59 replies to this topic

[Dolph Hoskins]

MadTulip, on 04 September 2014 - 04:06 AM, said:

Maybe i should therefor keep my secret on how to catch the biggest fish although im not good at aiming for myself then .

That was one of the first things I thought once I started noticing how accurate your representations were. All of those insolent IS freebirths aren't worthy of such knowledge.

Actually though, I thought your work on the matter came up at a very opportune time given all of the talk of balance, nerfs, and buffs that are soon to make a change, and it helped confirm and clarify a number of things I already had in mind.

[kapusta11]

Why are you trying to invent a wheel? People put a lot of effort into balancing TT you just need to take TT rules and implement them in a proper way. TWO most important issues MWO has right now, pinpoint alphas and worse than stock heat dissipation, is a direct result of not following TT rules.

Imagine TT as a system in which variables relate to each other according to simple formula A1=B1/(C1+D1)
Now increase A1 three times while keeping other variables and all the relations the same.
A2=3*A1; B1=B2; C1=C2; D1=D2;
A2=B2/(C2+D2)
The result would be broken relations between variables, for example B1=A1*(C1+D1) according to first formula but it would be B1=3*A1*(C1+D1) according to the second one, that's what broken system looks like, that's what MWO looks like.

Now relpace A with average damage that single component takes and B, C, D variables with tonnage, crit space, heat generation, heat dissipation, rate of fire etc. and you've got the picture.

Edited by kapusta11, 04 September 2014 - 05:41 AM.

[Tombstoner]

MadTulip, on 04 September 2014 - 03:38 AM, said:

It constantly doesnt feel good though. All the things you can not express explicitly are to be evaluated by statistics using game results as data. Whenever you try to crack an unknown system using statistics you should beforehand give as many known relations as you can. Then you should approximate in order to reduce variables where you can assume that your approximation doesnt include errors of relevant sizes. THEN is the point where you hammer the rest flat with statistics. Afterwards you can go back and release some of the approximations you did in order to compare different models for theire quality. Statistics wont tell you the exact solution. But they can tell you to what degree they are off. They also can be shown to be the optimal solution possible based on the data available while still not beeing accurate. I vote for choosing the optimal possible yet still error prone to a known extend solution instead of neglecting and doing whatever comes to mind.

What will kill any sort of assessment/ model building are non quantifiable factor interactions between map, player skill, mech design, PUG team work and a few others i can’t think of. I am not as mathy as some but i do use it at work for formulation optimization and stuff. primarily a program called JMP.

TT mech building, inter weapon stats are not conveniently linear. If they were then i can see modeling damage and range in a linear manner. It would make sense. but they are not. add to this a non symmetrical distribution of hard point type and location as tonnage increases further complicated things.
In short it cant be done correctly and thus becomes a subjective feel. then add in ghost heat.

[MadTulip]

Yo, but some relations can be expressed explicitly which helps reduce the number of variables. Not matter what method you use then to tune those uncertainties, even by hand, is more likely to produce good results then doing it for the full set of variables.

[poopenshire]

MadTulip, on 01 September 2014 - 12:34 PM, said:

Id say that exspecially the IS normal lasers are just not balanced well which is why they deviate from a linear distribution.

I dont understand the surface response for burn time idea, but if you have ideas there and like to join modelling i just could upload the matlab code to some svn or git.

Sorry been busy, not sure you can export from Matlab to JMP for surface responses.

A surface response is a DOE (design of experiment). its a method of mathmatical modeling in 3 dimensions that allows you to see multiple levels of interactions. Its a very very nice tools for data analysis. JMP is made by SAS Institute out of NC. Its much much better than Matlab, although it takes some getting used to.



Above is an example of the a Surface response graph, You see how 3d works to show optimum peaks based on multiple variables.

Hope this helps.

[Joseph Mallan]

poopenshire, on 04 September 2014 - 06:07 AM, said:

Thats cool! But what exactly is it showing me? A 3D bell curve?

[Lily from animove]

Iron Riding Cowboy, on 04 September 2014 - 03:50 AM, said:

I dont like this gyus vids, they are horribly biased.

Somethign that is unbalanced doesn't have a countertactic to beat it. if it does, it is not unbalanced. So he is using the term unbalanced way to easy, like all the cry kiddies scream "THATS OP"

wame with his rubbish "f2p is broken" Vid. Those companies running the "evil" p2w method are not short sighted as he claims leading to a death of that company. Those company plan this. They know this, They either create a new game when the old drops, or the "death" of a company is calculated and planned ruin. He lacks totally some simple buisiness knowledge of how buisness works outside the theory in the real hard profit driven capitalistic world..

[poopenshire]

Joseph Mallan, on 04 September 2014 - 06:09 AM, said:

Thats cool! But what exactly is it showing me? A 3D bell curve?

in this example yes, you see the optimal peak for maximum performance. You can also look for minimal performance, I call that PGI zone. The example below will tell you if a factor has no effect (no peak) and where maximum value (as defined by the study) can be obtained. Whether thats balance or optimal effects.

[Joseph Mallan]

Lily from animove, on 04 September 2014 - 06:16 AM, said:

I dont like this gyus vids, they are horribly biased.

Somethign that is unbalanced doesn't have a countertactic to beat it. if it does, it is not unbalanced. So he is using the term unbalanced way to easy, like all the cry kiddies scream "THATS OP"

wame with his rubbish "f2p is broken" Vid. Those companies running the "evil" p2w method are not short sighted as he claims leading to a death of that company. Those company plan this. They know this, They either create a new game when the old drops, or the "death" of a company is calculated and planned ruin. He lacks totally some simple buisiness knowledge of how buisness works outside the theory in the real hard profit driven capitalistic world..

So what you are saying is you don't like his opinion. Cause that is all he has given. You are free to disagree with his opinion.

[Tombstoner]

poopenshire, on 04 September 2014 - 06:18 AM, said:

in this example yes, you see the optimal peak for maximum performance. You can also look for minimal performance, I call that PGI zone. The example below will tell you if a factor has no effect (no peak) and where maximum value (as defined by the study) can be obtained. Whether thats balance or optimal effects.

This is what i thought PGI was up too in early closed beta before double armor..... boy was i wrong.

When you put in all the TT weapon stats and try to model it.... its fubar form the start. some weapons work well but its really an accident of history. If MWO had no hard points the medium laser would rule. it is incredably efficant for the heat/damage/ton. simply porting over TT stats and then trying to balance around that without some sort of presition cofactor to compensate for extreme accuracy is basically just pulling it all out of something’s behind.

The ability to make ac-40's and LRM-80's is fundamentally wrong in the context of mech durability. Thus the game becomes whatever feels good.

[Joseph Mallan]

Tombstoner, on 04 September 2014 - 11:41 AM, said:

This is what i thought PGI was up too in early closed beta before double armor..... boy was i wrong.

When you put in all the TT weapon stats and try to model it.... its fubar form the start. some weapons work well but its really an accident of history. If MWO had no hard points the medium laser would rule. it is incredably efficant for the heat/damage/ton. simply porting over TT stats and then trying to balance around that without some sort of presition cofactor to compensate for extreme accuracy is basically just pulling it all out of something’s behind.

The ability to make ac-40's and LRM-80's is fundamentally wrong in the context of mech durability. Thus the game becomes whatever feels good.

So the Hunchback IIC and Kraken with 120 tubes is wrong? How about the Longbow with 70 tubes?

[Fut]

poopenshire, on 04 September 2014 - 06:18 AM, said:

in this example yes, you see the optimal peak for maximum performance. You can also look for minimal performance, I call that PGI zone. The example below will tell you if a factor has no effect (no peak) and where maximum value (as defined by the study) can be obtained. Whether thats balance or optimal effects.

Laughed so hard.

[stjobe]

Tombstoner, on 04 September 2014 - 11:41 AM, said:

simply porting over TT stats and then trying to balance around that without some sort of presition cofactor

Yes, basically this is one of the fundamental mistakes PGI made with MWO. In the name of "e-sports" and "competitive gaming", they unwisely decided to implement perfect accuracy - basically hoping that they could somehow port over a weapon and armour system designed for semi-random accuracy and making it work.

They couldn't.

Pinpoint accurate, instant convergence, front-loaded alphas simply break the TT armour system. PGI knows this, but they can't do anything about instant convergence (due to HSR issues), they won't do anything about pinpoint accuracy (due to the aforementioned "e-sports" pipe-dream), so they're left with fiddling with heat, beam durations, and burst sizes to balance weapons.

It won't work.

Unless they either rework HSR to work with non-instant convergence, or implement some kind of less-than-perfect accuracy (e.g. movement-based cone of fire), we will never get away from the armour-breaking damage numbers we can casually place in one location - and that means we will never have weapon balance, because not all weapons are FLD (or non-FLD, for that matter).

Of course, the other major mistake PGI made was with the heat system, keeping dissipation at TT rates while increasing rate of fire several times (2-3 times on average), not implementing a sliding heat penalty scale à la TT, which means it is always more effective to alpha as often as you can than to chain-fire - especially with the above pin-point, instant convergence, front-loaded issues.

The most maddening thing about that is that all the mechanics for heat penalties is already in the game; movement speed reductions, reticule shake, ammo explosions, and shutdowns. They just have to couple them to the heat level and they're done.

Anyway, yeah. Porting half the TT rules and making the other half up on the spot didn't work so great.

[Tombstoner]

Joseph Mallan, on 04 September 2014 - 11:44 AM, said:

So the Hunchback IIC and Kraken with 120 tubes is wrong? How about the Longbow with 70 tubes?

Its not the number of missiles it’s the way they move and hit a mech. It becomes one giant blob of damage going for your CT.
lrm 20's typically delivered in TT 10-12 hits. 2D6 for number of missiles hit i forget the exact number. but delivered in groups of 5 damage randomly distributed. Altered for MWO.

You know very well its grouped weapons with perfectly precise shots that alters inter weapon balance. so no at face value 2 ac-20 clan ultras is not the problem, 2 of them linked guaranteeing hitting the same spot however is the issue. we might as well up date the weapons list to include multi weapon combos when discussing balance.

[Tombstoner]

stjobe, on 04 September 2014 - 12:12 PM, said:

Yes, basically this is one of the fundamental mistakes PGI made with MWO. In the name of "e-sports" and "competitive gaming", they unwisely decided to implement perfect accuracy - basically hoping that they could somehow port over a weapon and armour system designed for semi-random accuracy and making it work.

They couldn't.

Pinpoint accurate, instant convergence, front-loaded alphas simply break the TT armour system. PGI knows this, but they can't do anything about instant convergence (due to HSR issues), they won't do anything about pinpoint accuracy (due to the aforementioned "e-sports" pipe-dream), so they're left with fiddling with heat, beam durations, and burst sizes to balance weapons.

It won't work.

Unless they either rework HSR to work with non-instant convergence, or implement some kind of less-than-perfect accuracy (e.g. movement-based cone of fire), we will never get away from the armour-breaking damage numbers we can casually place in one location - and that means we will never have weapon balance, because not all weapons are FLD (or non-FLD, for that matter).

Of course, the other major mistake PGI made was with the heat system, keeping dissipation at TT rates while increasing rate of fire several times (2-3 times on average), not implementing a sliding heat penalty scale à la TT, which means it is always more effective to alpha as often as you can than to chain-fire - especially with the above pin-point, instant convergence, front-loaded issues.

The most maddening thing about that is that all the mechanics for heat penalties is already in the game; movement speed reductions, reticule shake, ammo explosions, and shutdowns. They just have to couple them to the heat level and they're done.

Anyway, yeah. Porting half the TT rules and making the other half up on the spot didn't work so great.

If we only had 80 million...... they cant link heat level with that stuff.... it ruins the FPS dynamic. no more fast paced combat. everything becomes heat managment. i expect they have tried some of this stuff and found it lacking. We how ever would love it, but we are not the target audiance. the heat neutral mech has definate advantages but that advantage was paid for in tonnage.

Thats the part i dont understand about PGI's advertion to heat neutral mechs. The game has them, but PGI chose to reinvent heat by increacing it, giving players the ability to accumulat it, but giveing them sub par disapation rates. personaly i think it's to sell cool shots.

[stjobe]

Tombstoner, on 05 September 2014 - 03:44 AM, said:

Thats the part i dont understand about PGI's advertion to heat neutral mechs. The game has them, but PGI chose to reinvent heat by increacing it, giving players the ability to accumulat it, but giveing them sub par disapation rates. personaly i think it's to sell cool shots.

It's because heat neutral 'mechs breaks MWO's badly designed heat system.

In TT, heat neutrality is commonplace, and not game-breaking at all. A lot of 'mechs are indeed designed explicitly to be heat neutral over large portions of their engagement range, which is only natural; I mean who would design a military vehicle that couldn't use all or even most of its weaponry without shutting down or suffer severely degraded performance?

To give an example I used in another thread: In TT, you can fire three IS medium lasers while walking for as long as you care to (3 * 3 heat for the lasers + 1 heat for walking = 10 heat) using only 10 Single Heat Sinks. In MWO, doing the same would shut you down from overheating in 20 seconds (25 if you've mastered the chassis).

The heat system is just plain broken, and it's a major contributor to why stock 'mechs suck so badly they had to be replaced by Champions.

[MadTulip]

Patchday!

Updated OP
- plots updated to current state of weapon values
- code V4 now supports reading from weapons.xml so i dont have to type in all the values each time they change weapon values
- history of the relevant game files is also stored with date and code

It looks like the 12vs10 plan just went over board. The small and medium clan lasers where brought "in line" with similar IS weapons. I assume this is the first step to nerf clan back to IS level. We are heading into the direction where IS and clan will be at equal strength.

Two weeks after the Nova came out for CB most of the weapons mountable on that platform have been nerfed down to "rest of the game" (balanced ) which is why i assume that in 4 weeks from now after the direwolf was out for 2 weeks for CB the clan ACs and UACs will be nerfed down to IS level as well. Mark my words . If so i guess it can be called a buisness model to sell 30% overpowered things for $$$ and nerf it to normal once they are "for free". i just recall the times when the IS UAC5 was nerfed 2 weeks after the ilya and firebrand where on sale for -35% MC but thats a different topic and lets not start fueling the fire here but just watch and plot .

The way the clan small and medium lasers where brought in line also suggest that they know where theire real value is located. Is that a good or a bad thing for the above mentioned UACs/Direwolves ?

Edited by MadTulip, 05 September 2014 - 11:35 AM.

[Mothykins]

I love the actual analysis being put into this.

All my love for the constructive thing you got going on here!

[poohead]

Can you add a chart that show only the variance between IS and Clan for each weapon? Where the origin on the chart is effectively the IS weapon

So if you plotted the variance for the CERSL it would be at roughly 2 on the x axis (range variance) and -0.1 on the y axis (efficiency variance), the CERLL would be about 1.5 on the x axis and 0.1 on the y axis. (this is the best I can read from the charts!)

You could plot every clan weapon variance on one chart which might be easier to read than seeing all weapons individual stats at once.

Or it may look awful, I have no idea!

[MadTulip]

poohead, on 05 September 2014 - 01:13 PM, said:

Can you add a chart that show only the variance between IS and Clan for each weapon? Where the origin on the chart is effectively the IS weapon

So if you plotted the variance for the CERSL it would be at roughly 2 on the x axis (range variance) and -0.1 on the y axis (efficiency variance), the CERLL would be about 1.5 on the x axis and 0.1 on the y axis. (this is the best I can read from the charts!)

You could plot every clan weapon variance on one chart which might be easier to read than seeing all weapons individual stats at once.

Or it may look awful, I have no idea!

Yea i can do that tomorrow. Want to field some CUAC20 now .

Edited by MadTulip, 05 September 2014 - 01:17 PM.