Overall I don't like the shoe-horning that is happening either... Demonology is only a tiny bit behind on affliction, just is harder to play. Doesn't warrant everyone going around saying you -should- be demonology.
that's subjective though. I personally believe demo is easier because it's more forgiving than affli as long as you're following the basics of lining up procs/cds for your meta. also it's very mechanic based just like fire mage. most fights this tier where demo has been "better" than affli is when certain burst (mainly AoE) is needed. feng is a great example of this. same goes for spirit kings and to some degree wind lord in HoF. most of the times avoiding enrage on the boss has to do with mechanics over actual overall dps output. for something like 1/5 fights you'll have to go demo to help score a kill but for parsing affli is most likely still gonna be better.
you are completely right, whydrood, while hit is simmed lower than haste and mastery in general what ppl forget is that it is nothing more than a simulation where the best conditions are met, meaning that you refresh dots instantly after a miss, and unless you have reaction times faster than a human being(which is an impossibility) is capable of, hitcap will ultimately benefit you more. [...]
Once again: SimC has very high reaction time, exactly for things like missing a dot application. It's a exgaussian distribution with mean 650ms. I wouldn't call that "faster than a human", on the contrary.
Whydrood's Min/Max numbers are a great example of interpreting more than just the DPS number. In Simc you shouldn't use the min dps, but rather something like the 5% percentile ( to get a statistically more realiable comparison ). Comparing the variance of the DPS can be very important for high-end raiding guilds which depend on constant performance delievered. On the other hand, questioning such fundamentals as "hit is always best" is a good thing, to clearly understand the reasons for ones gear decisions.
The whole thing really is a good example showing how people usually just look at one single number, instead of taking a closer look at all the detailed statistics SimC provides. Really can't blame the program for that.