PSA: Legion Prepatch Guides - Azortharion

[The Bastard of Bolton]

So here's a...bizzare question. I have a titanforged/socket tyrant shoulderpiece, and a titanforged/socket tyrant legplate (both mythic). None of the tier in analogous gear slots is any sort of warforge/etc. Now the thing is, breaking my 4 set to equip the pieces with higher agi/mastery ends up doing more damage for me....this seems odd. Has anyone else tried completely forsaking the set bonus(es) with titanforges to maximize agi/mastery?

[Azortharion]

Did you sim that or just make it up from hitting the dummy or a boss for 5 minutes

[3clipse]

So where are your logs, Skada, Recount info that refutes the sim results? Jesus fucking christ.

Well, the patch has been out barely a week, so...still aggregating?

You're missing my point, though. I've never said "there's no way the sims are right." What I have always said, or at least endeavored to say, is that there's no empirical reason to believe the sims are right or wrong right now. We need data to determine if the model is good or if the model needs further refinement. So far, no one actually has data, and right now we have no way of judging if the sims are sound or if there's a subtle error in them giving incorrect results.

If you do have a way to determine that empirically without data, I'm really interested in seeing what that is. Like, sincerely interested.

[Azortharion]

What you have attempted to claim is that Beast Mastery should use Barrage over AMoC over Volley on a single target, you based this on dummy testing. Show us the logs that helped you reach this conclusion, redo the test if you must and prove you're not a total idiot, scientist guy.

[Unusual]

Well, the patch has been out barely a week, so...still aggregating?

You're missing my point, though. I've never said "there's no way the sims are right." What I have always said, or at least endeavored to say, is that there's no empirical reason to believe the sims are right or wrong right now. We need data to determine if the model is good or if the model needs further refinement. So far, no one actually has data, and right now we have no way of judging if the sims are sound or if there's a subtle error in them giving incorrect results.

If you do have a way to determine that empirically without data, I'm really interested in seeing what that is. Like, sincerely interested.

The sims are the best info that can be used to determine the current "best way" to do things. That is the data we have, it will be the data we use. If the data is slightly off then how does that matter? It would STILL be a better way of doing things than "X feels better than Y" which is what other people consistently just throw out there.

How exactly are experiments not real world simulations? Experiments may be more accurate or less accurate but if you don't HAVE them or cannot preform them then you just don't have the data. Therefore people use simulations. Your saying the simulations might not be right..............ok they may not be perfect, but it is the data we currently have and the best to base things off of right now other than generic bullshit like "this feels better than this"

"only simulations are accurate, testing damage/DPS in game doesn't tell you anything," --
I don't think that anyone thinks this, but I think that if you can test in game then you could produce the same results with a sim AND have numbers for people to go over. Ya know, the evidence.

[3clipse]

What you have attempted to claim is that Beast Mastery should use Barrage over AMoC over Volley on a single target, you based this on dummy testing. Show us the logs that helped you reach this conclusion, redo the test if you must and prove you're not a total idiot, scientist guy.

What's in it for me to spend a bunch of time and effort doing that? Your posts make it very, very clear that you don't think target dummy or Boss log data are remotely valid. I suspect no matter what I come back with you won't accept the data, or you'll ask for more, or any of the other thousand things that people on the internet do when they can't actually back up their side of an argument with solid points.

I don't see why insulting me (and also like everyone who disagrees with your guides) is necessary, either. You even self identify as an asshole, which like...that should cause you to question why you behave in such a way. Are you just an angry person who lashes out at everyone around them? Or are you trying to neg me into doing it to try prove to you I'm smart, so that you can put me down more? Are you horribly insecure and trying to compensate? Or are you just so arrogant that you can't handle having someone question your advice and ironclad faith in simulations? You should probably introspect about this.

I couldn't care less why you're such an asshole, and the conversation was seeming like it was going to be interesting, but you don't seem to have any desire to actually argue about how one could prove or disprove sim data empirically, so I'm bored now. I've said my piece about the utility of simulations and their applications, you either won't or can't engage in meaningful discussion on the topic. Oh well.

[Azortharion]

So let me get this straight scientist guy; you'll join a forum, claim something different from the widely, generally accepted truth, not provide any proof, get yourself into a long argument about scientific method, act like a little bitch when you're challenged on what you say, and still not provide proof, but instead play victim and launch into some self-righteous psychoanalysis post?

I'll accept any methodology for figuring things out if there's a compelling reason to believe that it is actually the case, but you have yet to even try and provide proof.

Pretty typical case of "either it's an idiot, or it's a troll". I won't be responding to you anymore, I think anyone reading this can tell you're horribly incompetent. So long, "scientist" guy.

[Unusual]

What's in it for me to spend a bunch of time and effort doing that? Your posts make it very, very clear that you don't think target dummy or Boss log data are remotely valid. I suspect no matter what I come back with you won't accept the data, or you'll ask for more, or any of the other thousand things that people on the internet do when they can't actually back up their side of an argument with solid points.

I don't see why insulting me (and also like everyone who disagrees with your guides) is necessary, either. You even self identify as an asshole, which like...that should cause you to question why you behave in such a way. Are you just an angry person who lashes out at everyone around them? Or are you trying to neg me into doing it to try prove to you I'm smart, so that you can put me down more? Are you horribly insecure and trying to compensate? Or are you just so arrogant that you can't handle having someone question your advice and ironclad faith in simulations? You should probably introspect about this.

I couldn't care less why you're such an asshole, and the conversation was seeming like it was going to be interesting, but you don't seem to have any desire to actually argue about how one could prove or disprove sim data empirically, so I'm bored now. I've said my piece about the utility of simulations and their applications, you either won't or can't engage in meaningful discussion on the topic. Oh well.

I'm sorry I (and Azor) hurt your feelings. You still have not shown me data. I will argue data, not feelings.
(Well he is prolly not sorry, lololol)

[Deleted]

Azortharion #1 guildhopper.

Cheers for the guide tho

[Thum8er]

Hey Azortharion

First of all thanks for making this brilliant guide. I'll just get straight to the point. I was wondering if you have done some Simulations regarding races. To be more precise is there a race that pulls ahead of other races in the simulations ? I am referring to the MM Spec specifically.

I am not best Simulation-Crafter, thats why the question.

I am aware that we are in the "Pre-Legion" Phase and that at this moment the question might sound silly. I was just wondering.

Thanks for your any answer.

If other people did some sims on this then please feel free to post your opinion/results.


Thanks again, and keep up the good work.

[Deleted]

Well, I'm a scientist IRL, so I feel compelled to answer this post that's just trying to insult me, not actually ask questions, sincerely.

I am a scientist myself, so maybe I can give a comment on that.

Simulations aren't experiments! Simulations are simulations. You put in your constants, variables, and the framework you've programmed to model reality, and you get some results, but they're only as accurate as what you put in. If you're testing, say, allele frequencies in a large population, and you put in a frequency of 0.8 for an allele that's actually 0.2 in nature, your results are gonna suck dick. The simulations we can do accurately are actually very limited compared to what we actually test experimentally, because you have to be sure that every constant and variable you're plugging in accurately represents what you'd expect to see in the real world (for an extreme example, this is why NASA actually sends robots to Mars instead of just simulating Mars based on data from orbiters).

This is correct for "real science and simulations". The results are only as good as the model used in the simulation. Generally you tend to make a lot of assumptions, because we don't know things exactly, and/or simplifications, so that your calculation power is even able to handle the simulation in a reasonable amount of time.

But this is totally different from what SimCraft does. Simcraft is purely based on EXACTLY known game mechanics and numbers. So unless there is a mistake in implementing theses mechanics and numbers in Simcraft, Simcraft gives you very exact results. Simcraft has shown to be a very reliable tool in the past years, there is no reason to question its credibility.
But, of course, you have to apply these Simcraft results to actual encounter environments before reaching a final decision. Simcraft can not simulate whole boss encounters exactly (although there is a lot you can do with it). But it's not supposed to. It helps a lot reaching conclusions regarding priorities and rotations and so on.

The reason we use Simcraft is, because it is an exact model of the game world and can provide a lot more data than target dummy or raid testing ever can, especially if these are not out or only for a very short amount of time.

Of course, if you can come up with real data that contradicts the Simcraft data, you are welcome, but making wild assumption based on data you refuse to present here is not helping your credibility at all.

[Deleted]

I am a scientist myself, so maybe I can give a comment on that.



This is correct for "real science and simulations". The results are only as good as the model used in the simulation. Generally you tend to make a lot of assumptions, because we don't know things exactly, and/or simplifications, so that your calculation power is even able to handle the simulation in a reasonable amount of time.

But this is totally different from what SimCraft does. Simcraft is purely based on EXACTLY known game mechanics and numbers. So unless there is a mistake in implementing theses mechanics and numbers in Simcraft, Simcraft gives you very exact results. Simcraft has shown to be a very reliable tool in the past years, there is no reason to question its credibility.
But, of course, you have to apply these Simcraft results to actual encounter environments before reaching a final decision. Simcraft can not simulate whole boss encounters exactly (although there is a lot you can do with it). But it's not supposed to. It helps a lot reaching conclusions regarding priorities and rotations and so on.

The reason we use Simcraft is, because it is an exact model of the game world and can provide a lot more data than target dummy or raid testing ever can, especially if these are not out or only for a very short amount of time.

Of course, if you can come up with real data that contradicts the Simcraft data, you are welcome, but making wild assumption based on data you refuse to present here is not helping your credibility at all.

Simcraft simulations are only as good as the APL you use. The model you test is the APL, not the raw data implemented, because that data can be checked against the game data without any simulation needed. What simcraft does via the APL is mix and match the raw data of individual spells acording to a certain set of rules set by the user, which is exactly the same as any computational modeling does. If simcraft would be to work only based on spell formulas, it would be a formulation tool, not simulation

[Deleted]

But it gives exact results for the APL you use (which are your simulation parameters you try to optimize), which means you can optimize the APL with respect to the result and be sure that it will match "game reality" (which would be a corresponding experiment). That is not always true for scientific calculations, e.g. looking at quantum mechanical calculations for great molecules, the outcome depends significantly on which model you use for this specific problem.

What I meant was, that scientific simulations often give not exact results (with regards to matching reality aka matching experimental results) not only because of the parameters you set for that simulation are off but because the model itself does not reflect reality very well.

For me the parameter are experimental conditions and the model has to handle these parameters and give you a result. If the model is correct, your result will match reality. If the model is wrong, your result will be wrong.

TLDR: Whichever APL you use, the result Simcraft gives you will match the result you would get, when doing this as a practical experiment in the game world (within margin of error of course). This is not necessarily true for all models simulations use.

[danik9579]

Thanks for the guids

[Lucyinthesky]

There sure is a lot of "scientists" lurking around on the hunter forum haha.

[Shancii]

cheers for the guide

[Felfaadaern Darkterror]

based on reading the survival guide, the Best pet for survival is now a damn spider, for its ranged snare?

That has to be situational. Sounds like you're talking PvP?

[Deleted]

I wouldn't take anything Azortharion is selling as gospel. The guy is so insecure about himself that he has to resort to standard authoritative shout down debate tactics instead of discussing facts on their merit.

I'm not yet prepared to call him out on his bullshit in the Hunter guides, but I am willing to note that his and Effinhunter's sims show wide differences, and real world tests show even further discrepancies. We know for sure when we have more data.

[GT4]

I wouldn't take anything Azortharion is selling as gospel. The guy is so insecure about himself that he has to resort to standard authoritative shout down debate tactics instead of discussing facts on their merit.

I'm not yet prepared to call him out on his bullshit in the Hunter guides, but I am willing to note that his and Effinhunter's sims show wide differences, and real world tests show even further discrepancies. We know for sure when we have more data.

He's taking a Simulationcraft build that hadn't been updated for a few months as gospel. The first thing I noticed when checking Simulationcraft's current BM build was that a) the rotation is far from ideal and b) it's missing all the Beast Cleave damage from pet abilities which subsequently devalues Blink Strikes by a lot. I haven't checked anything else but I wouldn't be surprised if there were plenty errors like that still - errors you can avoid by not blindly trusting a tool.

[Azortharion]

I wouldn't take anything Azortharion is selling as gospel. The guy is so insecure about himself that he has to resort to standard authoritative shout down debate tactics instead of discussing facts on their merit.

I'm not yet prepared to call him out on his bullshit in the Hunter guides, but I am willing to note that his and Effinhunter's sims show wide differences, and real world tests show even further discrepancies. We know for sure when we have more data.

So basically, the guy (me) is a dick so he must be wrong? Are you basing this on anything real, or are you just talking shit to talk shit?

Link your real world tests, combat logging exists. Next, point out the differences in our sims and why it conceivably matters.

You know, discuss facts on their merit, rather than on your personal feelings about me.

Until you do all or either of those things, your post basically just boils down to a bullshit-loaded personal attack.
- - - Updated - - -

He's taking a Simulationcraft build that hadn't been updated for a few months as gospel. The first thing I noticed when checking Simulationcraft's current BM build was that a) the rotation is far from ideal and b) it's missing all the Beast Cleave damage from pet abilities which subsequently devalues Blink Strikes by a lot. I haven't checked anything else but I wouldn't be surprised if there were plenty errors like that still - errors you can avoid by not blindly trusting a tool.

It is updated continuously.

Blink Strikes is not used for any fights, not because it is devalued in sims by the error, but because there are no fights where you would conceivably have a larger benefit maximizing your AoE damage through it, than you would maximizing your single-target and priority target damage through Bestial Fury/buffing Bestial Wrath.

You have yet to point out how the rotation is far from ideal, so you're either trolling, or the idea of pointing this out simply didn't cross your mind, which makes you incompetent.

EDIT:

General sidenote, if you believe sims to be faulty, you're an idiot for not pointing out the specifics of why. The APL's are public through the tool itself (less updated) or trueshotlodge.ca (always updated), and my Pastebin has some of the bases for my own simulations: http://pastebin.com/u/Azortharion

If "real world testing" shows disprecancies, you can't just say that either. Show it. Else it's worth nothing.