Part of an open mind is being able to accept when you are wrong.
The methodology you have described is fundamentally flawed; the dependance of dps to weapons speed is demonstratable, I have proven it algebraically in this very thread. You need to accept the flaws and stop defending your white elephant.

Yes, but no.
If you want to go average, let's go.

In the >1.5-3s range, without haste effect, the CD is equivalent to "no proc if last roll was a proc", still following me?

So the probability P of any roll to be a proc is:
P x 0 + (1-P) x 0.2
(last roll was a proc) vs (last roll wasn't a proc).
Still with me?

So for weapon in the 1.5-3 range, the "average real proc rate" is 1/6.

If i'm not wrong, "on average", with no haste effect, 100% hit on target (no parry/miss/dodge), for weapon in the 1.5-3 speed range, the influence of the CD is to lower the proc rate from 1/5 to 1/6. So taking into account the CD, the average proc rate on a big enough sample is NOT equal to the theorical proc rate of 20%

Worth 100+ posts...

I'm not saying that, I only got till now: you are wrong without explanation. The Lactose one was one good post and it has a point, but now as I was suggested already I'm going to give up with this thread. I'll write Lactose a private message for that matter.

Poor Lactose...

We've established:

In a non Stormstriking, non hasted false reality:

1) Slower weapon's WF procs hits harder.
2) in the 1.5 to 3.0 range, the average of procs is the same.

So the question is:

Does Draghkar still maintain faster weapons = higher dps?

I have never stated that. And now I will ignore this thread as well, given the input I was looking for was not the one I get.

I wonder how many more times you'll post to tell us you're ignoring us.

You do see an average proc rate of 20% outside of the 3s cooldown. You don't see a proc rate of 20% with a 1.7 speed weapon, because 1.7s is inside the cooldown. Is that clear enough now?

WF is independent of weapon speed so long as your attacks are always more than 3 seconds apart. In practice, that is never the case even with a 2h weapon because of flurry, special attacks, etc. Therefore in every real situation, WF depends on weapon speed and benefits from a slow weapon.

It is as simple as that.

You are not "challenging one of the basic assumptions of how your class works". You are, although you seem not to realise this, demonstrating a profound lack of understanding of probability.

I wonder when I wrote I was going to ignore you. I just wrote I will give up to apport my any other argument to this thred, I do believe it's different than say I'm going to ignore you.

Let's sum up. WF for a single weapon between 1.5 and 3 speed will produce equivalent dps. Modeling WF as 1 in 5 swings is inaccurate, as it needs to be 1 in 5 eligible swings. Modeling WF on a single weapon while ignoring Flurry is simple and well explained. Modeling WF on a single weapon while ignoring Flurry is not useful or interesting because it is so well explained and it is not important to explain DPS for a single wield shaman. WF cooldown exists across both weapons in dual wield, rendering the conclusions from this foray to be unusable in the case of dual wield. This has been born out in testing. I don't understand what we are arguing about anymore.

Edit: The OP has succeeded in confusing me as to what exactly his point is. If it's that weapon speed isn't so important when you are single wielding, then he's right. But my question would be, why should I care?

You failed one important thing on your graphs. They don't tell us anything, at all. I too can draw a squiggly line on a piece of paper between two lines with tick marks and then gesture at it wildly, but without labels and units its worthless

So, a guy showed up with a simple simulator, modeling a very simple situation. You've got a guy using one weapon, it has Windfury, and he's autoattacking. He noticed that within particular ranges, weapon speed seemed to have no effect on Windfury damage. This is contrary to to the generally-help principle of "slower is better", with regards to Windfury.

So: the guy's building a simulator, and he's starting from the bottom up. What I would expect from these forums is critical analysis of his methodology, comparisons of equations between different simulators, and simple mathematical modeling to see whether his conclusions hold any water. Now, this is just math, and it's not hard math. I expected a few back-and-forth posts, ending in "oh! this is what I did wrong" from him or "wow, these results seem to actually be valid" from the community. Like I said, it's just math. Everything about this very simple model can be fleshed out and explained, and in the end, everybody learns a little bit and maybe you get a new tool for the community.

Instead, there's a chorus of "close this thread" and "you are wrong, period". I see this board operating at a higher level than that. When a colleague shows up with something that is (apparently) wrong, but he's obviously put some work into deriving his conclusions, you don't just show him the door - you talk it out until everyone's on the same page. Kneejerk reactions of "this doesn't jive with what I've already read so get the fuck off my forums" is how we, the community, miss stuff like how good weapon skill was/is. At the very least, if you disagree, try to be civil about it and try to explain your point of view. If there's math involved, don't say "I don't know how to do the math - but you're wrong," wait for someone who can pump out some equations to stop by and settle things.

Back to this guy's initial simulation. Very simple, one weapon with Windfury. He came to the conclusion that weapon speed doesn't matter. It turned out that his simulation was missing a critical part of the Windfury equation, that is, the cooldown. He added that in, and his results were changed somewhat - now weapon speed only didn't matter on certain intervals. It turns out that he's right - with a single weapon, in this simple model, Windfury dps is always a linear function of weapon dps, and that function changes depending on the interval you're dealing with.

S>3.0 implies Windfury DPS =.200 * weapon DPS
1.5<S<3.0 implies Windfury DPS = .167 * weapon DPS
1.0<S<1.5 implies Windfury DPS = .143 * weapon DPS

The math is simple, and I will be happy to provide it - if anyone still cares. Bottom line is, guy shows up with a simple simulation, makes a small error, corrects it, achieves a correct result . . . and get beat up. What's this forum coming to?

The initial post makes no comment in regards to how the two weapons are used and it must be deduced that he is talking about single-wielding which no one does, he uses weapon speed ranges that don't exist in real itemization, and seeks to reinvent the wheel that has in fact already been reinvented (or at least has been invented in two places about the same time).

The conclusion of "It seems to me that 2 weapons (with same DPS) do not affect the total damage after Y second (not using Storm Strike)." is horribly ambiguous (two weapons with the same DPS dual wielded, or two weapons in comparison to each other?) and not at all correct when the model is expanded to include Flurry, Dual Wield, miss/dodge/parry, or any of the other factors that influence real gameplay. It is, on its own, a terribly misleading post and its suggestions are highly contrary to the final result that has been presented by two different simulators and a great deal of parsing. The post should be the first step in creating a more complex simulator but is presented as a stand-alone conclusion that flies in the face of months of work by a large number of posters on this forum.

Yes, what he says is true for single wield. Do you know any enhancement shaman that regularly single wield to do dps in a raid? I don't, and I don't imagine I will in TBC. Testing has born out that dual wield is the superior dps method for enhancement shaman. The generally held principle that slower is better is true for dual wield with flurry, where two weapons (one of which hits for half damage) with variable speeds share a single WF cooldown. People came to this thread seeing enhancement shaman and thinking dual wield because modeling a single wield setup is fruitless. No one does it. There seems to be no point to this analysis, so I ask the OP again, what exactly are we to take from your post? If you are trying to show that faster is just as good for dual wield, then you are wrong because testing has shown this to be false. In other words, if you aren't talking about dual wield pve, then you must be talking 2H pvp, and in that realm, why are we having a dicussion about sustained damage?

I couldn't formulate better my tought. Also are nice the Reputation I received:

- you're a fucking idiot
- You are a silly man.
- Nice try but no cigar bucko
- Stupid at your level is an art form. I approve of art.

this for give you an idea of people lurking around.
The one with the greatest hint was Lactose; I have to admit he has a point but still I believe for big numbers it doesn't change the results (may be I'm wrong too); I can simulate his model (throwing the dice only when WF is not in CD ) but I believe nothing will change, think about it.

A profound lack of understanding in probability - oh dear. My degree is in Mathematics and my profession of choice is built upon an understanding of probability moreso than any other that I can name. But I digress. Here's some math.

Say you have a weapon with speed 1.5<S<3 and dps D. Let's model Windfury procs.

T=0 : Windfury proc!
T=S : No proc possible, since 1.5<S<3, we're within the cooldown.
T=nS: For all n until the next Windfury proc, P(proc) = 0.2.

How many swings will we have to wait, on average, until we see a proc? Writing it out, we get:

(0.2)(1)+(0.2)(2)(0.8)+(0.2)(3)(0.8)^2+(0.2)(4)(0.8)^3 . . .

Which is equivalent to

(0.2)(1 + 2*0.8 + 3*0.8^2 + 4*0.8^3 . . .)

Luckily, we have a simple equation to sum the expansion: 1/(1-r)^2. In this case, r=0.8.

Plugging in the numbers we get (0.2)(1-0.8)^2 = 5.

Thus, we expect to have to wait five swings until the next proc. The swing at T=S is ineligible and not considered. That mans that we expect to see a new proc on the sixth swing after the initial proc - at time T=6S. Please note that I'm not saying this will happen every time. Naturally, it won't - not even close. This is an average of all possible outcomes. Over a very long timeline, there will be an average gap of 6 swings or 6S seconds between Windfury procs.

Back to the math. With a proc, on average, once every 6S seconds, we're going to see (60/6S) = 10/S procs per minute. At each of these procs, the weapon swings twice for SD damage - for an average of (SD)(10/S) = 10D damage per minute, or .167D dps.

This number is completely independent of weapon speed, as long as the speed is within the range of 1.5<S<3. In this simple scenario, Windfury dps is a linear function of weapon dps. Does this clear everything up?

I thought the testing showed that even if you had two weapons with the same speed, they could easily get out of synch from a variety of causes. I believe parrying an incoming attack and server lag were the two most common. No, the two main reasons for dual wielding are:

A) Much easier to maintain 10% Attack Power buff. 2H requires roughly double the crit rate for the same uptime.

B) Better returns from attack power. Suppose your attack power tooltip says it increases weapon DPS by 100. Your can either increase 2H DPS by 100 or you can get 100 DPS from your mainhand and 50 DPS from your offhand. Your miss rate is higher, but the returns are still better. Plus hit rating is cheap on the itemization budget.

He's trying to get a grasp on the mechanics of the class. Modeling dual wield with different speeds with Windfury and Flurry is extremely difficult. Trying to build a pure mathematical model has me scratching my head, and there's apparently a great deal of variance in existing simulators. His input here down the line would be nice, and this simple problem seems like a great place for him to start. Correct response to him should have been: "That's a very interesting conclusion, apparently correct, but not very relevant, because shamans don't use two-handers in raids. Try your hand at modeling dual wield". End of thread. Actual response was "You're wrong - go away" - completely inappropriate and wrong itself, provoking all of this debate. In any case, the math has been provided, everyone should now be clear on what his results were, and I don't think there's anything else to talk about here.

I agree that those are two of the main reasons for dual wield. Completely. I wasn't saying anything contrary to that in what you quoted. I was saying that when you're dual wielding and benefiting from the things listed in your post, you want to be using slower weapons so that you can get more out of your single windfury cooldown.

Agreed.

Sir -
The reason I enjoy reading these forums is that when someone brings up an idea in conflict with the status quo beliefs they are asked to back up their proposed changes, be it with data or well-thought out, (relatively) assumption-free math. They then test it out, do better math, and report their findings.

The reason I read this thread this long was because I was waiting for you to do this, or do more than repeat your original post in different words. Instead all I have seen is the same example rehashed ten times over. I get it by now. Your model is correct based on your assumptions. All of the negative responses are telling you that the problem is your assumptions. Yet you keep repeating the same statements as your first post.

You claim you're challenging the status quo and that everyone is saying "no" before looking at your data. However, when the criticism with your 'theory' is that the assumptions are flawed for the purposes of finding intuitions for your class, you cannot run back to your data.

I read this far into this thread hoping you would amend your assumptions and model and move forward. Instead, there has been no change to your model to account for the realities of play.

In short, if you wish to improve your understanding, work with the people who are telling you why your model is wrong. You may find, as many have pointed out, that by reading the whole enhancement shaman thread your model will either improve or be unnecessary. I would put safe money on the latter.

Certainly.

As I mentioned, it's a fairly rudimentary piecewise function. However, the OP made no account of that originally, and thus invoked the wrath of the forum posters.

I wouldn't consider the OP to be mentally deficient, on the contrary his methodology seems very solid if you accept his assumptions. My critique would be that he failed to clearly state his project, and when objections were raised about what appeared to be overlooked in his assumptions he either misread them or ignored them.

Standard communication failure.

Also, to nitpick. The number is not independent of weapon speed - the number is fixed for a weapon speed inside that range. Not quite the same thing, and if we're going to be exact let's be exact.

Also, while this analysis is correct for weapons who's speed does not vary, as soon as you introduce flurry the analysis must change for some parts of that range. As soon as you introduce instant attacks the analysis changes. As soon as you introduce a second weapon and thus a second swing timer, the analysis must change. All of those changes are required before you approach reality, so the value of simulation/theorycraft at this level is minimal in application.

This is the entry level windfury analysis position, and I believe it's valid to say that it's been covered at least once, although at this point I wouldn't know where in the pages of archival windfury data it has been covered, so perhaps a revival was in order. I hope we eventually generate a very thorough write up of the mechanic for the planned ej wiki, if only to consolidate somewhat.

The simulator already provides a way to equip a second weapon:

Shaman::instance().offHand(myWeapon);

but in order to validate the simulator I started with a single weapon and I posted here the results, thanks to you and Lactose (by private messages ) I'm refining the model.
The 3 zones you are talking about are "proved" on the graph:



look at the lower line ( 0% crit one ), on the X-axis you have all the speeds from 1.0 to 8.0.

the simulator has a glitch still that is the contribution of WF on the hit that depends on the weapon speed (I'll correct this tomorrow ).