipated when MoP patch came in Cata. (It will be funny if Blizzard nerfs us for the OP single target spec we are then we found out we've been playing the spec wrong and remain a top 3 spec. The guy who finally figures this out could become a real hero.)
To put everything in short of what I would recommend I'd tell you to go through these steps:
1) Find out what amount of hit is good for you. This tier hit cap is most likely something most people won't obtain until close to full BiS. If you're fast at reacting to misses and feel comfortable playing with less hit then try find the best value for you. Otherwise just go for that cap. (I personally have went for 12% hit most of this tier, but I've had full raids where I've been hit capped because I've been tired of missing and it has made me lost my focus.)
2) Your next secondary stats to look at will be haste and mastery. Don't ask anyone what you should choose. First reforge your gear to 4717 haste and then as much mastery as possibly. Run this in SimCraft with plot/scaling. Now reforge your mastery to be 2-300 rating above haste until you can take haste thresholds (mainly 4717 and 6637). Run SimCraft again with plot/scaling. First use the reforge that gave you the highest DPS result. After this check the stat weights and add/remove 100/200 of each stat (haste/mastery) until they are as even as you can get (this last step is probably just gonna give you couple of hundred dps more in theory, but hey everything counts).
As a final word I'd just like to ask everyone who will keep running 4717 haste->mastery until absurd mastery levels and even encourage others to do so as well to please provide some proper theorycrafting instead of "this works for me I'm 12/14H and top 10 on most fights". All your results show us/you is that you're 1) really good or 2) had good RNG/were padding numbers. We need a huge sample of stats to prove that anything is better than other options. A couple of parses/high level players can't prove shit. Until this has been done SimCraft is what we all should rely on for now.