Warrior "buff" in 2.1: not what we are expecting?

[nguyenmb]

People really need to take theorycraft with a grain of salt.

No one has given a mathematical model for actual tanking in context. Only various ways to calculate EHP, enmity generated, damage taken over time, healing received, damage dealt, and side by side comparisons of cooldowns. Those are just components to tanking, but never the full picture.

Just because all of the individual components lead someone into thinking one way doesn't mean people who experience otherwise are exceptions.

I recommend googling:
"Widely accepted mathematical results that were later shown wrong"

Yes I know many of you will be willing to change your thesis based on counterexamples given. But just because noone has disproved your math yet doesn't mean there isn't a flaw in there. Actually, I think you should already know your flaw, the fact the you haven't accounted for the actual battle, party dynamics, and communication - the hard things to model yet many of you discount people's anecdotes when you haven't been able to do it theoretically.

How do you quantify the opportunity to inner beast after rockbuster to save yourself before mountain buster?
How do you quantify situations where using thrill of battle 3 times is more useful than 1 hallowed ground?
How do your correctly model Wrath uptime for battles that you tank swap like turn 2? Does a healer really need to heal you if you're going to swap in 3 seconds?
How do you correctly quantify the increased LB generation with War+Pal versus 2xPal?
I'm sure there are many more examples.


On a side note, if you have some computer science in you, i can better elaborate my points:

Have you ever learned about Splay Trees and Balanced binary trees? Individually splay tree operations can and easily be shown to be worse than balanced binary trees mathematically. But not until you mathematically model the whole picture do you see that the amortized running time is equal.

Computer Science is full of examples where people who use their gut feelings and intuition outclass strong theoretical results.

Look at JPEG compression (noone ever considered removing data the eye can't see, a theoretician can't quantify that when considering loss of data), look at GFX Lighting and Rendering (fastest algorithms are random approximations that work well in practice), look at skip lists (again practice works well, theory makes you think not), look at TSP (theory has never been able to achieve results of heuristics, yet heuristics can't show why they do well).

[Ruminate]

snip

I agree, someone who argues "5" then proves it with "3+2" is mathematically sound, but not always correct.

I disagree in that it has nothing to do with real application, but rather that the equation was never "3+2" to begin with.

Now if you want to do away with theory altogether and look at real application on whether or not WAR is on par with PLD, then you need only find the answer to these six questions:

How many groups have successfully gone through turns 1 to 5 with a PLD/WAR setup?
How many weeks of gearing did it take?

How many groups have successfully gone through turns 1 to 5 with a PLD/PLD setup?
How many weeks of gearing did it take?

How many groups have successfully gone through turns 1 to 5 with a WAR/WAR setup?
How many weeks of gearing did it take?

[Leiron]

THe majority of BC turn 5 groups that completed it were PLD/PLD.
War/War has been among the rarest combinations and I know of no group who used such a composition and completed BC turn 5

[Eardstapa]

THe majority of BC turn 5 groups that completed it were PLD/PLD.
War/War has been among the rarest combinations and I know of no group who used such a composition and completed BC turn 5

There WAS a WAR/PLD though, so there's that. I guess.

[Taemek]

People really need to take theorycraft with a grain of salt.

No one has given a mathematical model for actual tanking in context. Only various ways to calculate EHP, enmity generated, damage taken over time, healing received, damage dealt, and side by side comparisons of cooldowns. Those are just components to tanking, but never the full picture.

Just because all of the individual components lead someone into thinking one way doesn't mean people who experience otherwise are exceptions.

I recommend googling:
"Widely accepted mathematical results that were later shown wrong"

Yes I know many of you will be willing to change your thesis based on counterexamples given. But just because noone has disproved your math yet doesn't mean there isn't a flaw in there. Actually, I think you should already know your flaw, the fact the you haven't accounted for the actual battle, party dynamics, and communication - the hard things to model yet many of you discount people's anecdotes when you haven't been able to do it theoretically.

How do you quantify the opportunity to inner beast after rockbuster to save yourself before mountain buster?
How do you quantify situations where using thrill of battle 3 times is more useful than 1 hallowed ground?
How do your correctly model Wrath uptime for battles that you tank swap like turn 2? Does a healer really need to heal you if you're going to swap in 3 seconds?
How do you correctly quantify the increased LB generation with War+Pal versus 2xPal?
I'm sure there are many more examples.


On a side note, if you have some computer science in you, i can better elaborate my points:

Have you ever learned about Splay Trees and Balanced binary trees? Individually splay tree operations can and easily be shown to be worse than balanced binary trees mathematically. But not until you mathematically model the whole picture do you see that the amortized running time is equal.

Computer Science is full of examples where people who use their gut feelings and intuition outclass strong theoretical results.

Look at JPEG compression (noone ever considered removing data the eye can't see, a theoretician can't quantify that when considering loss of data), look at GFX Lighting and Rendering (fastest algorithms are random approximations that work well in practice), look at skip lists (again practice works well, theory makes you think not), look at TSP (theory has never been able to achieve results of heuristics, yet heuristics can't show why they do well).

Hence why parsing is the only true value of measurement anyone should even consider. Maths is and always will be nothing more then a guide.

By the way, I agree with everything you are saying.

[Adamar]

Hence why parsing is the only true value of measurement anyone should even consider. Maths is and always will be nothing more then a guide.

This would hold some value if parsing damage in ARR wasn't completely and irredeemably broken.

[Taemek]

This would hold some value if parsing damage in ARR wasn't completely and irredeemably broken.

Agree 100%, which is why they need to fix that parser to be 100% correct.

Not sure if that is even possible until they start allowing people to make their own add-ons through the add on feature and even then, they may not allow us full access to the API anyway.