Originally Posted by PiousFlea

Re: Tymir

I've spent much time over the last two days thinking about this problem (lol) and it's an interesting thought experiment. I've come up with at least 6 wildly different ways to look at the problem. I suppose it's all moot with 3.0 as SWP will refresh itself, but I am morbidly fascinated at this point.

I don't believe it's fair to consider Mind Blast as a 6second cooldown in this scenario. Mind Blast and SW:P can conflict with each other no more frequently than every 24 seconds (the duration of SWP). Whether you Mind Blast every
5.5 seconds or every 8 seconds, you can cause a delay in Mind Blast no more than once every 24 seconds.

-----
Consider the ways in which you can refresh SWP:
1) SWP falls off, MB is still on cooldown. You cast SWP. Your MB cooldown is unchanged, and you lose no SWP uptime.
2) SWP falls off, MB just came off cooldown. You cast SWP and then MB. Your MB cooldown is delayed 1.5sec.
3) SWP falls off, MB just came off cooldown. You cast MB and then SWP. You have lost 1.5sec of SWP uptime, and your next SWP will come 1.5sec late.

We will ignore situation #1 because it doesn't affect anyone. Let's again consider Alice (SWP first) and Bob (MB first). Let's say that over a certain period of time, both Alice and Bob have encountered 5 SWP/MB conflicts:

Alice: MB's cooldown has been delayed by 7.5 seconds. SWP has not lost any casting time.
Bob: MB has not lost any casting time. SW:P has lost 7.5sec of uptime.

Woohoo, you say - Bob has gained 2,900 damage from that extra Mindblast, while Alice has lost 2.5 ticks of SW:P for 1600 damage. It's good to prioritize MB over SW:P!

-----
Stop. Reality check.

Bob's total casting time is 7.5 seconds longer because he has let SWP fall off for 7.5 seconds! Alice's casting rotation takes less time because she can reapply SWP more frequently. We are effectively comparing a 24 second cast rotation to a 25.5 second cast rotation - apples and oranges! It is misleading to directly compare the two.

The truth is even trickier than this. The fact is, most of the time there will be no conflict between SW:P and MB and both players will refresh SW:P exactly the moment it falls off. (this is a thought experiment, so both of them can have 0msec effective lag. Maybe I should rename them Cameron and Cromartie.) Sometimes Alice will have a conflict and Bob won't - Alice will fall further behind in MB cooldown but Bob will not lose any SWP uptime. Sometimes Bob will conflict and fall behind, while Alice has no conflict and loses nothing.

If you look at a graph of total damage dealt, at every point from 1.5sec to 24sec, Bob will appear to be 1.5 seconds ahead of Alice. However, Alice gets to refresh SWP 1.5 seconds earlier. If you consider a DPS cycle as starting by refreshing SWP, Bob's DPS cycle is 1.5 seconds slower than Alice's. This is true if the conflict happens right away (24sec vs 25.5sec rotation), but also if the conflict takes several full cast rotations. (ie 72sec vs 73.5sec rotation)

-----
The Answer.

There is only one point at which Alice and Bob can be directly compared. This is after Bob has delayed SW:P by 24 full seconds - that's 16 SWP/MB conflicts. At this point, Bob is ready to cast his Nth SW:P. His SW:P timer is 24 seconds behind Alice. This means that Alice is ready to cast her (N+1)th SW:P.

How many additional Mind Blasts has Bob gained over Alice?
The answer is surprising: Exactly one.

Other than MB and SW:P, Bob and Alice cast exactly the same number of other spells. Therefore, if Alice has cast one additional SW:P, Bob will have cast one additional MB. No matter how improbable this seems, it is logically certain.

What happened the 24 seconds shorter cooldown on Mind Blast? It's an illusion. Bob purchased a 24 second shorter Mind Blast cooldown at the expense of spending 24 seconds longer to cast his spells. In the end he only comes out one Mind Blast ahead, and it doesn't matter what the cooldown of MB is .

Don't believe me? Run a sim or better yet, construct a simple Excel spreadsheet. You will find I am 100% correct.

I made a dead-simple spreadsheet where you can only cast two spells (Mind Blast and SWPain) and time is divided into 1.5second quanta. There are two columns, one which prioritizes MB and one which prioritizes SWP. You can plug in any values for Mind Blast cooldown and SWP duration and you will get one of three scenarios:
1) The "MB first" rotation is grossly more efficient than SW:P first. (less conflicts)
2) The "SWP first" rotation is grossly more efficient than MB first.
3) The rotations are roughly equal, and for every extra MB you get, you lose a full-duration SWP.

What's interesting is that scenario #3 is what happens under "realistic" conditions of 24sec SWP and 6sec MB CD. #1 and #2 are unrealistic in any case, because your filler spell rotation is unpredictable, and combined with lag this will tend to equalize the overall number of conflicts.