2 DMG. You'd need some pretty sad attack to make the 3% crit work, especially on a WS that has already boosted crit rate.
For highest average ws damage, that's probably true, but that doesn't necessarily mean it's the best to use.
A while back (year or so) milich mathed out cletine v thala (3 dmg, 2 dex, 12 delay, 3% crit) and cletine won provided you could still keep 6 hit or something like that. Obviously Thala will ws harder, but since the crit on cletine is never "wasted" it pulled ahead for overall damage. Obviously this is not the same thing being 1 handed v 2 handed weapons, but I'd imagine someone has already done the math here somewhere, maybe a ninja forum regular will link.
Except those are very high base damage weapons. It's a different story when the weapons are less than half of that base damage.
yeah, the point was that thal has a higher WS average, but cletine would do more total DoT (assuming you use drakesbane; if you use penta thrust--which of course can't crit--thal's WS advantage would tip the scales to thal). incidentally, though i seem to recall mentioning that the comparison was only valid if it didn't mess with a DRG's 6-hit setup (true or otherwise), i suspect that sTP and delay might negate what i wrote... i don't know though (it's hard to care, since fay lance and v.fork are usually better than both lances).
anyway, regarding the OP's question ("which will have a higher AVERAGE WS damage" caps lock key his/hers), 2 damage is the clear winner over 3% crit. brief illustration: [edit: well, i thought it was a clear winner b/c i thought 500ATT was low and suspected jin would usually have around base damage = 80... i'm thinking virtual equality on WS is more likely given realistic stats]
say you have weapon damage = 38, fSTR = 10, WSC = 52 (that's with 120 STR and 90 DEX; dunno if that's realistic, but it won't make a huge difference with such small mods), you'll then have 100 base damage for each hit on jin. so, 2 more base damage is a flat 2% increase to average WS damage.
here's a somewhat realistic lv82 gcolibri situation that's a bit generous to crit (b/c the ATT is kind of low, and means more vs lv82 than lv81). if you have 500 ATT, your cRatio is 1.179, so a crit raises damage by 85%~. .03 * .85 = 2.5%~, so if your base crit rate were 0%, 3% crit would win. up your crit rate to 25% and the increase drops to 2.1%~.
so, if you have low ATT, 3% crit will realistically hover right around the increase of 2 base damage (if you drop STR to 100 in the above example, the comparison doesn't change much; as long as you're not WSing in like 75 STR 75 DEX or something, 2%~ about what you get from 2 damage). if you have 600~ ATT, a crit becomes about a 66% increase, and wouldn't beat 2 damage even if you started at 0% crit rate.
incidentally, the base damage vs crit comparison favors damage more strongly when you're talking about melee DoT. this is because your base damage as ninja is usually a bit less than 50 for melee. 2 base damage is a 4%~ damage increase.
3% crit should never increase your damage that much. it's easy to find when 3% crit would raise damage by 4%. you saw above, i multiplied .03 by .85 to get .025 (2.5%), the damage increase you get from going from 0% to 3% crit when a crit raises damage by 85%. so, high school algebra gives us .03 * X = .04, where X = how much a crit needs to raise base damage for 3% crit to give you a 4% increase. .04/.03 = 1.333. in other words, a crit needs to raise your damage by 133% for 3% crit to raises average melee damage 4%. what would your cRatio need to be for this to happen? find it via 1/X = 1.33 (you're adding 1 to your cRatio with a crit, and you want the increase to be 133%). so,
1/X = 1.333
1 = 1.333X
1/1.333 = X
.75~ = X
so, you'd need .75~ cRatio to get there. you'd get that with about 360 ATT when fighting level 82 gcolibri (lower vs their lowby friends; about 340 ATT).
the upshot of all this is that getting a fudo with +1 damage makes senji obsolete. augmented fudo / perdu will beat senji / perdu, b/c they WS roughly equally (with the edge going toward aug.fudo / perdu if your ATT is decent), while the former always does more melee DoT than the latter.
edit: bsphil's post wasn't there while i was writing this one. interesting that we posited such similar values in our hypotheticals.
edit2: silly "homework" question for anyone reading the above as a little ffxi damage tutorial (as i wrote it, hoping it's helpful to someone for other damage comparisons): what's weird about the numbers i used for base damage? if we're talking gcolibri (either level), they're a tiny bit off given the parameters i stipulated, though making it realistic doesn't change the comparison at all (so i didn't fix it). Edited, Dec 14th 2009 7:52pm by milich