38 replies to this topic

[AnnoyingCat]

Too much info... brain hurts

[Kazairl]

I am still confused as to why damage is a category factor. Isn't the target damage "as much damage as possible" when dealing with weapons? Or Is it a lower limit to avoid data that includes misses?

Edited by Kazairl, 14 October 2013 - 07:51 PM.

[MustrumRidcully]

Kazairl, on 14 October 2013 - 07:51 PM, said:

I am still confused as to why damage is a category factor. Isn't the target damage "as much damage as possible" when dealing with weapons? Or Is it a lower limit to avoid data that includes misses?

of course everyone would love to deal 200 damage in 3 seconds. But this isn't possible, you can't equip enough weapons.
In addition, I want to model the impact of heat dissipation and the heat threshold. You can't do that with an abritrary number of weapons. You cannot even model it with just one weapon.

An example: Let's say you have 10 standard heat sinks and 1 PPC. Your dissipation is 1 heat per second and your heat threshold is 40. The PPC produces 10 heat every 4 seconds (and the first time at 0 seconds as we start tracking heat and damage the moment you shoot, not before.) After 20 seconds, you would overheat with this mech.
But this doesn't mean you could install 10 extra heat sinks and 1 extra PPC and also last 20 seconds. You would actually overheat after 16 seconds.
And it gets even more complicated with double heat sink mechanics.

So you can only really consider your damage potential and your sustainability in the context of at least an approximation of a build.

That is one of the issues with the current heat system. The table top heat system makes such efficiency calculations simpler, becuase you don't have a giant heat threshold you will be able to fill up. Or rather, you may be able to fill it up, but you will lose a lot of effectiveness in the process as your mech suffers heat penalties. That means that it's usually desirable to run a close to heat neutral mech. A mech like the Marauder that generates 6 waste heat in one turn (10 seconds) just when using its long range weaponry (1 AC/5 and 2 PPCs) is already a very hot build. (Thought here are hotter ones.)

Compare that to the Quad PPC Stalke back before PPC nerfs and ghost heat. That one produced over 60 points of waste heat in 10 seconds, and it was deemed quite effective - in the same time it delivered 60 waste heat, it also delivered 120 damage, which is significant in a game where a 65t mech has no more than 126 armor points at its CT (including internals and rear and front armor)...

[Almond Brown]

keith, on 14 October 2013 - 12:15 PM, said:

reading this pgi kinda did their job. though the energy based weps should be balanced around lasting 15 or so secs. a great quick and easy fix would be tweak the DHS value.

A Blasphemer! We have among us a Blasphemer! Get the torches lads. This one will burn brightly indeed. lol

P.S. Not everyone thinks PGI doesn't understand what they are doing.

[Jman5]

Hey, thanks for labeling the Y axis.

Would it be possible to space out the weapons a little more so they aren't so bunched together. It can be a little tough to read when they are all squashed together like that. Perhaps if it's a space limitation, you could do 3 different graphs for each weapon type (Ballistics, Energy, Missiles). Then you could mash the two original graphs together.

[MisterFiveSeven]

MustrumRidcully, on 14 October 2013 - 01:02 AM, said:

Just to see if my charts could be of any use at all.

Some People claim that ballistics are OP. I am not sure if that's actually true, but let's see.
snip...

You missed the most important caveat:

Pinpoint vs duration.

It really does trump most other considerations, but I wonder how much of it is psychological impact and skill vs. the efficiency of the mechanic.

[MustrumRidcully]

MisterFiveSeven, on 17 October 2013 - 08:15 AM, said:

You missed the most important caveat:

Pinpoint vs duration.

It really does trump most other considerations, but I wonder how much of it is psychological impact and skill vs. the efficiency of the mechanic.

I don't think it trumps most other considerations. It's an advantage that we haven't really tried or managed to quantify.

If we keep failing to do so... well, I am not opposed to changing all "pinpoint" weapons into beams or streams of bullets (whichever is appropriate).

When I am a grown-up, I will have my own MW:O servers to test stuff like this.

Mustrum "Metaphorically speaking, I am 32" Ridcully

Edited by MustrumRidcully, 17 October 2013 - 09:14 AM.

[MustrumRidcully]

Jman5, on 17 October 2013 - 07:52 AM, said:

Hey, thanks for labeling the Y axis.

Would it be possible to space out the weapons a little more so they aren't so bunched together. It can be a little tough to read when they are all squashed together like that. Perhaps if it's a space limitation, you could do 3 different graphs for each weapon type (Ballistics, Energy, Missiles). Then you could mash the two original graphs together.

Would you be interested in the original files or data? I have uploaded earlier versions already to Google Drive.

---

A bit of motivational idea behind the whole deal:
As long as the heat system makes heat neutral builds impractical and is instead based on a "damage race to shutdown" model, it will be difficult to find a good spot here. By nature, energy weapons are lighter than ballistics. Their heat only comes into effect after some time. It's basically impossible to pick energy weapon damage and heat in such a manner that they would be able to compete long-term with ballistics, and if they ar ecompetitive to long, ballistics will be always inferior - you can't burst, and your sustained DPS is only superior after you've already keeled over dead.

The original table top model tried (not really succeeded, but tried) to balance weapons around the heat neutrality state. The heat system made gaining heat always punishing, not just because you overhated, but because of penalties that started early.
The balance model was basically that a weapon must be worth the weight of the weapon itself, the heat sinks required to neutralize all heat, and the ammo sufficient to last a match. That made it easy to balance on a per weapon basis (and yet, FASA didn't completely succeed, and once they added powegamer's wet dream Clan and Level 2 Tech things got worse.). You balanced the weapons around their heat neutrality state, and building a mech would be about trade-offs between sustained damage (by actually equipping all the heat sinks needed to be haet neutral) or burst damage (by not actually equipping them all.) A build that was 5 heat sinks short of heat neutrality was already hot.

In M:WO, even the pre-PPC nerf and ghost heat Quad PPC Stalker with 20 DHS was basically 40 heat sinks or 20 DHS short of heat neutrality. But it didn't matter, because you could deal 120 damage before you shut down, which can destroy any 65 ton mech (and while that's assuming no misses, it's also assuming you don't use cover occassionally to cool off). How can you hope to balance weapons now? Certainly not with the TT model.

You can't. You either need a different heat system, or you need a completely different mathematical model. The one I describe above is such a model. It's hardly perfect or sufficient, it's still only a start. It can't be done to the sufficient level by some arm chair developer at home that doesn't have access to any server data and metrics to refine his model.

Edited by MustrumRidcully, 18 October 2013 - 12:39 AM.

[Corvus Antaka]

good post! thanks!

[MustrumRidcully]

Here are the charts without the missiles, for better readability.
High Damage Output Chart



Low Damage Out Chart

[MustrumRidcully]

Hey someone else is taking math in his hands: http://mwomercs.com/...es-and-oranges/

I really wonder what kind of models PGI has, if any.

[JimboFBX]

You need to integrate over a normalized bell curve centered at 500 meters with a standard deviation of 150 meters

I base that on absolutely nothing

Edited by JimboFBX, 15 November 2013 - 02:20 AM.

[Ngamok]

William Mountbank, on 11 October 2013 - 03:50 AM, said:

When I have mechs that focus on a few long range weapons but still have energy slots free (Stalker, Quickdraw etc...), I find the SL to be very effective and heat efficient as a backup close range weapon for when the action comes to close, though I'm not sure I would want to run them as a primary weapon on anything other than a maxed engine Jenner.

The new engine speed for the HBK-4P makes it a nice faster 9xSL punch.

[MustrumRidcully]

When discussing the advantage ballistic/PPC pinpoint precision vs beam duration lasers, people always bring up the trade-offs of ammo need and heat requirements and all that. These all exist, but what kind of impact do they have on a mech build? Are ballistics so much heavier that for the same weight investment in energy weapons and heat sinks I get more firepower, or do energy weapons actually need so many heat sinks that their weight advantage is neutered?

This thread is trying to give an answer on that.

I wonder if I can actually also put an approximation of the value of pinpoint precision into this - I could redo the math to also include weapon damage utilization, e.g how much percentage of a weapons potential damage do you actually inflict on average? We don't have any general statistics on that, however, only the statistics of individual posters. Still, those seem to show a trend of something like 5 % to 10 % advantage for ballistics and PPCs. But I think it would be better to look at the per weapon level, since pinpoint isn't actually all that is affecting precision in my experience, but also rate of fire (the more often you have to shoot, the less time you have for aim, the worse your precision becomes.)

[Noesis]

The damage ratings are based on potential as derived from a spreadsheet not actual real applied effectiveness in game so a weapons real effectiveness are not being measured.

===

Also to use the spreadsheet numbers looking at practical realities

2 AC10 with 6 volleys, takes 15 seconds and produces 36 heat. Heat cap 50. Dissipation (10 * 2) / 10= 20 /10 = 2.

So after 15 seconds it would dissipate 30 heat so in that time we'd expect 6 heat. Taking 3s to restore to 0 and still have lots of ammo left.

For equivalence as per the spreadsheet for MLs:

6 MLs need 4 volleys, equally takes 16 seconds but produces 96 heat.

With 6 external DHS in addition to 10 DHS = 6 * 1.4 + 10 * 2 = 8.4 + 20 = 28.4 = 2.84 Dissipation.

We also have a heat cap of 62.

Time 0s, 6 MLs = 24 heat, Heat = 24

Time 4s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 24 - 11.36 + 24 = 36.64

Time 8s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 36.64 - 11.36 + 24 = 49.28

Time 12s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 49.28 - 11.36 + 24 = 61.92



So at the end of this time the AC10's are able to continue to be fired since their ammo has not run out, which is very interesting in itself as it says that 12 AC10 shots are needed but 7 tons provide 15 shots per ton which would be 105 ammo, yet the calculation alots all of the 7 tons for the efficiency calculations not 1 ton as needed. So this extra tonnage skewing the amount of weight needed for the calculation.

And yet at the end of this time as a result the lasers have reached full heat and will no longer be able to be fired since it is their balancing ammo by equivalence.

So 2 AC10's will be firing at the end of this engagement, whilst 6 MLs will need to cool down and lose heat before being able to perform on a similar basis again, and needing to reach low heats taking about 20seconds to do this.

I don't find this a very realistic comparison of efficiency as a result.

And in real terms likely find more Heat sinks needed to be of use for multiple ML use but this of course would bias considerable more weight needed in relation to weapon weight for the ML use thus making them less efficient than the spreadsheet data would imply.

---

This in some helping to demonstrate why builds associated with lasers are more efficient associated with more managed arrangements of weapons and doing damage over time than just concentrating on the big alpha. But does help to demonstrate an advantage Ballistics have with more longevity for same damage output if the ammo is there.

Edited by Noesis, 11 January 2014 - 10:05 AM.

[MustrumRidcully]

Noesis, on 11 January 2014 - 09:23 AM, said:

The damage ratings are based on potential as derived from a spreadsheet not actual real applied effectiveness in game so a weapons real effectiveness are not being measured.

===

Also to use the spreadsheet numbers looking at practical realities

2 AC10 with 6 volleys, takes 15 seconds and produces 36 heat. Heat cap 50. Dissipation (10 * 2) / 10= 20 /10 = 2.

So after 15 seconds it would dissipate 30 heat so in that time we'd expect 6 heat. Taking 3s to restore to 0 and still have lots of ammo left.

For equivalence as per the spreadsheet for MLs:

6 MLs need 4 volleys, equally takes 16 seconds but produces 96 heat.

With 6 external DHS in addition to 10 DHS = 6 * 1.4 + 10 * 2 = 8.4 + 20 = 28.4 = 2.84 Dissipation.

We also have a heat cap of 62.

Time 0s, 6 MLs = 24 heat, Heat = 24

Time 4s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 24 - 11.36 + 24 = 36.64

Time 8s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 36.64 - 11.36 + 24 = 49.28

Time 12s, 6 MLs = 24 heat, Dissipation = 11.36, Heat = 49.28 - 11.36 + 24 = 61.92



So at the end of this time the AC10's are able to continue to be fired since their ammo has not run out, which is very interesting in itself as it says that 12 AC10 shots are needed but 7 tons provide 15 shots per ton which would be 105 ammo, yet the calculation alots all of the 7 tons for the efficiency calculations not 1 ton as needed. So this extra tonnage skewing the amount of weight needed for the calculation.

Let me clarify why the AC/10 uses the amount of tonnage it uses:

The analysis constructs the weight allotment needed to satisfy these conditions:
- The listed target damage number must be reached or exceeded
- within the time frame selected
- without overheating
- without running out of ammo
- and it must be done repeatably
- assuming the weapon is fired over the listed duration, starting at 0 seconds and then always "on cooldown", even if it satisfies the target damage number earlier.

The "without overheating" constraint is often trivial with the free DHS alone for ballistics, meaning they rarely need additional DHS investment, while the repetition clause is irrelevant for lasers, since they don't run out of ammo and for each repetition are assumed to start at 0 heat, no matter whether you were close to shutdown after the alloted time for an engagement or not.

So any ammo based weapon will have more ammo then needed for one engagement, since it has to last for multiple engagemenents, and it will also often (usually) have more ammo then is needed for the minimum requirement.
The efficiency of the weapon however also accounts for the actual ammo/damage delivered, not just for the minimum. The only ammo that is wasted is really the one that is needed because you need to equip ammo in units of 1 ton and you might not utilize a fraction of the ammo.

[Varent]

what most charts fail to take into account is the different uses for the different weapons as far as overall versatility as well. That is a fairly fatal flaw in most lines of thinking.

Example.

A Sig226 Handgun. Vs. a Tank (extreme example).

You can argue over the ability of the tank to simply level a city. However if someone stands on top of the tank with the handgun they cannot be shot and the person can simply work there way inside with a crowbar and kill the pilot with the handgun. Also while the tank allows massive armor and damage it is huge and you cant stick it in your back pocket to use as a weapon.

Obviously this is an extreme example and not even related to MWO but its to prove a point that basically.

Before you compare numbers you should also consider how those weapons will be used, in what mechs and in what situations and consider how changing them will effect overall balance.

Lets say we want to make the handgun produce the same firepower as the tank for example.

Think about it.

[MustrumRidcully]

Will you also think about it - how do you incorporate the different uses in such an analysis?
What are the different uses you speak of specifically? What traits would be important for those?

My attempt at this so far is that I don't just take one single set of values for damage/time/engagements, but different ones, and see how the weapons differ with the different parameters.
The parameters are guided by different uses. For example, an engagement time of 5 seconds - that's more or less a typical sniper/ridgehumper scenario. (But still too long for jump-snipers, you won't find multiple shots during your jump, after all.)
15 seconds? That's more like a brawl or other form of sustained engagements. (15 seconds can be enough to kill a mech in a straight firefight - not necessarily an Atlas, but possibly a medium or heavy.)


For a jump sniper scenario for example, the analysis might need something like a "jump jet recharge time", the time the "poptart" spends between pops to rest - which might not be enough time to fully cool off (waiting for that is nice, but you can't always afford it - you don't want to give the enemy the time to get into safety / cap point / your face.)


There are also properties that I don't take into account - Single damage Projectile (ACs, PPC, Gauss) vs DOTs (Laser, MG) vs Projectile Cloud (LRM, SRM, LBX). Also not how rate of fire, range or projectile speed can affect precision.
I think a lot of these projectiles we could probably approximate via weapon accuracy stats and statistical comparison of damage fired * weapon damage vs weapon damage actually dealt.

[Varent]

MustrumRidcully, on 11 January 2014 - 03:20 PM, said:

Will you also think about it - how do you incorporate the different uses in such an analysis?
What are the different uses you speak of specifically? What traits would be important for those?

My attempt at this so far is that I don't just take one single set of values for damage/time/engagements, but different ones, and see how the weapons differ with the different parameters.
The parameters are guided by different uses. For example, an engagement time of 5 seconds - that's more or less a typical sniper/ridgehumper scenario. (But still too long for jump-snipers, you won't find multiple shots during your jump, after all.)
15 seconds? That's more like a brawl or other form of sustained engagements. (15 seconds can be enough to kill a mech in a straight firefight - not necessarily an Atlas, but possibly a medium or heavy.)


For a jump sniper scenario for example, the analysis might need something like a "jump jet recharge time", the time the "poptart" spends between pops to rest - which might not be enough time to fully cool off (waiting for that is nice, but you can't always afford it - you don't want to give the enemy the time to get into safety / cap point / your face.)


There are also properties that I don't take into account - Single damage Projectile (ACs, PPC, Gauss) vs DOTs (Laser, MG) vs Projectile Cloud (LRM, SRM, LBX). Also not how rate of fire, range or projectile speed can affect precision.
I think a lot of these projectiles we could probably approximate via weapon accuracy stats and statistical comparison of damage fired * weapon damage vs weapon damage actually dealt.

I think a better study would be to individually break down weapons and mechs vs weapons and other mechs and then create scenarios that show the different interactions. But then again I think you missed my point. My point being the weapons actually do have a sense of balance right now based off of different scenarios. I would say there are imbalances created with jump jets. And the large lasers, erlarge and large pulse could use some tweaking.