Ask Mr Robot "Find Upgrades" is wrong?

[Zorenith 0]

Hello, I am new to this forum but I had a question regarding AskMrRobot's Find Upgrades feature. When I search for upgrades in SoO LFR it displays BBoY trinket as my best upgrade (+1056), however I already have the LFR version of PBI and the flex version of KTT, both fully upgraded. Everything I read agrees that PBI > KTT > BBoY for destro locks. Is AskMrRobot just wrong or am I missing somthing?

 

Here is a link to my armory page:

http://us.battle.net/wow/en/character/area-52/C%C3%A0lyps%C3%B2/simple

 

Thanks! 

[Lockybalboa 618]

AMR is commonly wrong about a lot of things. This is one topic, of many, that it is very wrong on....kind of.

 

Normally it should be assumed that KTT>BBoY for destro, unless you're very good at using BBoY. If you have some advanced skill and don't mind using it, you can eek out more damage with BBoY then with KTT, in theory. It's not easy and takes some work, but it can make for some smexy damage.

 

Under ~97 percentile skill level, KTT>BBoY

[Liquidsteel 279]

It's not necessarily wrong, it just isn't searching for what you want.

The default AMR profile favours haste over crit, and its gear rating system is based off those weights.

Personally, I would never use AMR to find upgrades. Look through the dungeon journal and see what I need. Simcraft is a good tool for comparing items.

[Terris 0]

AMR is wrong, multistrike > haste for destro.

[Zagam 1,982]

AMR is wrong, multistrike > haste for destro.

Not always true.  The variance is very small.  Most players will see KTT > BBoY, but some will be able to get BBoY to do more than KTT.  It's not a fair analysis to compare Multistrike to Haste.

 

KTT = Multistrike, which could range from 2% to 8% of your total damage in a fight - the RNG of it is very skittish.  There is also a flat Intellect proc for 10 seconds.

 

BBoY = Haste which you can reforge into Mastery which can result in a 4-6% damage gain just on the Haste and Mastery alone.  Then you have a 10 stacking buff that is more powerful than KTT from 6 stacks until 10 stacks where it DOUBLES KTT's proc.  This is why if you can utilize BBoY to a VERY high level, it can and will outperform KTT.

[Liquidsteel 279]

The past few months have really cemented the ideal that we can't really deal in absolutes anymore when it comes to gearing etc. 

 

I love it so much.

 

I was honestly taken aback by how well I did with no LMG, crit reforge (I was uncomfortable with it and often canceled chaos bolts by trying to move too early as a muscle reflex) and BBoY, especially on a fight like Thok which can be tricky to time that 10 stack chaos bolt.

 

I've gone back to Mastery > Haste for the time being, but I have a new found love for Crit (ember gen felt insane sometimes when you're used to calculated gains from haste).

[SparkSovereign 61]

"Wrong" is one of those words that, as a computer scientist, always bugs me.  It gets so misused.

 

"On two occasions I have been asked [by members of Parliament], "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?"...I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question."

 

 

AMR is not "wrong" in that it is telling you what is, definitively, correct... for what it is being told to do.  There is no error in their code that is causing this result you perceive as "wrong".  It is working exactly as intended.  The "problem" and source of your frustration is that you are asking the wrong question.  It's not your fault, the defaults are not well communicated (no small part of Locky's perpetual ire for it), but that's what's happening.

 

What you wanted was to know what gear will maximize your dps.  This is not what AMR actually does; AMR optimizes the sum of the stat weights on your gear.  What those stat weights are is entirely irrelevant; it will optimize for whatever you tell it.  If you told it to weight hit past cap 30x higher than any other stat, it would give you the absolute most hit rating you can squeeze onto your gear, and it would not be "wrong", even though the result would be useless in practice.

 

"+1056" means that by switching to that item, your summed stat weights (including whatever value it gives to procs and what have you) would go up by 1056.  This is not wrong.  They would.  You can do the addition by hand, if you want to verify it.  AMR is not "wrong" here.  Knowingly or not, you asked it to maximize the sum of stat weights, and that is what it has done.

 

However, trinket procs are somewhat more involved that a pure stat average, as evidenced by the useful responses earlier in the thread.  A good comparison is why Engineering has better pure throughput than other professions; the average stat gain works out exactly equal to the others, but you can stack it with any burst you have for greater effect.  I would NEVER attempt to use any simple math (here, adding raw weights) to figure out which trinkets to use; nothing less than a full simulation will get you the right picture, and it is really hard to write a sim that properly demonstrates the impact of skillfully making use of trinket procs.

 

Basically, AMR correctly answered the question given to it.  It is not "wrong", it's just not the right tool for the job.  It's like giving a computer two pictures of the ocean and asking which is more blue, and then looking at the result and saying "but the other one is prettier".  Hrm, I actually like that analogy, I'll have to use it more often.

 

 

 

Rant aside, I'm not a warlock expert, but since KTT doesn't give multistrike to pets (notice how most of the warlocks in this thread are using Grimoire of Sacrifice) and you're using Grimoire of Supremacy, it is entirely possible that even with "average" skill LFR BBoY is actually better for you than Flex KTT.  Rather than using Ask Mr Robot to try and answer that one, though, I suggest using this fabulous tool called "Ask Mr Zagam" found in this very section's pinned threads.

[Astynax 24]

"+1056" means that by switching to that item, your summed stat weights (including whatever value it gives to procs and what have you) would go up by 1056.  This is not wrong.  They would.  You can do the addition by hand, if you want to verify it.  AMR is not "wrong" here.  Knowingly or not, you asked it to maximize the sum of stat weights, and that is what it has done.

 

I have not bothered to do the math on the scenarios that Locky has presented in previous posts. That said, as I understand them: No, it did not. It physically did the math incorrectly. 

 

I've gone back to Mastery > Haste for the time being, but I have a new found love for Crit (ember gen felt insane sometimes when you're used to calculated gains from haste).

 

 

So much this... I was soo surprised when I switched to crit. My AOE feels like shit... but the embers man... the embers... And I still have 20% haste raid buffed due to not being able to drop enough of it.

[Kazistrasz 216]

"Wrong" is one of those words that, as a computer scientist, always bugs me.  It gets so misused.

 

 

 

AMR is not "wrong" in that it is telling you what is, definitively, correct... for what it is being told to do.  There is no error in their code that is causing this result you perceive as "wrong".  It is working exactly as intended.  The "problem" and source of your frustration is that you are asking the wrong question.  It's not your fault, the defaults are not well communicated (no small part of Locky's perpetual ire for it), but that's what's happening.

 

What you wanted was to know what gear will maximize your dps.  This is not what AMR actually does; AMR optimizes the sum of the stat weights on your gear.  What those stat weights are is entirely irrelevant; it will optimize for whatever you tell it.  If you told it to weight hit past cap 30x higher than any other stat, it would give you the absolute most hit rating you can squeeze onto your gear, and it would not be "wrong", even though the result would be useless in practice.

 

"+1056" means that by switching to that item, your summed stat weights (including whatever value it gives to procs and what have you) would go up by 1056.  This is not wrong.  They would.  You can do the addition by hand, if you want to verify it.  AMR is not "wrong" here.  Knowingly or not, you asked it to maximize the sum of stat weights, and that is what it has done.

 

However, trinket procs are somewhat more involved that a pure stat average, as evidenced by the useful responses earlier in the thread.  A good comparison is why Engineering has better pure throughput than other professions; the average stat gain works out exactly equal to the others, but you can stack it with any burst you have for greater effect.  I would NEVER attempt to use any simple math (here, adding raw weights) to figure out which trinkets to use; nothing less than a full simulation will get you the right picture, and it is really hard to write a sim that properly demonstrates the impact of skillfully making use of trinket procs.

 

Basically, AMR correctly answered the question given to it.  It is not "wrong", it's just not the right tool for the job.  It's like giving a computer two pictures of the ocean and asking which is more blue, and then looking at the result and saying "but the other one is prettier".  Hrm, I actually like that analogy, I'll have to use it more often.

 

 

 

Rant aside, I'm not a warlock expert, but since KTT doesn't give multistrike to pets (notice how most of the warlocks in this thread are using Grimoire of Sacrifice) and you're using Grimoire of Supremacy, it is entirely possible that even with "average" skill LFR BBoY is actually better for you than Flex KTT.  Rather than using Ask Mr Robot to try and answer that one, though, I suggest using this fabulous tool called "Ask Mr Zagam" found in this very section's pinned threads.

 

Zag and Locky have begun using BBoY recently due to frustrations with the ridiculous RNG inconsistency of KTT ranging from 2% to 9% damage on any given fight. They've determined the better you are as a player, the better BBoY is for you. If you're of average skill, though, you may still be better off with KTT.

 

As another computer scientist (well, former, yay major change with a year left to graduate), I agree. The computer will never do something its code doesn't somehow, somewhere tell it to do. I am totally with you on the computer not breaking the rules. I just think their rules are inaccurately written, as I have had it, regardless of different stat weight settings, put me below caps, put be significantly above caps pointlessly, improperly calculate post-amplification rating values (on a character not affected by the bug identified yesterday), and otherwise give me setups that were worse than what I've come up with by hand.