You can criticize Lhivera for many things.

Knowledge of the mage class is not one of them.

Yes and no. I absolutely think Lhiv does a respectable job with the math. Practical use on the other hand no. Don't get me wrong, I think he is a good guy who means well. And that is why he has green text. And I do agree that he deserves it. That said, you also have to take what he gives with the understanding that there is a side of the game that he does not understand and does not take into consideration.

There are many people who could attain better results than most of us due to mastery of mathematics yet not even bother logging into the game. Not saying Lhiv is one of those. But this 'practical' bullsh!t is getting old.

There is more to it then that. Don't get me wrong, I LOVE math and think you can use it for many things, both in game and in the real world. However, just looking at numbers will not tell you all of it. Let me give you an example:

Yes...yes it will. The weakness in 'numbers' is the limitations of your model. If you can account for every variable, your model will accurately predict reality.

Rarely do we have a perfect model, though. However, the debate is about whether our models are accurate or not and the degree of accuracy.

You have a group of mobs being tanked, lets say 6-7, by 1 tank. Puddles form randomly forcing the tank to shift position every little bit. He doesn't move much, but he does shift. Now, math will tell that the highest damage possible is to put NT on every one of them. And yes, theoretically it is. However, for those that have raided much, you know that putting it on everyone is a different story. Tab-targeting is problematic, especially when the mobs shift positions. Click targeting is troublesome because the shift, get in each others way, etc. Because of delays in getting the right ones targeted, it can very easily mean a dps lose to try and dot all of them. That is something that mathing out the 'best' wont really tell you.

It's not problematic if you use the correct addons. TidyPlates is a good example. Theoretical situations may be obtained through optimal use of tools given to us. Don't think for a minute that models do not account for situations like the ones you've just described, either.

The other thing you have to think about anytime you are dealing with math is number of variables and undefinable variables. The game may run itself entirely on math, but there are two problems to mathing things out. First, by design, you don't get all the variables.

The only valid point you have so far. We are not privy to all of the information that the developers are and therefore we do not have the entire picture.

Second, players (particularly other players) cannot be put into a nice equation. You cannot math out all the solutions since you don't have all the variables. Practical experience can fill in what the math leaves out.

I'd love to impart some knowledge about ODE/PDE/NLDE-type courses to you because it does exactly what you just described.

You are arguing that the theoretical model has discrepancies when you compare it to the model including player error. Curve fitting is an excellent way to 'fit players into a nice equation'.

All of this can be explained with mastery of mathematics and statistical analysis. Granted, very few of us have the mastery I speak of (myself included).

Stating that everything may not be understood through the basics, though, is wrong, outright.

Game design isn't a purely mathematical exercise, and becomes debased when it devolves into that. It is an art, and what Blizz is doing here is painting by numbers.

Would love to see your proof for this.

Otherwise, I'm pretty sure game design involves assessing current environments and adjusting numbers in a game engine...which involves pure mathematics. No art is involved.

It's such a nuisance when people start mistakenly classifying mathematics as art.

It's not problematic if you use the correct addons. TidyPlates is a good example. Theoretical situations may be obtained through optimal use of tools given to us. Don't think for a minute that models do not account for situations like the ones you've just described, either.

Funny, I have tidyplates. I see those health bars swing all over the place when the tank takes even a few steps. So, yes it can still be problematic. Sure, addons can get you closer to the theoretical perfect, but just closer, not there.

Yes...yes it will. The weakness in 'numbers' is the limitations of your model. If you can account for every variable, your model will accurately predict reality.

Rarely do we have a perfect model, though. However, the debate is about whether our models are accurate or not and the degree of accuracy.

This is true. And exactly my point. The problem with only doing the math of the abilities is the fact that there is a great deal it does not account for. There is a great deal that you have to experience in order to correctly apply the equations. The point is, by the nature of things, the math of the abilities leaves far to many variables out. Because of that its degree of accuracy suffers.

You are arguing that the theoretical model has discrepancies when you compare it to the model including player error. Curve fitting is an excellent way to 'fit players into a nice equation'.

All of this can be explained with mastery of mathematics and statistical analysis. Granted, very few of us have the mastery I speak of (myself included).

I feel like you don't really get what statistics are. When we start dealing with situations, such as the world/societies/etc, we run into a situation where all the variables cannot be known. In fact, most of the variables cannot be known. You can't use a formula to predict anything when you have an unknown number of unknown/unknowable variables. You just don't have what you need. That is where statistics come in. It looks for patterns in past behavior to guess at future behavior. By its very nature, it is guess work. Fortunately most systems are fairly consistent. You can make reasonable guesses. But, they are still guesses. And by the nature of statistics, there are always outliers, situations that produce different results then the average. So, even with the best statistics out there, you cannot math out how X person will act. Oh, he will fit on the curve somewhere. Maybe the middle, maybe an outlier. An individual is like rolling the dice. If you have 500 people, the average is reasonably predictable. If you roll the dice 500 times, you know your average. But taking 1 person is like taking 1 roll of the dice.

Ironically, practical experience works much more like statistics. You learn previous behavior of a system and learn to anticipate similar behavior in the future. So, saying that a person needs practical experience should not be taken as wrong.

Stating that everything may not be understood through the basics, though, is wrong, outright.

**IF**you could possibly have every single variable. You can't! There is a reason labs do experiments. You can put everything down on paper, think you have all the math worked out and know exactly what is going to happen, run the experiment and something completely different happens.

Sims are like the theoretical system. We do the math and figure how things should work. We try to account for variables we know. We get results. Then we run the experiment. Think of raidlogs as the results. The results differ from the sims. This means there are unaccountable variables. To really understand why they differ, to figure where the extra variables are, we need to observe the whole situation. This is where experience is key. Getting in and doing the raid fills in the wholes as to why numbers differ. It is the equivalent of running the actual experiment.

So, you may be good at theory. You may have a good grasp of what 'ideally should happen', but until you run the actual experiment, you will be lacking. The theoretical has its place and

**is**valuable. But it is far from the end.

03/30/2013 05:58 AMPosted by NeeberFunny, I have tidyplates. I see those health bars swing all over the place when the tank takes even a few steps. So, yes it can still be problematic. Sure, addons can get you closer to the theoretical perfect, but just closer, not there.

If there's some way to prevent this I'd definitely like to know (not too stray too far offtopic here)

03/30/2013 05:58 AMPosted by NeeberFunny, I have tidyplates. I see those health bars swing all over the place when the tank takes even a few steps. So, yes it can still be problematic. Sure, addons can get you closer to the theoretical perfect, but just closer, not there.

Do you understand what a model is? The last sentence would suggest you don't.

03/30/2013 05:58 AMPosted by NeeberThis is true. And exactly my point. The problem with only doing the math of the abilities is the fact that there is a great deal it does not account for. There is a great deal that you have to experience in order to correctly apply the equations. The point is, by the nature of things, the math of the abilities leaves far to many variables out. Because of that its degree of accuracy suffers.

So...I have to experience getting electrically shocked to understand Ohm's law? That analogy doesn't make any sense.

What you are essentially arguing is that you need experience to actually know what you're talking about which is, as demonstrated, a complete croc. No experience in-game is needed to understand the mathematics and physics that the game engine runs on.

Math does not leave any variables out. Math accounts for all of the variables. Simplifications are made to make the equations more manageable and these assumptions usually are quite valid. Go ahead and try to 'solve' one of Einstein's field equations without making brash assumptions. It's neigh impossible.

Mathematics describe the situation perfectly: the limitation is on the human looking at the numbers. It suffers no loss at all by virtue. The minute you make an assumption, the actual equation becomes just an approximation.

Is that approximation good or not?

**That is entirely dependent upon the person looking at the data!**

Furthermore, do you know how many equations have been derived just from experimental results alone? Look at Rydberg and the atomic spectra. He made a great discovery but only could account for it due to his observations via experimentation. Only later did we accurately describe the situation with an equation.

03/30/2013 05:58 AMPosted by NeeberI feel like you don't really get what statistics are.

You are telling an industrial engi that he does not understand what statistics are. Please just stop.

Statistics allows us to pinpoint the correlation between variables which is invaluable for the topic we are currently discussing. It allows you to determine if the correlation between two topics is weak or strong or non-existent. It allows us to quantify parsed results (such as WoL) and form coherent hypothesis based upon variation in the numbers.

So much more but...stat is extremely useful when you end up making too many assumptions in math.

03/30/2013 05:58 AMPosted by NeeberWhen we start dealing with situations, such as the world/societies/etc, we run into a situation where all the variables cannot be known. In fact, most of the variables cannot be known. You can't use a formula to predict anything when you have an unknown number of unknown/unknowable variables. You just don't have what you need. That is where statistics come in. It looks for patterns in past behavior to guess at future behavior. By its very nature, it is guess work.

Be careful there: you are oversimplifying a field that is invaluable to many scientific fields. Without statistics, scientists would be up the creek. Just look at how important probability is in modern physics.

03/30/2013 05:58 AMPosted by NeeberFortunately most systems are fairly consistent. You can make reasonable guesses. But, they are still guesses. And by the nature of statistics, there are always outliers, situations that produce different results then the average. So, even with the best statistics out there, you cannot math out how X person will act.

They are 'guesses' (a rudimentary term) backed by analysis of variation (like ANOVA), correlation between variables, and so on. These guesses allow us to determine whether there is a strong correlation between two variables.

Outliers are very much accounted for in statistics as well. One of the first things you cover in a basic statistics class.

Barring that, you are claiming people are just random and are not predictable. How does this help you? Lhiv seems to at least try to apply in-depth analysis to mage problems where others take a backseat and throw around their own personal opinions not backed by facts. You wouldn't be able to predict behavior any better in this case.

So what is better: someone who makes inferences based upon a field of mathematics specifically designed to account for EVERY variable, EVERY outlier, and every....thing OR a person who just throws their opinion around, citing their in-game experience as proof that they know what they're talking about?

03/30/2013 05:58 AMPosted by NeeberOh, he will fit on the curve somewhere. Maybe the middle, maybe an outlier. An individual is like rolling the dice. If you have 500 people, the average is reasonably predictable. If you roll the dice 500 times, you know your average. But taking 1 person is like taking 1 roll of the dice.

Called expected value. Not hard to actually apply to a single person.

This is the very reason we have...duh duh duh...confidence intervals.

03/30/2013 05:58 AMPosted by NeeberIronically, practical experience works much more like statistics. You learn previous behavior of a system and learn to anticipate similar behavior in the future. So, saying that a person needs practical experience should not be taken as wrong.

What previous behavior do you refer to? All you've described is, "Well, if you do this x amount of times for y duration, you will be able to predict things in the future." Uh..what you're doing is applying statistical concepts: trimming your outliers, getting closer and closer to the true population mean, and determining which variables influence others.

Someone with a mastery of statistic does not need to have 'experience' to do what you described.

03/30/2013 05:58 AMPosted by NeeberIFyou could possibly have every single variable. You can't! There is a reason labs do experiments. You can put everything down on paper, think you have all the math worked out and know exactly what is going to happen, run the experiment and something completely different happens.

I work in a lab almost all day. The simulation and pre-calculations are dependent on the person, yet again. The experiment is to confirm the calculated and simulated results. Now, if you're talking about upper tier science and research, you may have a point.

But this is WoW and WoW's gaming engine. Not as complex as astrophysics yeah?

Sims are like the theoretical system. We do the math and figure how things should work. We try to account for variables we know. We get results. Then we run the experiment. Think of raidlogs as the results. The results differ from the sims. This means there are unaccountable variables. To really understand why they differ, to figure where the extra variables are, we need to observe the whole situation. This is where experience is key. Getting in and doing the raid fills in the wholes as to why numbers differ. It is the equivalent of running the actual experiment.

So, you may be good at theory. You may have a good grasp of what 'ideally should happen', but until you run the actual experiment, you will be lacking. The theoretical has its place andisvaluable. But it is far from the end.

This is a little off but I see where you're coming from.

However, raidlogs is a form of simulation. It most certainly does not account for all of the variables. Furthermore, if you run the 'experiment' (by doing some raids), you will never be sure if your results accurately reflect the sim/calculations because your sample size is too small.

What yields results is large amounts of data that can then be assessed. What methods are you applying during the assessment? Well, that should be obvious.

Case in point: no one actually has to play this game to gain this invaluable experience you speak of if their aptitude for analysis is spot on. Easy to do now days considering how much information we have available to us.

http://us.battle.net/wow/en/forum/topic/8518532477?page=1#8

IMO, Flamestrike/Blizzard could use a huge buff, the damage needs to be shifted back to our nukes, and I'd like to have LB cleave back or be better at target-switching.

I'm competitive with my guild but I've always been a super hardcore min-maxer type and a lot of them only log in for raids and they're gone to the next one. When Fire was near top of the pack nobody every came close to me, but now that we're middle of the pack I generally just tend to be near or at the top, not just 100% of the time anymore. Unless its an AOE fight, then I get crushed =p I currently have a great weapon now but I'm still lacking the 2pc or 4pc t15 that most of my guildies have. Blew all my DKP on the Jinrohk ring and pants. (not really vying for it, I just bid on stuff that's an upgrade)

Do you understand what a model is? The last sentence would suggest you don't.

I really don't think you are getting what I am saying. There are added variables, such as a tank moving the pack as making targeting more difficult and more time consuming that purely mathing the best wont account for. Addons can in fact alleviate some problems. But not all of them. For example, tiddyplate does make targeting easier. But those plates still swing around like crazy during movement. So, targeting still has variables the model does not cover.

What you are essentially arguing is that you need experience to actually know what you're talking about which is, as demonstrated, a complete croc. No experience in-game is needed to understand the mathematics and physics that the game engine runs on.

No, what I am essentially arguing is that if you want a complete picture you do in fact have to have experience. Lets say you want to learn to shoot a bow. Archery is pure math. Force, ballistics, friction, etc, etc. A great deal to learn. Study all of it in books and you should be an expert, right? Do you think you can speak to the difficulty of a shot? Do you think anyone would give you much credence? No on all counts. Sometimes, you have to experience something to get the whole picture. Doing math on paper is a useful tool and it provides useful information. But it cannot provide the entire picture.

Math does not leave any variables out. Math accounts for all of the variables. Simplifications are made to make the equations more manageable and these assumptions usually are quite valid. Go ahead and try to 'solve' one of Einstein's field equations without making brash assumptions. It's neigh impossible.

Mathematics describe the situation perfectly: the limitation is on the human looking at the numbers. It suffers no loss at all by virtue. The minute you make an assumption, the actual equation becomes just an approximation.

That is kind of the point I am making. Knowing every variable is not possible. Some variables get left out. By the nature of doing math just the math with no practical experience is that you will leave out variables that you shouldn't. You make assumptions based on a limited understanding. Therefor you equations by nature will be lacking.

Why do you think scientists perform experiments? They are testing the assumptions they make. Its simple really. If the results do not meet the expectations, a false assumption was made. Without actively testing your assumptions, you cannot know if you made correct ones or not. So, saying someone who just does the number crunching but does not experiment does not have a complete picture is in fact correct.

But this is WoW and WoW's gaming engine. Not as complex as astrophysics yeah?

Maybe not, but I wonder if the people who play it aren't. But that aside, less complex then astrophysics does not mean simple either. ;) And it doesn't mean that you can have every significant variable without experience.

Furthermore, if you run the 'experiment' (by doing some raids), you will never be sure if your results accurately reflect the sim/calculations because your sample size is too small.

I do agree. However, running the raids will help you understand the variables involved better. Just like preforming an experiment in a lab wont guarantee you covered every variable. However, it does provide insight into how things work.

Case in point: no one actually has to play this game to gain this invaluable experience you speak of if their aptitude for analysis is spot on. Easy to do now days considering how much information we have available to us.

You can analyse data and come to some very solid conclusions. Even some very accurate ones. However, you are effectively removing a data source. You increase your likelihood of false assumptions when looking at the data only. When you experience it, you are more likely to apply the correct assumptions.

So, if I have someone who has no practical experience telling you 'the math says' you have to take understanding that wrong assumptions are very possible. The data must be taken in that light.

Again, the math most certainly has value. But if you do not have experience you miss a valuable data set. You increase the likelihood of a false assumption. And thus, you increase the likelihood of a false conclusion.

Remember when we used Inferno Blast to spread LB in PvE?

Man that was awesome.

Man that was awesome.

04/01/2013 11:18 AMPosted by NeeberI really don't think you are getting what I am saying. There are added variables, such as a tank moving the pack as making targeting more difficult and more time consuming that purely mathing the best wont account for. Addons can in fact alleviate some problems. But not all of them. For example, tiddyplate does make targeting easier. But those plates still swing around like crazy during movement. So, targeting still has variables the model does not cover.

What you have described is user-error. It is quite reasonable to assume most mages have mastered tidy plates well enough so that targeting isn't an issue in raiding environments. These variables are accounted for: which is what you seem to not understand.

04/01/2013 11:18 AMPosted by NeeberNo, what I am essentially arguing is that if you want a complete picture you do in fact have to have experience. Lets say you want to learn to shoot a bow. Archery is pure math. Force, ballistics, friction, etc, etc. A great deal to learn. Study all of it in books and you should be an expert, right? Do you think you can speak to the difficulty of a shot? Do you think anyone would give you much credence? No on all counts. Sometimes, you have to experience something to get the whole picture. Doing math on paper is a useful tool and it provides useful information. But it cannot provide the entire picture.

You seem to have lumped a sport that requires extreme precision and dexterity with WoW: not a sport that does not require much besides pressing keys in the right sequence and using your addons effectively.

You absolutely do NOT have to experience WoW to predict the linear WoW mechanics you seem to think are so complex.

There is also the difference between a human-written engine (which makes assumptions of their own to alleviate strenuous mathematical algorithms) and a sport that has to deal DIRECTLY with the laws of physics.

This is exactly what I've been telling you: these assumptions have absolutely been made and these outliers you are citing have been accounted for. Do you think WoW's game engine models the real world to the T? Definitely not! Can you collide with another person in real life and merge through them? No. Why can you do that in WoW? It makes life easier on the programmer.

That is kind of the point I am making. Knowing every variable is not possible. Some variables get left out. By the nature of doing math just the math with no practical experience is that you will leave out variables that you shouldn't. You make assumptions based on a limited understanding. Therefor you equations by nature will be lacking.

Knowing every variable is not NEEDED when you are speaking of mechanics in WoW for the love of Cthulhu. The algorithms governing this game engine are not revolutionary. They're not some impossible to solve problem like Einstein's field equations. We have data (from both the developers and logs) that accurately describes what the WoW engine does, how it does it, and why it does it.

You are entirely incorrect when you say you make assumptions based on limited understanding. It is entirely the other way around. Hear this clearly: you make assumptions because you comprehend that, without these assumptions, the mathematical workload would bog down the analysis by quite a bit. Good assumptions are made when you truly understand what you are talking about and the constraints of your model.

The equation will be lacking but do you know what it is lacking? Are you able to assess whether the assumptions made are valid or not? No probably not. Which is why I tell you: it is best left to people well-versed in the field of statistics and mathematics to decipher what is going on with WoW simulations and analysis of logs. Lhiv seems to have proven useful in this regard.

04/01/2013 11:18 AMPosted by NeeberWhy do you think scientists perform experiments?

Scientists perform experiments to confirm what they suspect to be the proposed solution to a problem. This implies that they have done considerable work in preparation for this experiment in the design and simulation phase.

04/01/2013 11:18 AMPosted by NeeberThey are testing the assumptions they make.

And these assumptions: what do you think they base them on?

04/01/2013 11:18 AMPosted by NeeberIts simple really. If the results do not meet the expectations, a false assumption was made. Without actively testing your assumptions, you cannot know if you made correct ones or not. So, saying someone who just does the number crunching but does not experiment does not have a complete picture is in fact correct.

Apparently not as simple as you'd like to believe. How do you know that, if the results do not meet the simulated and calculated results, that a false assumption was made?

You do not know. A good scientist will continue to perform the experiment and gather data to refine their assumptions further (if possible).

As established prior, WoW is not an accurate representation of reality. You seem to think that it is complex to program due to the complexity of reality itself. This is a false assumption in itself.

A person who analyzes the mass amounts of data out there for WoW can easily identify which assumptions hold true and which do not without having performed any of the experiments themselves: especially with engines like WoL. The variability you speak of will be readily apparent to the person performing the analysis. You would expect something of a normal distribution: a few outliers but most of the data is concentrated within 1 standard deviation of the true mean.

**No experiment (no in-game experience) is required at all**and this is what you truly do not understand.

04/01/2013 11:18 AMPosted by NeeberMaybe not, but I wonder if the people who play it aren't. But that aside, less complex then astrophysics does not mean simple either. ;) And it doesn't mean that you can have every significant variable without experience.

No it does not. It has its own physics engine. And that engine can be programmed in such a way that makes it easier on the programmer AND easier on a person analyzing the data because fewer variables are introduced!

We're missing the critical link here: every single variable is not needed to assess a range of values extremely close to the true mean. Our goal is not to find the true mean. It would be nice but it is often not given. We are then tasked with what range of values is within reasonable distance from the true mean.

As your sample size increases as it does with many logging programs like WoL, your interval becomes closer and closer to the actual mean value! With so much information provided to us, you REALLY do not need to step foot in-game to see what is going on because all the information is laid out there for you! All you have to do is assess the variance and determine which factors, based on your data, influence gameplay.

04/01/2013 11:18 AMPosted by NeeberI do agree. However, running the raids will help you understand the variables involved better. Just like preforming an experiment in a lab wont guarantee you covered every variable. However, it does provide insight into how things work.

What makes this different from watching a YouTube video? Do you actually think our tasks in WoW are so complex that they require you to play the game to understand what is really going on?

If yes, please provide an example to me? You may be on to something but I am unsure.

04/01/2013 11:18 AMPosted by NeeberYou can analyse data and come to some very solid conclusions. Even some very accurate ones. However, you are effectively removing a data source. You increase your likelihood of false assumptions when looking at the data only. When you experience it, you are more likely to apply the correct assumptions.

Provide an example of this please? Maybe I can help.

Let's take a random MoP raid like Elegon. There are strategies for the fight written all throughout the internet. There are YouTube videos of the fight. There are lists of his abilities, information about what he does in which phase, and even suggested tasks for certain classes.

What is this experience you are referring to when you do Elegon? What are you actually gaining? How not to fall in a hole? How movement can impact DPS?

Well I hate to tell you but this is all reflected in the logs. Fall in a hole and your DPS plummets. Up time is kept track of so if you are not managing your movement during certain phases properly, it will definitely show in the log.

Which variable in this fight can only be account for by actually doing the fight? It's an honest question.

So, if I have someone who has no practical experience telling you 'the math says' you have to take understanding that wrong assumptions are very possible. The data must be taken in that light.

Again, the math most certainly has value. But if you do not have experience you miss a valuable data set. You increase the likelihood of a false assumption. And thus, you increase the likelihood of a false conclusion.

How likely is this false assumption that you claim can appear? Not very.

Furthermore, do you honestly not think that this false assumption would remain false if you looked at another set of data to see if your results mirror it or contradict it? Again, we still haven't entered the game. What is this invaluable in-game experience? All you need is right in front of you and is actually a better representation of what is more likely to happen (since you're examining a large sample of data) compared to the in-game encounter you have which is just n=1 and has no relevance?

In fact, anything to do with multi-dotting just makes me a sad mage. There are "dot" classes for a reason, and mages should never EVER be one of them. Just remove the ability to dot more than 1 enemy at a time, and you solve a major mage problem.

Then all you have to do is fix the level 90 talents...