+25 agility to 2H weapon

[Concillian]

7 Strength = 14 AP = 1 DPS :: 10 Strength = 20 AP ~1.5 DPS
10 Agility = 0.5% crit ~1% DPS

Because Flurry currently applies to hits and not just swings, it essentially gives you  2 extra hits when you crit, so 1/2% crit is about 1% DPS. Of course, this will change in 1.10, and there are some flaws in this math because of the hit penalty on DW, but I try to stack up +hit gear to compensate.

If you accept the above math (and don't sweat the 1.10 changes yet), Strength is a better than Agility on a 1:1 basis as long as your DPS is below 150. Above 150DPS, agility is better 1:1 than strength for a warrior with 3/3 Flurry.
[right][snapback]104165[/snapback][/right]

It's more complex than that. Actual DPS gain from STR is more than the flat gain from AP. Your AP is applied on instants and with Bloodthirst. It can be anywhere from 1x to over 2x that gain depending on the build and what you're fighting (Damage reduction from armor). Also, your measured DPS takes into account damage reduction from armor, but the AP calc does not. You're not comparing apples to apples in this comparison.

Consider one build with 1% higher crit than the other but the other has 20 higher AP. The one with 1% crit will get 1% more flurries (and flurry is 30% increase for 3 swings, so ~0.9 extra attacks, not 2 extra attacks, and you have to varieties of flurry bonuses, one which gives bonus to 2 off-hands and 1 main hand, and one which gives 2 main hands and 1 off-hand), the one with less crit will get fewer flurries, but each time flurry procs anyway, they are doing more damage. Each benefit has advantages and each has disadvantages.

Also, the extra % crit will increase the likelihood that you will cut short your own flurry (2 crits in a row effectively negate the flurry bonus from one of your crits), so the high AP gearset will get more complete flurries. Thus there is a factor to the extra crit that will effectively decrease your increase in damage.

+1% crit is not +1% damage either, even on white damage. Consider a 25% crit rate....

100 swings, 75 hits and 25 crits...
total damage: 125 * whatever the average swing damage is

100 swings, 74 hits and 26 crits...
total damage: 126 whatever the average swing damage is
126 / 125 = 1.008 (0.8% increase, not 1%)

Also, if you have deep wounds in your build, that one skill highly favors AP in a dual weild scenario, where crit per unit time is more important than crit % chance. Each crit resets the timer so the Deep wounds won't tick until 3 seconds beyond the time you crit. If you often crit once, then crit again 2 seconds later, your first deep wounds gave 0 extra damage, because it never had a chance to tick, and the second one will wait another 3 seconds to tick damage. A high crit chance with fast weapons will rarely even see a tick from deep wounds, because you're always resetting the timer.

There will be a "right" balance of the two numbers, but it's impossible to determine this balance with such simple math. Warrior DPS is not a simple equation. The only way you can say that DPS gain from X AP = DPS gain from Y% crit is through empirical means, then reverse-engineering the equation from that empirical data. "Forward-engineering" the equation involves too much speculation on what each hit will do.

The statistical variation of hits and crits is difficult to model. Especially in a DW flurry build where you have different amounts of damage bonus depending on which weapon crits. You have an off-hand that crits more often than main hand, which means you'll have more 2-main hand / 1 offhand flurry bonuses than 2 off-hand / 1 main hand flurry bonuses. Does this mean that it's better to have a slow main hand and fast off hand than two fast weapons? I have no freaking idea. I also have no idea how to approach modeling that mathematically without at least some data as a starting point.

It's one of those things where the more you know about it, the more you realize you don't know about it.