Arcane Design - Page 12

Kavan · 09/14/10, 3:59 PM

Actually Kalgan has expressed almost exactly what I was saying all the time, except maybe in a way easier to understand for masses. And him saying that it'll be very hard for players to optimize is not really a sign that they don't understand how the mastery will work out. It's a hint to what I was saying at the very announcement of the mastery, that by its very design of creating an indefinite quadratic problem it is extremely hard to prove that something is optimal.

Woeful · 09/15/10, 11:34 AM

Double post sorry

Woeful · 09/15/10, 11:39 AM

My observations from the PTR:

Arcane Missiles is still tied to the Arcane Blast debuff. I hope this is working as intended, but I kinda doubt that is the case.

1. Mana gem is bugged as I could not find it in my bag to test.

2. The mana conservation cycle is sustainable on the PTR under static conditions. At no time was I without tools to hold my mana pool at nearly any level I desired. The "Mana Conservation" phase was peculiar. ABx2 Abarr with Mage Armor, resulted in a slow depletion of the mana pool. (I need to find a way to quantify this) When AM procced I would actually gain mana regardless of when it procced. (I need to quantify this phenomenon) Arcane Missiles seems to be tied to Elemental Mastery, as the individual bolts that resulted in a crit also produced a mana return. Thus, I was able to "tread water", and actually make small increases in my mana pool with a lucky streak of AM proccs over time.

3. Arcane Missiles is bugged. The arcane blast debuff seems to apply to damage, and the mana return from Missiles appears to be tied to either my miss rate, (character sheet calls it a 3.26% miss rate) or elemental mastery (Which is odd since the spell has no mana). Either way I ended with a net mana gain with missile procs, and a net mana loss when using ABarr to break the stacks.

4. Arcane Barrage still feels rather lack luster. Its the damage feels low when compared to arcane blast, and does hold a candle to the bugged Arcane Missile procs, When casting Arcane Barrage in a mana conservation phase, it appeared to be the main mana hog.

Logix · 09/15/10, 1:45 PM

Originally Posted by Woeful

When AM procced I would actually gain mana regardless of when it procced. (I need to quantify this phenomenon) Arcane Missiles seems to be tied to Elemental Mastery, as the individual bolts that resulted in a crit also produced a mana return.

I have noticed this phenomena on the Beta as well. I haven't, yet, been able to identify the source, but somehow, something is causing a large mana return on what seems to be critical hits, and it isn't AM.

At first I thought they had finally fixed the Master of Elements talent to work well with higher Arcane blast stacks, but I can confirm that this is not the case.

I also checked whether this mana return is coming from AM criticals, but that is not the case either (just tested and confirmed it this morning). AM criticals are not returning any mana (as expected). ABr criticals are returning the expected amount, nothing close to the observed mana return.

I intend to do some deeper testing to see what is causing this. One issue is that it isn't a consistent event, but its noticeable when it happens. It could end up just being a product of a chain of procs combined with mage armor returns.

I will report back if I find anything.

edit:
Just tested it with "misses". Misses are not returning mana for any spell.

Logix · 09/18/10, 6:34 AM

Woe take, for if it comes. Then give.
Also known as; A new build is out.

Arcane

Arcane Blast base damage increased by 25%.

Arcane Barrage base damage reduced by 8%.

Arcane Missiles base damage reduced by 14%.

Arcane Power now increases spell cost by 10%, down from 20%.

Focus Magic now increases your chance to critically hit with spells when the target critically hits.

Arcane Flows now reduces the cooldown of your Presence of Mind, Arcane Power and Invisibility spells by 12/25%. (Down from 15/30%)

What do these changes mean?

More overall damage of Arcane is being offloaded onto AB. Not a wholly good idea in my mind. All this does is force a lot more of Arcane's damage into its "burn" phase (100%mana to 40%mana with CDs popped), while taking away damage from the non burn phases (which is what Arcane will be in for most of the time).
This means that Arcanists who are unable to pull off a well timed evocation and fully sustain a good 'burn' phase will just get penalized more.
In short, it increases the penalty (by a lot) of a badly timed Evocation. Making the proper use of Evocation almost critical to Arcane's success (much more so then it was before).

Also, a rather obvious attempt at making ABr more DPS viable then AM. This is a superfluous change, since there is no choice or thought in using ABr over AM (or vice versa). We will just use whichever one is available to maintain the AB stack we need to be at (i.e. if, say, we want to maintain a ABx1 rotation, we will be choosing ABr over AM not because of some inherent DPS potential, but because it will be the only spell available. For moments where both are available, we will pick AM, since we will only be using DPM rotations when trying to conserve mana, hence causing us to pick the best DPM spell at the time, which will always be AM). This is a core issue with these two spells (ABr and AM) right now. That being, that there is no choice involved in really using them. We just use whichever is available at the time since they are both, by themselves, unreliable.

Arcane Power and Presence of mind now stand at 1min30 seconds cooldowns. Invisibility stands at 2min15seconds.
Currently AP is being used in sync with mana gem, to essentially start the "burn phase". Since mana gem is a 2 min cooldown, there will be little "gameplay" effect from this change.
To be fair, I think the Arcane flows change is mostly just to help line up the cool-downs a bit. An issue may arise in Arcane PvP, but not by much I think.

The Focus magic change makes no sense. We will probably have to wait and see what the change actually is, since it already does what these changes claim it to be doing.

On a separate but connected note. I mentioned a little while back that Mage Armor will most probably be the defacto armor for Arcane mages. I realized that this actually makes a lot of sense due to the existence of Focus magic.
In sacrificing Molten Armor, we will essentially be able to regain what we lost through focus magic (3% crit), which, due to its placement in the Arcane tree, is unattainable by other Mage specs.
If nothing else, I feel this adds more support for the idea that Arcane mages are meant to be using Mage Armor as their primary raiding armor at all times. This is a good thing, since the mana returns from Mage Armor, by all my calculations, are far to important, and allow for concepts such as "maintain your mana at X% while waiting out cooldowns". Which will be the mindset for Arcane post-burn phase in Cataclysm.

Tyrian · 09/18/10, 4:50 PM

Given that Critical Mass is getting a (compelling) selfish bonus, doesn't it seem a little inconsistent that Arcane Tactics currently doesn't provide one? Both talents offer raid buff utility: 5% crit vs 3% damage. If the difference is intended, how are the designers deciding which Raid Buff talents are worthy of selfish bonuses, and which aren't?

Logix · 09/19/10, 2:18 AM

So I got a bit bored today waiting for the Beta servers to come up, so I decided to create a little sim to test out some ideas surrounding how viable high DPM rotations will be, especially since it seems we will need them to be able to maintain our mana at reasonable levels while we wait for things like Mana Gem to come off cooldown.

I ran my sim, telling it to run the most efficient rotation possible (ABx1 ABr/AM) from 100%mana to 82%mana (essentially, replicating a post-burn "wait out your cooldowns" phase). To get a reasonable data set, I got it to iterate over this process 200 times.

The results are interesting, though not wholly unexpected

Time Spent (secs)
Min: 19
Max: 275
Average: 92.63
Standard Deviation: 52.95

This is one of the outputs, telling me how long was it before I reached 82% mana. As you can see, the variance is absurdly high. From the looks of it, most of this deviation is due not only to the unreliability of the AM proc, but also since Clearcasting has such a massive effect on the outcome of a particular trial. Naturally, when I turn those off or dial them down in the sim, things start evening out.

As I commented before, this unreliability will end up being an issue, especially without tools to counteract 'bad luck'. At the end of the day, it just seems a bit unintuitive that Arcane relies so heavily on chance.

As many have surmised so far. the maintaining of mana levels to line up to a burn phase is paramount. Since the latest patch notes have shown that an even larger proportion of Arcane's damage will come through AB, this only makes getting off a 'full' burn phase critical to performance success.

I just don't think we have enough tools to be able to handle some of this unreliability. Blizzard mentioned this as the "little bit of RNG" thrown into Arcane (mentioned in this post). But from everything I can see, there is a lot more RNG here then they expect.

edit:
Forgot to mention.
I threw the little sim up on the web in case anyone else wanted to play around with it. It isn't much, but it at least lays out, to some extent how rotations and procs will effect cycle viability and how, if at all, we can use rotations to conserve mana. I only spent a few hours on it so it really doesn't have any fancy cycle optimization functionality (i.e. it wont tell you what the best rotation to use is, but it will pick the best spell to cast given your input requirements). I think I based the damage numbers off a mage with ~4k SP. Most of the calculated damage was up to date from last beta patch.

The casting and rotation based stuff is pretty ok though I think. It was more about just seeing how much or little things like AM procs and Clearcasting procs will affect viable rotations from a "mana management" standpoint. Basically to explore the somewhat "rngy" nature that Arcane may be taking on in Cataclysm.

Here's the link.
http://manamadesimple.elementfx.com/

edit 2:
The sim could be pretty buggy, so let me know if it suddenly deletes your hardrive and runs away with your wife and kids. Or something equally foolish.

Obeast · 09/19/10, 1:03 PM

This is one of the outputs, telling me how long was it before I reached 82% mana. As you can see, the variance is absurdly high. From the looks of it, most of this deviation is due not only to the unreliability of the AM proc, but also since Clearcasting has such a massive effect on the outcome of a particular trial. Naturally, when I turn those off or dial them down in the sim, things start evening out.

As I commented before, this unreliability will end up being an issue, especially without tools to counteract 'bad luck'. At the end of the day, it just seems a bit unintuitive that Arcane relies so heavily on chance.

Two thoughts on this:

I think the high variance you observe may in some sense be an artifact of the rotation you've chosen. There's no way to know what the preferred low MPS rotation will be until they finalize the constants and we can compute optimum cycles, but I'd guess it's going to involve rather more AB casts on average than the one you've chosen. More AB's will obviously up the MPS, but it will also narrow the distribution, since a smaller percentage of your casts will be "free" ones.

This may sound like scant consolation, since you're burning more mana anyway, but at least it will be more predictable. I'd also imagine that in real raids, in real gear, you'll have access to some mana management tools that your simulation isn't including (although maybe I'm doing it an injustice).

My second thought is that you're not considering an aspect of control that players should have, namely the point at which they switch to one or more higher-dps, higher-mps rotations. From the math I've read, it looks like optimal play under the effects of Mana Adept will involve using your low-dps, low-mps rotations first, and the "burn" stuff later (cooldowns may flip this at points, but in general, in long fights, it should be true). This should imply that if you observe an annoying statistical fluctuation that makes your low-mps rotation actually consume a lot of mana (a long time without MB procs and/or few clearcastings) you can compensate by using less of your high-dps, high-mps rotation before you reset with mana cooldowns.

That doesn't sound like a problem to me -- quite the opposite. It sounds like a opportunity for a good player to control bad luck where a bad player might not.

Elimbras · 09/19/10, 2:41 PM

I've been trying to check the current intuition that arcane mages will be spamming AB, then wait with a low-dps cycle until they can regen. Or, in other words, I've been trying to compare a full K-AB cycle, compared to a rotation with mixed (K-1)-AB and (K+1)-AB cycles.

Model:

I'm simplifying a little bit the model, and consider taht the only meaningful part of the cycle is the AB spell. In otherwords, AM or ABarr do not contribute significantly to the dps nor to the mana expenditure.

Call T the total fight length, and M the total mana you can spend (including regen / procs / etc.).
Call X and Y the dps and mps of the K-cycle.
Consider that when using a N+1-AB cycle, the dps and mps are both increased multiplicatively, with multipliers denoted by a and b.
In other words, we will compare a cycle with X dps and Y mps, to a mixed rotation with partly (aX dps, bY mps) cycle and (X/a, Y/b) cycle.

Assume for simplification that T*Y = M, ie. that the K-AB cycle fits perfectly for your mana and fight lenght.
K-AB cycle has total damage equal to T*X, by definition.

For the mixed rotation, the first point is to consider the proportion of each cycles. Let T1 denote the time spent using the low-dps cycle.
We have obviously that T1* Y/b + (T-T1) * b*Y = M, which we can rewrite as
T1 = (T*b*Y - M) / (Y*b - Y/b) = (Y * T * b - Y*T)/ Y * (b-1/b) = T * (b-1)*b/(b*b -1) = T*b/(b+1).

Hence, the total dps is T1*X/a + (T-T1)*X*a = T*x * [b / a*(b+1) + a/(b+1) ]
The mixed rotation damage is higher iff we have
b/a + a > b+1, which is equivalent to b (1-1/a) < a-1, or more simply, a>b.

First "approximated" conclusion:
The mixed rotation is interesting only if the AB-stacks increases the dpm ratio of the spell, which it does not currently.

Limits of the model:
1/ I didn't take any proc into consideration. When using a trinket (or any effect that modify the dpm ratio), there is more interest in using a burn cycle. This is the same for phases where you have a high incentive to burn due to fight mechanism, even if it does not change your dpm ration.

2/ Obviously, AM and ABarr have an impact of the rotation. However, the number of these are close in both rotations (the ratio of these spells (in favor of mixed rotation) is about 1 + (b-1)/(b+1) * 1/K + 1/K^2.

3/ Clear-casting procs are more interesting when using a high-stacks cycle. In practice, this doesn't change in our model: it saves the same fraction of mana in both case (this fraction is the proc rate), and total mana expenditure is by definition identical.

4/ The constant are not exactly the same when going from a K-1 cycle to a K-cycle, or from K to K+1. More on this below:

5/ I've treated only a single case, where one deviates from the "medium" cycle only of one stack. However, it does indicate a trend, and I see no reason for this trend to change further away.

Multiplicative cycles assumptions:
Call X0 and Y0 the dps and mps of the base AB spell. Call a0 and b0 the multiplicative effect of a stack.
What are the multipliers aK and bK for a K-stacks cycle (and their dps XN and mps YN) ?

We have
XN = (X0 + a0 X0 + a0^2 X0 + ... + a0^K X0 )/K= X0 * (a0^K - 1) / K*(a0-1).
and similarly,
YN = (Y0 + b0 Y0 + b0^2 Y0 + ... + b0^K Y0) / K = Y0 * (b0^K - 1) / K*(b0-1).

Hence, aK = (a0^K-1)/K*(a0-1) ~ a(K-1) * a0 * (K-1)/K.
The multiplicative assumption does not hold perfectly, but for "large enough" K, that should be fine.

Fair conclusion:
Currently, I'm not sure that the spam-and-lull rotation will be interesting if there is no outside incentive to go for it. However, this will be possible at low sustained dps loss, and arcane mages will have to learn how to best deal with it given their cooldowns and each specific fight mechanism.

Logix · 09/19/10, 7:13 PM

Originally Posted by Obeast

Two thoughts on this...

I think there are a few issues that you are missing in your analysis, ones that are critical to the issue I am trying to describe. These issues do not require caculating the exact DPS of the cycles, just their mana sustainability.

Also note, that while damage is being tuned, mana costs for spells are constant on the beta, so calculations concerning mana usage will be just as accurate now as they will later. True, this may change, but there is currently no indication that the core Arcane spells, blast and barrage, will have their mana adjusted.

Let me lay out the issues here, and try to show how they are connecting with each other to form a bigger problem. This is a slightly involved problem, so please bear with the length of the post.
Let us begin.

A. The AB change offloads an extremely large portion of Arcane's overall damage into Arcane Blasts (while simultaneously removing damage from other spells). This, naturally puts a very high premium on ABxmany cycles, which in turn, puts a lot of importance in Arcane's "burn phase" (100% to 40% with gem+trinks popped). Simple inference shows that an ABspam rotation with trinkets popped provides a very large amount of damage in the time its active. Consequently, any negative impact on the 'burn' phase (be it by reducing the amount of time it is active, or having it be active on non-optimal mana levels, i.e not between 100-40% mana) will have a compounding negative impact on Arcane's overall performance.

B. The question then becomes; since the 'burn' phase is so important, what must happen to ensure that it is pulled off smoothly? We see that there are 2 core ideas that are linked to properly executing a 'burn phase':

1. The pre-burn phase must be managed correctly.
The 'pre-burn' phase is the phase just before you pop your mana gem and trinkets. This phase is the one where you temper your MPS output in order to, ideally, ensure your mana is at, or slightly above, the amount of mana returned from mana gem when it comes back off of cooldown. This is so that you can start your burn phase at 100% mana (hence fully benefiting from Mana Adept). More on this phase later.

2. The post-burn phase must be executed correctly.
The 'post-burn' phase is the phase where you evocate. This phase needs to be executed properly so that you can enter the 'pre-burn' phase of the next cycle at 100% mana.
For example, a badly executed post-burn phase is one where you start evocating at <40% mana, leaving you starting your next pre-burn phase at <100% mana. The reason this is key, is since Arcanists do not have a mana positive rotation, they have no tool or method to regain the mana they need to start the burn phase at 100%. So say they enter a pre-burn phase of the next cycle being at 80% mana, this will naturally have a negative impact of the burn phase of that cycle, since you will lose all the potential damage of 'burning from 100%'. This will, due to AB being such a big part of Arcane's damage, have a massive impact on the overall damage output during that fight.

With these two ideas in mind, the question then progresses to seeing what factors influence the pre-burn and post-burn phases, be it positively or negatively.

The first thing that immediately (I hope) jumps out, is that much more so then before, pulling off Evocation correctly is of paramount importance to Arcane. A badly pulled off Evocation has too massive a negative impact on Arcane's overall throughput, since it directly impacts not only the current cycle, but subsequent cycles (and burn phases) as well. Though this point we already know, it is just important to emphasize that if you thought Evocation was critical to Arcanists now, it really will be connected directly to their lifeline in Cataclysm.

The second thing that we must see, is what are our options in the pre-burn phase. This is where my sim comes in.

In essence, it is a simulation to see what realistic rotation options we have when at 100% mana, trying to get no lower then say 82% mana while waiting out the mana gem CD or waiting to line up our next burn phase. In an ideal sense, we will have some way to ensure that for the ~94 seconds minimum (the maximum will vary due to fight mechanics and when we want to line up our burn phases) from when we start our pre-burn phase to when we start the burn phase, we have some way to keep within the required mana levels.

What my sim does is see which possible cycles can pull this off, as well as how reliable they are. Reliability is very key here since (as shown above) if we screw up the pre-burn phase, that has a cascading effect upon the subsequent burn phase (which in turn has a negative effect on our overall output).

To start with, I checked the most mana efficient cycle we can have, ABx1 AM/ABr. In my mind, if this cycle cannot reliably pull off staying between 100% to 82% mana for up to 2 mins, then there is a serious problem, since no other cycles would.

Take, for example, the output of the sim for an ABx2 AM/ABr cycle from 100% to 82% mana, over 200 iterations:

Time Spent (secs)
Min: 19
Max: 149
Average: 52.48
Standard Deviation: 22.38

2 things to note. Unlike the AB1 AM/ABr rotation, this rotation is a little more stable (not by much). There is a much smaller variance in how long it takes to this rotation to get from 100-82% mana, however, the key thing we see is that the average is not ideal, it is only 52 seconds whereas we need, at minimum a cycle that can hold out for at least 94+ seconds.

Now, the AB1 cycle does have a 90-ish average, but it has a massive variance. The issue then becomes that even though we have a rotation that will allow us to properly manage a pre-burn phase, that rotation itself is so very unreliable.

This unreliability, in any other scenario, would not be too bad an issue. But in the context of arcane, and the context of how important a preburn phase is to Arcane's overall throughput, makes it a very very bad situation for Arcane.

Couple this with the fact that there is no mana positive rotation, or any other method by which the Arcanist can react to 'bad luck' we start seeing the issue more clearly, e.g. Say you are in your pre burn phase using an AB1 rotation and you are getting unlucky with clearcasting and/or AM procs. You see you are burning your mana fast and will overshoot the 82% limit well before your mana gem is back up. At this point, there is nothing you can do. You have no tool that will allow you to handle this situation. Neither can you 'downshift' to a more efficient rotation, you are already using the most efficient one you have.

A key aspect is, that this situation did not come about due to an error you made, or something you did not plan for. It came about by pure chance. And there is nothing you can do to rectify it. Whats worse, is how massive an impact this 'bad luck' has on your overall output.

That is the core issue here.
The fact that you will, quite regularly, by looking at the variances, be coming across situations that arise purely out of chance and you will not have anything you can do about it (its not like you can switch to a more mana efficient rotation from an AB1 ABr/AM one) but which will impact your performance greatly. The fact that this unreliability is within the core pre burn phase, means that you will be carrying the negative impact of this pure chance all the way to the end of the fight.

Ideal solutions would be either to;

- Reduce the impact 'chance' has on the overall output of Arcane, or,
- Ensure that we have tools that we can use to counteract this 'chance' (e.g. something that lets us have a mana positive rotation), or,
- Not inject such variance in such a key aspect of Arcane's DPS process (i.e. remove the chance).

One of these needs to happen.
We need to either lessen the impact of bad luck, get tools to deal with bad luck, or remove bad luck from the equation altogether.
As it stands now, Arcane looses too much due to "bad luck".

Obeast · 09/19/10, 8:13 PM

Logix, I think you misunderstood me. Lemme try again:

I'm not saying that specific DPS values have anything to do with the point you're making, which is that the lowest MPS rotation available to arcane has a huge variance in its mana consumption that the player cannot control. I'm saying that without final spell DPS values (and mana costs, which may also change) there's no way to know what low-MPS rotation will actually see use.

I rather strongly suspect that the rotation you call "ABx1 Abr/AM" will not be used as a significant low MPS rotation. This isn't because there will exist a rotation that uses less mana per sec on average, but rather because the loss in dps in going from my hypothetical rotation -- let's call it "Efficient Rotation X," because it sounds ominous and impressive -- to ABx1 Abr/AM too large to justify the savings in MPS.

If Efficient Rotation X rotation uses more AB casts, there is less variance in mana expenditure. You show this yourself in your second post, which considers 2 ABs instead of one, and I'd expect adding more AB casts would further narrow the distribution. That's what I meant by the extremely high variance being "an artifact of the rotation you chose for your simulation," or however I phrased it.

But this is by far the less important of the two points I attempted to express. No matter what Efficient Rotation X turns out to be (it may turn out to be optimal to use several different ones), arcane will still, by construction, see a larger variance in MPS use than other specs. This is true on live, it's true in beta, and it seems like it'll be true in the expansion, period. You're absolutely right about that, qualitatively. I only meant to suggest that it might not be quite as drastic as your first simulation suggests.

Here's what I'd truly like to dispute:

A key aspect is, that this situation did not come about due to an error you made, or something you did not plan for. It came about by pure chance. And there is nothing you can do to rectify it. Whats worse, is how massive an impact this 'bad luck' has on your overall output.

That is the core issue here.
The fact that you will, quite regularly, by looking at the variances, be coming across situations that arise purely out of chance and you will not have anything you can do about it (its not like you can switch to a more mana efficient rotation from an AB1 ABr/AM one) but which will impact your performance greatly. The fact that this unreliability is within the core pre burn phase, means that you will be carrying the negative impact of this pure chance all the way to the end of the fight.

As I understand it, you're making an error in assuming that arcane's optimal playstyle will be a series of "burn phases" from near 100% mana followed by stalling with efficient rotation until evocation and/or other mana cooldowns come up.

The posts I've read that addressed this mathematically, by Roywyn, Kavan, etc., all state the opposite, irrespective of the details of the cycles in question: under mana adept, all other things being equal, you want to cast your lower DPS / MPS rotation first. This means that, for example, if you plan to use AB spam and Efficient Rotation X in a given time frame, you will use X first and AB spam last. Cooldowns or the an encounter-specific need for very strong burst may distort this conditionally, but most of the time it should be the rule.

So I don't agree with you when you say "there is nothing you can do to rectify it." There is: when Efficient Rotation X turns out to be pretty inefficient over a period of time because the RNG gods bent you over a table and had their untender way with you, you can compensate by delaying your transition(s) to a higher DPS, higher MPS rotation. Good players will be able to ride out the high MPS fluctuation near-optimally, while bad ones will really foul up their mana management. By the same token, an observant player will be better able to take advantage of the times where RNG works in your favor and gives you an unusually low MPS -- s/he will switch earlier to the high DPS rotation(s) and profit.

Logix · 09/19/10, 8:50 PM

Originally Posted by Obeast

As I understand it, you're making an error in assuming that arcane's optimal playstyle will be a series of "burn phases" from near 100% mana followed by stalling with efficient rotation until evocation and/or other mana cooldowns come up.

Then this is where we differ.

Take note, while still an assumption, I do not think it is an erroneous one, since, I based the assumption off of exactly what Blizzard themselves have stated their intent is for 'how the spec should play."

I point you to this post by Kalgan.

Quoting Kalgan...
I think that's more accurate. Generally speaking, the approach is something along the lines of...

Evocation and mana gem are available at the start so you burn some mana quickly, use gem and dps cooldowns, burn more down to 40%, evoc back to 100%, play conservatively until mana cooldowns are available, burn more, then come back up and play conservatively, then nearing the end of the fight you try to time your burn down once again to match the end of the fight.

Furthermore, I think your understanding of Roywyn's work is slightly inaccurate.

Roywyn's proposal of a two cycle theorem does not negate the optimal play I am proposing. In fact, according to his own post (found in this thread, post #82), the possibility that I am 'assuming' to be the case works, and as mentioned later in the same thread, could turn out to be the optimal (numbers willing, and if Blizzard has a say, the numbers will be very willing).

Taking the fact that we have blizzard's intent to have arcane play like I describe, and the fact that our current mathematical models actually allow for this 'style' to be the optimal, I feel you may be misunderstanding what Roywyn's analysis actually was, and what it implies. At the end, I am sure blizzard can tweak the numbers so that this is exactly how arcane will end up playing (since, it is their idea of how it should play). I am actually certain, that this is precisely the reason they are offloading so much damage onto AB, to widen the gap and increase the potential damage gained from a "burn phase".

(As a side point, all that Roywyn's two cycle theorem states, in simple words, is that if you will be casting cycle A and cycle B over period C in time, you should use whichever one is more efficient first. This works perfectly well in the overall picture of optimal play I have described. For example, in the "preburn" phase, you can use an AB1 cycle followed by an AB2 cycle as you approach closer to the mana gem cooldown, assuming you have enough mana to be at 82% by the end of time period C).

In short, I think you are misunderstanding what the current math is actually saying about Arcane. From which you are discounting my 'assumption', and through extension, all issues that arise with the existence of that type of optimal play.

Now true, all this could change, after all, it is still beta. But as far as I can tell, this is how things are playing out right now, and hence, the issues I bring up are still valid.

On a separate side note. Everything that I am bringing to the table right now does not include my personal experience with Arcane on the beta (I like to keep things objective and keep the use of empirical evidence to a minimum). But everything that I am mentioning actually correlates almost 100% to how arcane is playing in the beta right now.
As an experiment, I urge you to take your arcane mage in the beta to the dummies and test out this 'assumed' optimal play that I have been describing. Test it over a 200-300 second casting period (which usually, according to my tests, will include ~2 rounds of the "burn-postburn-preburn" cycle) and compare the results to any other 'possible optimal' rotation or play that you can think of. (a word of caution, you should run this test multiple times, since there will be times that you will not be able to pull it off correctly since your pre-burn phase drops you outside of gem range due to RNG, which is the issue I am describing).
I think you may be surprised how well this 'assumed' cycle works already, especially with the new beta patch which has increased the gap between Arcane Blasts and other spells.

In short, pretty much all my testing on the beta is aligning with the 'assumed' play that I have described. On the beta forums, there are other mages reporting the same thing.

edit: Clarity

Obeast · 09/19/10, 10:19 PM

How does that post of Roywyn's (#82, which is precisely what I was referencing) not conflict with, e.g., this statement of yours:

Now, my assertion is that this 60% mana point is nothing more then an artificial "oom" point for the mage. I.e. The mage will burn till he reaches it, then switch to a mana neutral rotation and wait out his cooldowns.

Roywyn's mathematical argument says that, all else being equal, you do more damage by using low MPS rotations before high MPS ones. The statement I just quoted -- specifically "burn till he reaches, it then switch to a mana neutral rotation..." -- is the opposite: high MPS ("burn") then low ("mana neutral rotation").

This makes a difference, as I've now argued twice. When you "burn" first, then everything you say is true -- you have to wait for evocation with a low MPS cycle, and you can get screwed by RNG in a way you can't control. But you shouldn't burn first, except the beginning of a fight where evo is off cooldown anyway.

I'm willing to listen to arguments about how I'm misinterpreting the math (although pray be specific -- you accused me of butchering the math quit a few times in your last post without specifying how and where), but could you please precisely explain which of these two statements you disagree with, and why:

1) In a given time frame in which you plan to use two different rotations in a specific proportion, in the absence of cooldowns, arcane mages maximize damage by using a low MPS, low DPS rotation before a high MPS, high DPS rotation. (Roywyn's result).

2) Given that the low MPS rotations are used before the high MPS ones, good mages will be able to compensate for lost mana due bad RNG in their low MPS rotations by adjusting the balance between the two (or more) rotations in the same fixed time frame.

Kalgan's post, incidentally, does not contradict what I'm saying -- in fact, it echoes it. We're talking about is the part where he says "play conservatively until mana cooldowns are available, burn more..." and so on, not the begininning. The beginning, where a mage deliberately burns mana to get everything on cooldown ASAP and to maximize damage under cooldowns, has no bearing on your argument, since there is no place where a mage can be screwed by RNG during a low MPS cycle. It's hardly possible to have your DPS ruined by fluctuations in MPS when you are deliberately trying to get to a specific % as fast as you can so that you can get evocation on cooldown.

Once you've used your CDs, are back near 100% and are waiting for evo/gem/whatever to come back up, then yes, if you somehow burned too much mana before your regen came off cooldown you'd be in trouble. But this is where 1) and 2) apply, and where skill enters the spec.

You now have a specific time frame: time to evocation coming off CD. You have two or more cycles with varying MPS and DPS values. You use the low MPS cycle(s) first, and you can compensate for bad RNG by adjusting the proportions of the cycles. Since you haven't started casting the high MPS ones yet, this is not a problem -- you just end up using less of those cycles. And this matches what Kalgan says: you use the "conservative" rotation first, and ramp up your MPS / DPS when you get close to getting evocation off CD (what he calls the "burn").

This will take skill, knowledge, and planning -- all good things. Where is the place where a good mage can get screwed by RNG with no recourse?

On an almost irrelevant side note, Roywyn's result ignores dps cooldowns, which will almost certainly make high MPS / high DPS cycles optimal near 100% at various point in the fight as they come off cooldown. But as soon as the buffs expire, you're back to the same situation, and you should switch immediately to low MPS cycles, and there won't be any problem.

Logix · 09/20/10, 1:05 AM

Originally Posted by Obeast

Kalgan's post, incidentally, does not contradict what I'm saying -- in fact, it echoes it.

It is interesting that you say this, since I am essentially echoing what Kalgan is saying.

Firstly though, it is important to note that the quote of mine that you posted is not only being taken way out of context (that quote is from an entirely separate discussion, one where not only were we were discussing hypothetical scenarios on "how to make Arcane a bit more fun" but also trying to figure out "what is Blizzard's idea of how Arcane will play", since, at that time, there was little to no communication from Blizzard on Arcane. The running theory at the time was that we would be balanced around a specific mana point due to Mana Adept, which was later dis-proven by Blizzard themselves). But it is also very important to note, that that post of mine was made before Blizzard came in and told us exactly how they think the spec should be played, i.e. much before Kalgan's post that I have linked in my previous post.

So yes, I am sure you can find other posts of mine, from further back, that are saying something different, but, it would not be prudent to point to those in an endeavor to find me contradicting myself, since they were made in different times, well before any of the current changes to the spec. That does not mean that they weren't relevant or accurate at the time I made them, but just that similar to how the beta itself has changed, ideas and theories change as well, given the light of new information.

So no, I think it is best to focus on what I am saying now concerning how the spec will play, which is pretty much identical to what Kalgan is saying since it is he who essentially inscribed the idea in the first place.

Understanding this, you will understand that I am not trying to change this "vision of how Arcane plays". In fact, I am totally resigned to the fact that it will play exactly like Kalgan is saying, i.e. we will burn a little mana then use gems/AP/trinkets and burn from 100%-40%, then evocate, then try to maintain our mana in wait for our cooldowns and the next burn phase, and rinse and repeat (which is pretty much exactly what I have been saying for my past few posts, many many times over).

What I am trying to do, is explore this "vision of how Arcane will play" and find weaknesses and issues with it. One of the issues I found, as you concede to, is the unreliability of pre burn rotations.

Now, I understand that your solution to this issue is "If you get unlucky in your preburn phase, you can just burn less to not spend too much mana since you got unlucky in conserving it". What I am trying to show, is that this solution does not actually work given the most recent changes to Arcane.
These being, that the greater chunk of Arcane's damage output will now come through AB, naturally meaning, that if we spend less time in the "burn" phase (for whatever reason) that it will have a significant enough impact on our DPS output to warrant issue. Essentially what I am saying is that the DPS loss from not maintaining a full burn phase is too great, simply due to how skewed Arcane Blast's damage is. This makes firing off a full burn phase very important.

What I did then, is take the next step, and try to identify what the possible issues could be that would be preventing us from getting off a "full burn phase". My post on the previous page was this investigation.

To make it easy to conceptualize, think of the "burn phase" as the real meat of Arcane's damage. 'Skill' will come in knowing when and how to use it given encounter specific mechanics. This is good, since this is in our control. However, what we see is that this "burn phase" is too critical to our overall damage output, so we, as skilled Arcanists, must labor to reduce anything that would prevent us getting proper burn phases off (which is also good). Unfortunately, the bad is what my previous post and sims suggest, is that there are things that exist that are out of our control but which can still negatively impact our burn phase. The idea being that we can line up the phase to encounter specific mechanics perfectly. We can time our cooldowns and get our positioning just right. That we can play as skillfully as is in our control, but we can still fail because RNG was against us. This is what the worry is, coupled with the fact that we have no tools to counteract this bad RNG. Skill is about knowing which tool to use, when and for what purpose, as well as using the tool to its full potential.

The solution then is not "just burn less", because the entire point for Arcane in Cata is to "ensure you burn every-time it is possible to do so and for the full duration since that is really where our damage is coming from".

Roywyn · 09/20/10, 8:21 AM

Originally Posted by Obeast

On an almost irrelevant side note, Roywyn's result ignores dps cooldowns, which will almost certainly make high MPS / high DPS cycles optimal near 100% at various point in the fight as they come off cooldown. But as soon as the buffs expire, you're back to the same situation, and you should switch immediately to low MPS cycles, and there won't be any problem.

I reckon that this is just speculation on your part? The level 85 premade gear has 12 (8 base + 4 gear) mastery points, which makes Mana Adept a +18% damage effect at 100% mana. (The post has been written with an assumed +30% bonus, so the exact numbers are invalid. What matters are the qualitative effects that occur just the same.)

Assuming you burn your mana down to 40% and Evocate back to full, you'll lose damage to two sources. First is a 9% average damage loss (from 130% to 112% at the end) since your mana decreases while casting, then there is the extra time lost to Evocate your mana back up. Also, remember that using a full Evocation is far less efficient now than using just two ticks of it, since the first tick occurs immediately. 2 ticks in 2 seconds is 50% better than 4 ticks in 6 seconds, so there may be an incentive to get to 70% mana after Mana Gem, but I doubt you'd want to reach 40% mana.

Interestingly, when your gear improves, two things will happen. First, Mana Adept improves, so the incentive to burn mana is even smaller. That means that the burn cycles have to get comparatively better, to keep up with the increased damage loss from Mana Adept. Also, your intellect increases, which also increases your mana regeneration linearly (up to 4.1% MP5 with glyphed Mage Armour and Replenishment). Do those effects even out? I don't know, I don't even have the slightest idea that would warrant a guess. And unless they have a good simulation/sheet/tool, no one can say how Arcane is going to function and what a good casting strategy is.

Most casters can easily whip up a sheet that calculates their DPCT and the immediate strategy "1) Use the best spell. 2) Hope that you don't run OOM." will likely put you within 5% of your optimal performance. Thanks to proper tooltips, you can do that by just eyeballing numbers in the client even. Arcane simply cannot do that, for the better or worse. From those estimates, playing Arcane as close to 100% as possible with Mana Gems and 2-tick-Evocation seems like a good start. For anything else, I'd consult 3 tools and hope that 2 of them give a similar answer.

Originally Posted by Logix

The solution then is not "just burn less", because the entire point for Arcane in Cata is to "ensure you burn every-time it is possible to do so and for the full duration since that is really where our damage is coming from".

We'll see how this prediction holds up against the final results. I'm pretty sceptical, I must admit.

[Edit]: Premades have 8+4 mastery points for a total of only +18%. And Vontre also has serious doubts whether burn phases can be realised with the current Arcane.

Also, Lhivera delivered a new batch of spell power coefficients:

Arcane Barrage: 71.43% -> 65.43%
Arcane Blast: 71.43% -> 90.29%
Arcane Missiles: 69.65% -> 60.00%

Fireball: 96.51% -> 100.00%
Living Bomb DOT: 80.00% -> 83.15%
Scorch: 49.29% -> 53.91%