Double Attack - Triple Attack

[Spider-Dan]

Byrthnoth is correct. Any person who says that 2A does not have diminishing returns would also logically have to believe that Haste does not have increasing returns, for exactly the same reasons.

On a 300 delay weapon, Haste+5% removes 15 delay. It doesn't matter whether that is 0% Haste -> 5% Haste or 75% Haste -> 80% Haste; it's still Delay -15. The obvious difference is that in actual application, going from 75 -> 60 delay is a hell of a lot more significant than going from 300 -> 285 delay.

[Shinryuu]

I think technically haste doesn't have increasing returns in the classical sense (Economics major here, we invented the whole ordeal). Increasing returns are for exponential functions until whatever amount, but haste cannot go past 100% ever.

[Shamaya]

Yeah, as long as we all understand what the other means, it doesn't matter what we say. The mistake I don't want newbies to fall for, as they often do, is thinking they're going to get the same increase from a double attack piece on their war/sam as their mnk/nin, etc etc etc. So when noobs ask the question as to how double attack works, the important thing for them to understand is that you get diminishing returns as you add more, as with so many other things in the world.

More importantly, I thought d.a. for WS was always on the first two hits. Is there any source available that verifies this over "it can proc a maximum of two times"?

[Liquidedust]

Yeah, as long as we all understand what the other means, it doesn't matter what we say. The mistake I don't want newbies to fall for, as they often do, is thinking they're going to get the same increase from a double attack piece on their war/sam as their mnk/nin, etc etc etc. So when noobs ask the question as to how double attack works, the important thing for them to understand is that you get diminishing returns as you add more, as with so many other things in the world.

More importantly, I thought d.a. for WS was always on the first two hits. Is there any source available that verifies this over "it can proc a maximum of two times"?

Pretty much only anecdotal evidence, but there was a discussion/test about it in regards to Guillotine and Raging Rush here I think 1 or 2 years ago (I'll see if I can find and dig out the old thread about it).

[Hookah]

To say that it has diminishing returns implies that the more you add the less you get, which is not at all the case.

While it is mathmatically correct, it gives the wrong idea to those who don't know any better.

It's a poor application of the phrase.

[Shamaya]

Well I respectfully disagree. But perhaps because I usually say "decreasing marginal returns."

[Hookah]

Yeah, as long as we all understand what the other means, it doesn't matter what we say. The mistake I don't want newbies to fall for, as they often do, is thinking they're going to get the same increase from a double attack piece on their war/sam as their mnk/nin, etc etc etc. So when noobs ask the question as to how double attack works, the important thing for them to understand is that you get diminishing returns as you add more, as with so many other things in the world.

It all depends on whether or not you understand what the phrase really means.

As I said, those who don't know any better (IE those newbies that you refer to in your post quoted) are more than likely going to assume that it means the more you stack the less it does, as in the more DA gear you stack the fewer DA's you actually get. That is totally untrue, the effect does exactly what it says it does.

The phrase itself has a poor connotation. To simply tell someone it has diminishing returns without any further explanation is really giving bad advice, you cannot assume everyone is going to know what you're talking about.

[Masamune]

More importantly, I thought d.a. for WS was always on the first two hits. Is there any source available that verifies this over "it can proc a maximum of two times"?

Not as far as all the researches i did on the subject + the thread we got with CDF, but could be in theory testable with PentaThrust like Byrthnoth suggested:

AutoWS non stop on Fortifications, with /echo KParser: my TP return=<tp> after each WS.
Sort each type of DA'd TP returns in 3 columns "No DA" "1x DA" "2xDA".

Once get a significantly high number for each DA'd TPreturns, search for a mob having exactly the HP for 4 WS hits (DA'd or not doesnot matter, 4th hit HAVE to kill it).

Once get significantly high numbers in that 2nd test also, Compare results for "2xDA".

Comments: While in theory may sound ok, there r a few flaws in practice:
How to make sure while sorting TPreturns, we don't mix "No DA" with a WS that missed 1 hit but got one DA, or a WS that missed 1 hit but got 2 DAs in "1x DA".
or
in 2nd test, what says that a WS could simply land 3 normal hits w/o a single DA, or missed 1hit but DAed 2x ?

Unless someone get a better idea reliable to test DA occurances without mixing missed hits and DA'd hits, we have to admit this conclusion:

Conclusion: Giving the benefice of doubt, better follow the postulat of "2nd DA can occur once on ANY additional WS hit"

[CDF]

Strictly speaking, you can't "prove" that DA procs only on the first and second hits only, but you can be a reasonable person and accept this claim in light of other plausible mechanisms being very unlikely given collected data.

Regarding DA interaction with multi-hit weaponskills:

Given these hypotheses:

(1) DA may proc only on the first and another hit (probably second) of a multi-hit WS, and
(2) DA may proc on any of the hits, subject to the condition that DA procs two times max, and
(3) DA may proc on any of the hits, subject to the condition of 8 hits maximum

You can do a probability exercise to obtain, for each of the above, the probability distribution of the number of hits of any multi-hit weapon skill and then collect some data (TP return) to see which hypotheses are obviously wrong with fairly small sample sizes than just focusing on 7-hit Penta Thrusts (or looking for the non-existent 8-hit).

Actually, I already did this after I came across this thread discussing TP return for Penta Thrust given 95% hit rate and 10% DA. The results (n = 196) were:

3 hits: 3 (.015)
4 hits: 42 (.214)
5 hits: 120 (.612)
6 hits: 30 (.153)
7 hits: 1 (.005)

Hypotheses (2) and (3) are similar, so I compared (1) with (2) for Penta Thrust given 95% hit rate and 10% DA. Here's an illustration of the probability distributions under (1) and (2).

You can see that the probability of a 6-hit return is much higher under (2), and the mystical 7-hit return would be about a 6 in 100 event if (2) were actually true. The above data obviously shows that (2) is really unlikely that DA can proc on any of the hits of Penta Thrust w/ a 2 DA max. You don't really need to use a statistical test to see that.

So, being a practical person, I assert (1) is the reasonable conclusion without collecting a bajillion samples.

[Cadsuane]

The mistake I don't want newbies to fall for, as they often do, is thinking they're going to get the same increase from a double attack piece on their war/sam as their mnk/nin, etc etc etc.

That is the case, actually.

[CDF]

In this game, the relative efficiency of anything depends on initial conditions. I doubt anyone ever said you would get the same percent increase from a double attack piece on WAR/SAM vs. MNK/NIN. But I see people are obviously confused when they say "the more DA you stack, the worse it gets," which is nonsense.

Due to the nature of damage calculations, which just involve multiplying a bunch of factors together, sure, there will be situations where changing one factor will be more efficient than changing another.

For example, in the dark ages of no Marinara Pizza, it could have been argued that meat was more efficient than sushi even when hit rate wasn't capped. This is was not because of "diminishing returns" per se but because, at some point, additional attack became more efficient in the presence of high levels of accuracy than additional accuracy added on top of existing high accuracy.

[Cream Soda]

"the more DA you stack, the worse it gets,"

Again, like my dollar analogy on page 1, it's still the same increase, it's just that increase is worth less of a % increase to your damage when you have more to begin with.

[Cadsuane]

Again, like my dollar analogy on page 1, it's still the same increase, it's just that increase is worth less of a % increase to your damage when you have more to begin with.

Whatever grows by the same % per point is growing exponentially even if you want to call it a "constant percentage increase". However the correct (and actually useful, non-relative) definition of return demands that we refer to that as increasing or exponential returns just like it demands that we call it a constant or linear return in the case of DA.

[CDF]

Again, like my dollar analogy on page 1, it's still the same increase, it's just that increase is worth less of a % increase to your damage when you have more to begin with.

You know it's not worse, merely a smaller proportion of total damage if you have a higher base quantity to begin with, but those who conclude "the more DA you stack, the worse it gets," don't get it. In a vacuum, the dollar analogy is trivial--nothing wrong with adding more of a positive thing without regard to opportunity costs and trade-offs is fine, except there are usually opportunity costs and trade-offs.

[pahnphoenix]

the reason for mentioning that each further point of DA gives a lower % increase is that the value of a piece of equipment is the % increase in damage it yields. this is how you compare pieces of gear. people who cry about this well-explained, obviously relevant point have something up their ass.

(also, the way haste works has misled a lot of people to dream about "DA builds" or to make dumb comments like "i only use pole grip /WAR". this is why it's important to explain to people that the more double attack you have, the less value it has relative to competing options, not more)

if anyone wants to argue about this, just read the alla thread and relive the good old days. you aren't going to post any points that aren't already in the thread, so i wouldn't bother.

[Shamaya]

How far into the future will we have to progress, where such simple semantic hairballs will no longer make us all seem mutually and inexorably retarded.

[Onorgul]

Until language no longer is necessary to convey ideas. Someone with 1% of your DNA might eventually experience that, but I doubt it.

[Cadsuane]

the reason for mentioning that each further point of DA gives a lower % increase is that the value of a piece of equipment is the % increase in damage it yields. this is how you compare pieces of gear. people who cry about this well-explained, obviously relevant point have something up their ass.

(also, the way haste works has misled a lot of people to dream about "DA builds" or to make dumb comments like "i only use pole grip /WAR". this is why it's important to explain to people that the more double attack you have, the less value it has relative to competing options, not more)

if anyone wants to argue about this, just read the alla thread and relive the good old days. you aren't going to post any points that aren't already in the thread, so i wouldn't bother.

I guess if the entire point of that excercise was to solidify in the minds of the swinish masses that it's better to build for haste than DA, then it was a noble endeavour even if it was definitionally wrong, incosistent with the concept of a cartesian function and supported primarily by offensive posturing, but, I'll take your advice and read an old thread for a verdict.

... says some clown who doesn't know the difference between percent difference and rate of change.

[FailureMidgard]

As I've stated before, it's only a linear increase if you are only interested in your dependent variable as damage gained, and your independent variable as double attack percentage. Realistically however, double attack percentage is not a very useful independent variable, and some made up "stat points" should be your independent variable. These "stat points" can be used to gain DA, accuracy, attack, haste, etc.

When someone says it has diminishing returns, IT IS FUCKING OBVIOUS that they are not referring to what you happened to choose. As was also used in a previous discussion when this came up, if you look at a crop example commonly used, and say the second batch of seed only produces half as much, you can either stop there, and call it diminishing returns, or you can look at the fact that wherever you buy the seed, your 2nd batch is actually half price, then suddenly the produce:dollar spent no longer has any diminishing return yet the produce:seed used does.

But, if Cadsuane was able to read any other time this topic came up, this argument wouldn't keep recurring.

Edit:

Furthermore, @ 0% DA, 1 DA adds 1 attack per 100 attacks. @ 50% DA, 1 DA adds .67 attacks per 100 attacks.
@ 0% DA, 1 DA adds 1 attack per 100 attack rounds. @50% DA, 1 DA adds 1 attack per 100 attack rounds.

OMG I made the same thing both diminishing and not diminishing. *world explodes*.

[Cadsuane]

Honestly, I'm still at a loss as to what specifically you're talking about or trying to demonstrate, which leads me to believe as I've stated before that you don't know what that is exactly.

And furthermore, that's spectacular, but conveniently misleading, since we'd be talking about attack rate which is number of attacks with respect to time.