Warriors need MORE damage

[Lyth]

I was kind of expecting to have to do this, because I knew there would be at least one person who would be obstinate about it. Here you go.

Quick disclaimer: the formatting on this is going to be difficult, so assume that integrals are taken as limits from 0 to T, where T is total fight duration.

The "expected value" of your total damage output is given by

C∫ p(t)w(t) dt

where C is the crit multiplier, p is the crit rate at a given time t, and w is your damage output as a function of time. If you're playing a job with a fixed crit rate, this simplifies to

Cp∫ w(t) dt

Xeno's argument implies that w(t) influences this value. My point is that it doesn't, so long as the area under the curve is the same. It doesn't matter if you rearrange WAR's rotation from one with heavy burst to one that has more consistent dps as long as the total damage is a constant.

Now you might say: WAR's p(t) does vary with time, because it is linked to their resource gauge. Which is correct, and is also why this is an underestimate. WAR actually is more likely to crit on higher weighted portions of their rotation than this estimate would predict. Which is why they're actually privileged, not penalised.

I know that this might seem counter-intuitive, but probability theory often is. That's why people get obsessed with perceptual errors like "winning streaks" in spite of the fact that we've proven them to be, time and time again, mathematically false. If you fail to crit on your Berserk, you'll do less damage. I get that. But it all comes out in the wash when you look at averages.

[Chrono_Rising]

I was kind of expecting to have to do this, because I knew there would be at least one person who would be obstinate about it. Here you go.

Quick disclaimer: the formatting on this is going to be difficult, so assume that integrals are taken as limits from 0 to T, where T is total fight duration.

The "expected value" of your total damage output is given by

C∫ p(t)w(t) dt

where C is the crit multiplier, p is the crit rate at a given time t, and w is your damage output as a function of time. If you're playing a job with a fixed crit rate, this simplifies to

Cp∫ w(t) dt

Xeno's argument implies that w(t) influences this value. My point is that it doesn't, so long as the area under the curve is the same. It doesn't matter if you rearrange WAR's rotation from one with heavy burst to one that has more consistent dps as long as the total damage is a constant.

Now you might say: WAR's p(t) does vary with time, because it is linked to their resource gauge. Which is correct, and is also why this is an underestimate. WAR actually is more likely to crit on higher weighted portions of their rotation than this estimate would predict. Which is why they're actually privileged, not penalised.

I know that this might seem counter-intuitive, but probability theory often is. That's why people get obsessed with perceptual errors like "winning streaks" in spite of the fact that we've proven them to be, time and time again, mathematically false. If you fail to crit on your Berserk, you'll do less damage. I get that. But it all comes out in the wash when you look at averages.

While I do not 100% agree with your expression for the expected damage, I do not believe a more detailed analysis would lead us to a different conclusion.

But this is not the reason for my posting here.

I would actually like to point out that analysis like this actually has a strict inequality for the type of functions we are dealing with, meaning that not only is this type of analysis pointing out that the lower bound is not achieved, it simultaneously points out another advantage of burst damage, raid buffs. Since raid buffs are short duration the ability to put the majority of your damage into a buff window is buffing expected damage more than a class which has flat damage through out the encounter.

This has a particular consequence, as our damage base increases through the expansion the damage between the tanks will continue to spread apart meaning we should expect the difference between warrior, paladin, and dark knight dps to separate more in the coming patches. In particular we should see warrior damage growing faster than the other tanks and dark knight growing less in comparison.

[Aana]

I was kind of expecting to have to do this, because I knew there would be at least one person who would be obstinate about it. Here you go.

Quick disclaimer: the formatting on this is going to be difficult, so assume that integrals are taken as limits from 0 to T, where T is total fight duration.

The "expected value" of your total damage output is given by

C∫ p(t)w(t) dt

where C is the crit multiplier, p is the crit rate at a given time t, and w is your damage output as a function of time. If you're playing a job with a fixed crit rate, this simplifies to

Cp∫ w(t) dt

Xeno's argument implies that w(t) influences this value. My point is that it doesn't, so long as the area under the curve is the same. It doesn't matter if you rearrange WAR's rotation from one with heavy burst to one that has more consistent dps as long as the total damage is a constant.

Now you might say: WAR's p(t) does vary with time, because it is linked to their resource gauge. Which is correct, and is also why this is an underestimate. WAR actually is more likely to crit on higher weighted portions of their rotation than this estimate would predict. Which is why they're actually privileged, not penalised.

I know that this might seem counter-intuitive, but probability theory often is. That's why people get obsessed with perceptual errors like "winning streaks" in spite of the fact that we've proven them to be, time and time again, mathematically false. If you fail to crit on your Berserk, you'll do less damage. I get that. But it all comes out in the wash when you look at averages.

Im not disagreeing with the 'expected' value. War's damage is very good. The area under the curve is large. Maybe I'm misinterpreting xeno and the related posts about war. The 'issue' being raised is not that redistributing damage from less bursty will make war do more damage or increase the area under the curve. But a redistribution would smooth out the variable performance of small RNG samples of single fights.

Extreme Scenario: (lets pretend you always hit your crit rate for simplicity)

A: Job has 1 attack GCD for 100 pot and you hit it 100 times before a fight is over with a 10% crit rate. Crits do 50% more damage. Total fight damage is 10,000 pot + 500 crit=10,500
B: Burst Job has 1 attack GCD for 10 pot and you hit it 100 times in a fight, but 10 of those hits are UltraZerk hits that do 910 potency. Total fight potency is 10,000 pot. Crits add between +100 pot and +4550 pot.

Both have the same overall potency. In an infinite number of parses steady job and bust job will average to the exact same damage. But on any 1 parse burst job could be 10,100 and 14,550 while the steady job would consistently do 10,500.

The statistical average is the same. But any 1 off can be drastically different and theres nothing you can do about it but pray the RNG is in your favor to benefit during the burst window.

This is an extreme example but the principle remains. The more a jobs damage is funneled into small enhanced windows the less reliable their damage will become. Every job has 'some' level of burst, even if its just popping a potion and hitting during a TA. But not to the same degree.

The 'issue' I see Xeno articulate is that some wars dont find this uncontrollable RNG playing so large a role in your damage on any individual instance. Not that the space in the curve isnt big enough. You cant apply the expected outcome to a singe instance when the strong burst itself is the polarizing aspect. It works out over time, sure, but it doesnt 'feel' good when it doesnt and you watch your parse go down the tubes for that run, but the tradeoff is you sometimes get the opposite man mode parses.

Your looking at the average and saying 'its fine', the complaints are aimed at the fight to fight volatility inherent in burst jobs combined with Crit/DH rng, not power. I see that volatility as right in line with war's theme and style personally, but i get why some dont like it.