For what it's worth.

Hmmm... Im thinking about it on the back of my head for a while, but I still can't formalize the problem of crtD vs crtH.

Leaving aside factors like positioning, weapon power, crth based weapons proc etc: crtD is a real value and will always stay real, i.e. given 1 crit shot you will always deal your crtD damage.

CrtH however is just probability of crit event occuring or not.

Assuming 100 shots with 10 damage per shot

With 0% crth 0 crtd = 1000 damage

Base crth: 22% crtd: 80% (180%)

22 * (10 * 1.8) + 78 * 10 = 396 + 780 = 1176

5 Locators crth + 9%: crth 31% crtd 80%

31 * 18 + 69 * 10 = 558 + 690 = 1248

5 exploiters crtd + 40%: crth 22% crtd 120% (220%)

22 * (10 * 2.2) + 78 * 10 = 484 + 780 = 1264

--

n = number of shots

d = damage per shot

crth = 0.22

crtd = 180

n * crth * d * crtd + n * (1 - crth) * d => n * d * ( crth * crtd + (1 - crth) )

assuming n, d are scalar constants the damage function becomes

crth * crtd + (1 - crth) => (crth : x, crtd : y) =>

Our final function of damage over crth and crtd

dmg = x * y + (1-x)

--

Lets take 3d function plotter, as im too lazy to do derivative calculus for 2 variables:

http://www.math.uri.edu/~bkaskosz/flashmo/graph3d/
x * y + (1-x)

Parameters:

x = 0 - 30, y = 0 - 220

Bummer... this is a surprise. There's a little bump from 0 to zmax, which proves the claim that there's in fact sweet spot ratio of crtd and crth. Something which I didn't belive for a while.

--

Also worth noting my friend's little experiment: he parked 9km from ISE gate and for an hour setting his weapons on autofire. One hour with locators, and one hour with exploiters. His results were rather surprising: out of somewhat 1,200,000 dmg dealt in both cases the difference was 40k damage: less than 3%, which is well within experimental error.