Fractal Softworks Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Starsector 0.97a is out! (02/02/24); New blog post: Simulator Enhancements (03/13/24)

Pages: [1] 2 3 ... 7

Author Topic: Vanilla Weapon Balance Sheet  (Read 25891 times)

xenoargh

  • Admiral
  • *****
  • Posts: 5078
  • naively breaking things!
    • View Profile
Vanilla Weapon Balance Sheet
« on: April 18, 2018, 10:26:58 PM »

This is the latest version of my attempts to map out weapon balance.  This comes with a chart showing relative efficiency and outliers, for people who just want the Cliff Notes.  This formula will get used for the next version of Rebal; I'm fairly certain it's reasonable and sane, as it built on previous work on Rebal, but includes a lot of improvements to my approach.

Google Sheet: View Here

I'm not 100% certain, but I think people should be able to make a copy of the Sheet and view / edit the formulas on their copies.  I went back to basics on everything; there aren't many fudge-factors (and those that are there are clearly explained).  

Why are there any fudge factors at all?  Isn't this just math in, math out?

1.  Well, first off, any comparison exercise needs a par value.  We don't have any "ideal weapon" candidates, so I had to choose something.  Every weapon is par-balanced against the Pulse Laser.  

I think that it's about the closest thing to par we can find in the core game's weapons; it's about 1:1 efficient on Flux, 1:1 on damage, perfect accuracy, medium range, and a slight bonus from having faster-than par shots (i.e., it'll hit smaller targets than par a little more reliably than usual).  If anybody wants to argue for a superior case, please make that argument, with some math to back it up.  But, as we all appear to agree that it's All Right, that's what I ended up using.

This meant adjusting the value of raw damage output and the relative value of range for all other weapons to compare to par.

2.  Valuing different types of damage requires judgement.  The absolute value of KINETIC vs. HIGH_EXPLOSIVE vs. ENERGY can be argued until the cows come home.  Basically, when the dust settled, I decided to treat ENERGY as "neutral" and all other types compared to it.  I may have over/undervalued KINETIC or HIGH_EXPLOSIVE or FRAGMENTATION, but I suspect it's reasonably close.

My previous version of this used a different approach, whereby I was measuring TTK, instead of measuring average damage outputs.

3.  Like the previous versions, this ultimately measures efficiency for each OP spent.  The total_efficiency value does not tell us how "good" a gun is.  That's largely about designing the parameters of the weapon to suit a particular role.  This value simply tells us whether the gun's "worth" more, per OP spent, than other weapons, using the combined data.  Think of OPs as points we'd spend, in any game system where we build up game objects (OPs are essentially like Tonnage in Battletech, which is clearly where Alex started the idea).

I understand that this approach seems pretty confusing to a lot of people.  They're hoping to see a simple chart that says, "X is OP, here's why, with Math".  

But it's not that simple.  Weapons aren't just their efficiency values; a weapon can be "efficient" but have a design with marginal utility, like the AM Blaster.

Some weapons are great bargains for their OP costs, like the Light Mortar (OP cost of 2, surprisingly-good stat lines); it's just too bad they're not terrific in any way that would make them attractive vs. other choices.  

The biggest surprise in the findings, at least to me, was seeing that the Thumper is kind of a serious bargain, lol.  It's just too bad it has stats that make it pretty lousy for any particular niche role.

Other weapons are designed pretty well for their intended use case, and are also OP bargains, like the Railgun; it has perfect accuracy, above-average range, attractive Flux stats, pretty decent DPS in all categories for a lightweight Kinetic.  total_efficiency helps us understand when a gun might be over / under-priced; it won't tell us to fix it in any particular way.  That's the right way to use something like this; don't limit people's creativity by forcing weapons design into pigeon-holes, just give folks a tool to construct reasonably-balanced stuff with.

4.  I haven't engineered the formulas to arrive at pre-built assumptions about what's "best" or to prove my biases.  Some stuff really surprised me, frankly, but the math's been looked at over and over again and I'm fairly certain there aren't any big mistakes in the core formulas.  

The factors regarding damage-type weighting can be explored and argued about if people want, but I feel I was quite fair about damage to Armor; that and the value of range and accuracy are all arguable points.  I feel like this is a very good starting-place, though.

5.  Accuracy is an average between min and max.  I realize that's not perfect, as the vast majority of weapons are 100% accurate on their first shot and degrade.  But it's extremely difficult to write anything more satisfactory.  For what it's worth, in Rebal, I don't use different min/max values or use accuracy degradation; if a weapon's supposed to be inaccurate, it just is- this keeps balance easier to maintain.

6.  Alpha and rapid-fire guns are inherently covered by kill_power, as is Flux efficiency; kill_power is not just TTK; it's DPS with Flux efficiency applied.  This produces some interesting results; a weapon can do tons of damage but have a pretty lame kill_power, because it's so inefficient on Flux; this represents a gun that can only be used situationally, and that's why it's penalized vs. guns you can fire all day long and expect good results.

7.  How can this be used to arrive at better overall balance for weapons?  It's pretty simple, honestly:

A.  Simple buff / nerf on OPs could get things roughly par without changing anything else.  That would mean a few guns got cheaper, a few got considerably more expensive, etc.

B.  But, if we're trying to keep OP values the same as Vanilla (which is usually what people have said they'd prefer), then buff / nerf of the "god stats" (i.e., range, DPS factors, Flux efficiency, accuracy) gets us there.

C.  For guns that are already bargains, OP-wise, but still suck because they have no clear niche, define that niche through better design.  For example, I made the Thumper actually relevant in the last private builds of Rebal by making it a rapid-fire "burp gun"- high Frag DPS, mediocre accuracy, long bursts with long pauses between.  Suddenly, I had a weapon that killed fighters half-way decently and was efficient enough that putting one on a ship wasn't a huge waste.

8.  Why is the Plasma Cannon dead last?  I mean, that gun rocks, right?

That's not terrifically complicated.  Basically, the bigger guns that eat more OPs are all pretty low on the efficiency curve; this was a consistent finding.  This is something I've brought up before, and it bothers me quite a lot from a game-design POV; a weapon using a rare, precious Large Slot should be more efficient than a common Medium, not less, imo.  

But the numbers explain things pretty clearly.  While the Plasma Cannon is great for what it does, it's not terrifically efficient.  

Average kill_power, if we drop the huge outliers (see below) is around 4.5 or so.  It's at 3.6; it does a lot of damage, but it eats a lot more Flux than the damage it outputs.  Other than that, the weapon's just a little better than average in terms of range, etc.  So, while it has perfect accuracy and great core DPS, it's not really all that super, unless it's pounding a target that's near Overload or doesn't have a shield facing the weapon.

The Plasma Cannon's a perfect example of designing a gun to do a job well... and then pricing it incorrectly.

9.  Why'd you drop TTK?

I didn't; the DPS vs. various things is really super-clear now.  Kill_power represents TTK vs. Flux efficiency, with accuracy factored in; if your shot misses, then Flux is being wasted; this is why inaccurate guns often feel much weaker than their core stats would otherwise suggest.

10.  What's with the huge outliers?

They're just... there.  People have been saying for years that, under the right circumstances, LMGs and Vulcans make great assault guns.  They're totally right, because these weapons are massively cheap for what they do.

Honestly, seeing those outliers made me question whether 600 range (the Pulse Laser's range) was the right value to hinge the value of range in general on, because sure, they're great guns, if you don't mind closing to belly-button ranges.  But as I said, just because total_efficiency is "good" doesn't magically make the weapons good; in the end, it's just a numeric tool that tells us whether the weapon's too expensive or cheap for what it does.  Game designers make the weapons good or bad; that's a separate issue from whether they're par-balanced (these things are related, though).  I'm pretty sure that these weapons can be made par-balanced and yet still do their main jobs (being cheap PD, not crazy-good assault guns).

11.  You've been saying Energy weapons suck, but they're pretty evenly distributed on the chart.  What's up with that?

First off, re-read #3 until you get what I'm saying.  Really understand that total_efficiency isn't a magic, "this gun is OP" tool; it merely flags weapons as over/under-priced for their OP costs, compared to the par weapon.

Second off, I don't think all Energy weapons suck.  Just most of them.

I was a little surprised at the ranking of the IR Pulse, personally; but the shot speed bonus helped it a lot (it's double the range- the shot arrives at the end of its range in half a second).  So that makes some sense; it doesn't miss much against speedy targets; that's kind of valuable.  If it had a standard shot-speed, it'd probably need some help to be OK.

I was really surprised by the ranking of the Mining Laser.  It still sucks, but not because it's inefficient, but rather, that it has no clear job it does well.  It lacks the burst alpha to kill missiles or damage fighters reliably, yet it's also not Flux-efficient enough to use as a trader unless you can win with practically anything.  It's quite efficient, but not well-designed.  Can it be in a better place?  Yes, it can; either accept a range reduction in return for much better Flux efficiency, much higher damage output in exchange for being pulsed with long pulse cycles, etc., etc.- there are plenty of valid choices.



Anyhow, that's it.  I was actually pretty happy with the current formulas for Rebal, warts, odd factors and all.  But this seemed like a good time to take one more whack at a purist set of formulas and this is probably as close as we're going to get; now I have to apply all of this to Rebal and see what happens in playtesting; I'll report back if the fudge factors needed tweaking, etc., if anybody's interested.  
« Last Edit: April 18, 2018, 10:29:48 PM by xenoargh »
Logged
Please check out my SS projects :)
Xeno's Mod Pack

intrinsic_parity

  • Admiral
  • *****
  • Posts: 3071
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #1 on: April 18, 2018, 10:45:19 PM »

Your 'kill power' parameter seems flawed. I skimmed through and all the weapons with high kill power are either kinetic, PD or both, with the last three weapons (lmg, dual lmg and vulcan) having a kill power an order of magnitude higher than almost anything else. The graviton beam also has a well above average kill power. Clearly this is not actually reflective of the ability to kill average size ships. These weapons tend to have high dps and low flux cost which is likely why they appear to be so effective via this metric. The fact that the lmg and dual lmg are rate 10x higher than 90% of the other weapons indicates that this definitely need tuning. Perhaps it is a useful metric but I think 'kill power' is likely not an accurate term.
Logged

AxleMC131

  • Admiral
  • *****
  • Posts: 1722
  • Amateur World-Builder
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #2 on: April 19, 2018, 12:35:22 AM »

Another attempt at a "universal weapon comparison"?  :-\ Those are supposed to be bad, right?

> Forgive me in advance for sounding a little harsh, but I'm getting progressively more and more confused by your ideas of "balance", xenoargh.



Now, I haven't looked at the spreadsheet because I'm honestly too scared to, but even without that I have a few problems with some of the statements you've made in this post. Here are my thoughts on the biggest issues that caught my eye:

The biggest surprise in the findings, at least to me, was seeing that the Thumper is kind of a serious bargain, lol.  It's just too bad it has stats that make it pretty lousy for any particular niche role...

... I made the Thumper actually relevant in the last private builds of Rebal by making it a rapid-fire "burp gun"- high Frag DPS, mediocre accuracy, long bursts with long pauses between.  Suddenly, I had a weapon that killed fighters half-way decently and was efficient enough that putting one on a ship wasn't a huge waste.

The Thumper has great stats. Where are you getting this "pretty lousy" from? It's got decent range for a medium weapon, outstanding burst damage (2,000!), amazing flux efficiency, and a sweet spot of accuracy that gives a nice balance between on-target pounding and area denial. And all of that for 9 Ordnance Points.

I think you're underestimating the potential of Fragmentation damage here. Sure, it only does 25% against shields and armour, but what if a target is overloaded, and its armour is already stripped? Or if it has front shields and weak armour and you can duck behind it to fire into its engines? A Thumper can delete a frigate in the right situation. (And when it comes to comparing weapons, situation is everything.)

Then there's this line of yours: "making it a rapid-fire "burp gun"- high Frag DPS, mediocre accuracy, long bursts with long pauses between." Forgive me if I've missed something massively obvious, but that's exactly what the Thumper already is???


5.  Accuracy is an average between min and max.  I realize that's not perfect, as the vast majority of weapons are 100% accurate on their first shot and degrade.  But it's extremely difficult to write anything more satisfactory.  For what it's worth, in Rebal, I don't use different min/max values or use accuracy degradation; if a weapon's supposed to be inaccurate, it just is- this keeps balance easier to maintain.

I hope you somehow factored projectile speed into your accuracy scoring. Especially when it comes to beams.

It's also not at all perfect to average accuracy between min and max, because again, it depends on the situation and weapon's behaviour whether it's going to be firing constantly or only periodically. The Light Dual Autocannon, for example, is the bane of me because of its horrendous accuracy. It degrades quite fast and ends up with a massive divergence, which would be fine if it wasn't a weapon I expect to be firing consistently for long periods of time (even if it's under Autofire control). However shields are generally bigger targets than the ships they protect, so in a way it makes sense, right?

In contrast, if a weapon has great or even perfect accuracy, but isn't going to be firing more than one shot at a time (Antimatter Blaster for instance), then is accuracy a good marker to judge it on? Is it even relevant?


8.  Why is the Plasma Cannon dead last?  I mean, that gun rocks, right?

That's not terrifically complicated.  Basically, the bigger guns that eat more OPs are all pretty low on the efficiency curve; this was a consistent finding.  This is something I've brought up before, and it bothers me quite a lot from a game-design POV; a weapon using a rare, precious Large Slot should be more efficient than a common Medium, not less, imo.

You seem to be forgetting that the sort of ships that can comfortably mount a large weapon (mostly cruisers and capitals) generally have far better flux stats than ships that can only comfortably mount a medium, and the same goes for smaller and smaller ships and weapon sizes. The size of a weapon shouldn't have any effect on its relative efficiency, since the bigger the ship, the bigger a gun it can handle.

The Plasma Cannon has the highest sustained flux cost of any stock weapon, and is also the most expensive weapon to mount OP-wise, but surely that's all as designed? Don't forget it does 1,000(!) energy damage per shot, in bursts of 3, and is the second best non-missile weapon for cracking armour after the Hellbore. It's like 3 energy-based Atropos torpedoes every four seconds. If you want that beast to be cheaper on flux - or OP for that matter - then your ideas of "balance" need some rethinking.

(Ever seen an SO Plasma Cannon Sunder?)



I'm sure you mean well and all, and that you have sensible motives for making such statements and decisions in your Rebalance Pack, but I really do have to question such things. Heh, it's in my nature to ask about things I don't understand, and for the record xeno, a lot of stuff comes out of you that I just don't understand. :-X

... That's probably enough.


Logged

TaLaR

  • Admiral
  • *****
  • Posts: 2794
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #3 on: April 19, 2018, 01:16:18 AM »

LMG is great under ideal conditions - sure, but when does this happen?
Lasher is the only vanilla ship able to properly capitalize on LMGs. While being probably the most cost-efficient ship for campaign play (Lasher is common, weapons are open market) I wouldn't call it overpowered.
Any sufficiently faster frigates can kite LMG Lasher to death (which is not hard, since Lasher has only average base speed), while DEs that have enough firepower to overload it during approach (=any properly outfitted) are safe from it.
Of course, AI often has weak variants or gets distracted, but that's not LMG Lasher's fault. It just enjoys it's high risk - high reward niche.
If you were to nerf LMG to bring it in line by your metric, it would become completely useless (it's already bad as PD anyway).

Similar considerations go for Vulcan - it's going to miss a lot as PD, so firing it has to be very cheap. And with so short range it needs massive dps to intercept anything.
As anti-ship its good for finishing hull at short range after shield and armor have been stripped. Well, I think you deserve to win massively, if you managed to satisfy all these prerequisites.

Or take Mining Laser - high tech ships tend to be underslotted, if they use mostly cheap and weak weapons, they just end up not using their flux allotment.
And what exact scenario is it good for? As PD it doesn't do enough during missile approach (though I didn't use them that much, 600 range *seems* enough to have at least some effect).
As offensive weapon being inefficient soft-flux makes firing it worse than holding fire against most shields (and dubious against armor due to extremely low damage) - it just nears your loss in flux war in most scenarios.

Same high tech underslotted-ness relative to flux can make Plasma Cannon attractive choice. Apogee just can't effectively use all it of it's flux in any other offensive way (medium rear slots are non-starters for Heavy Blasters due to horrible placement and arcs).

So I guess the moral here is - many outliers by your formula are there for good enough reasons.

I hope you somehow factored projectile speed into your accuracy scoring. Especially when it comes to beams.

So much this. Perfectly accurate, but slow projectile can be useless against most non-Capitals, especially at long range. I tend to view projectile speed, range and spread as single effective stat boiling down to "chance to hit maneuverable target at given range".

Also combined high speed + lack of windup has somewhat non-linear benefits. Being able to hit enemy faster than they can raise shield is very beneficial (in jump-shot, phase frigate or enemy re-raising shield scenarios). Or to provide passive ( free! beat this HE :) ) anti-armor threat by just pointing Heavy Blaster at target.
« Last Edit: April 19, 2018, 01:48:03 AM by TaLaR »
Logged

Tartiflette

  • Admiral
  • *****
  • Posts: 3529
  • MagicLab discord: https://discord.gg/EVQZaD3naU
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #4 on: April 19, 2018, 02:00:41 AM »

Radical idea: Range is irrelevant when comparing what weapons to mount since it acts as a threshold.

(put down the torches and hear me out)

The range isn't exactly a parameter you compare the weapons with because you generally build loadouts with consistent range: for a slow ship you want long range weapons while for fast ships you can afford to have lower range. And a lone sledge gun on your cruiser is almost useless (in non-pursuit battles) if the rest of your build has 300 range. So the range isn't the parameter you use to compare the weapon but the one to sort them out! For a given range band you can compare all weapons more ranged as equals regardless of the extra range (except maybe to break a tie).

Effectively, if say Mauler is a "8" (totally arbitrary value for the sake of argument) HE weapon at 1000 range, it is also a "8" at 500 range. The heavy mortar on the other hand is a "9" at 500 range (way cheaper, similar performances) but exactly a "0" at 1000 range. So if you build a SO knife fighter Enforcer, the Mortar is better, but if you want to poke enemies at a distance, it's useless.

In other words all the other stats are trade-offs, but you could say range is the all-important one that serves to sort which weapons to look at, not to compare them, since a weapon that is shorter ranged than your target engagement distance is but wasted weight.

All the benefits from range are factored in when you choose what type of loadout you want: the opportunitiy benefits from range are abstracted when you decide "nope, I will build a SO Dominator for knife fighting range". But your choice of engagement range is not decided on a weapon-per-weapon basis, it's an overarching idea, and factoring range isn't that usefull to compare weapons.

All it boils down to is that you don't think "I want the best gun regardless of range" when building a ship, but you do think "what would be good 800+ range HE weapons to mount in those three slots?".
Logged
 

TaLaR

  • Admiral
  • *****
  • Posts: 2794
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #5 on: April 19, 2018, 02:24:30 AM »

@Tartiflette
You do have a point that different range offensive weapons rarely compete for a mount, but shorter ranged ones definitely need to have significant edge at their ranges to make whole approach of a knife-fighter viable (either due to inherent stats or stats of ships that use them).

Then again - take Desdinova (I found no relevant vanilla comparison). I can take long ranged Maulers or shorter ranged, fast projectile HE guns with higher single-shot damage (forgot their name, also BRDY). Both can be viable (because I dictate engagement range anyway) within mostly same variant(in terms of other weapons/hullmods/etc).

There are also some weapons sharing same niche, but having different ranges.
Light AC = Dual AC < Railgun < Light Needler.
- Dual AC: shortest range, OP cheap. Worst one, but serviceable unlike Light AC.
- Railgun: shorter range, can threaten frigate/fighter armor, highest dps.
- Light Needler: max range, best efficiency, worst against armor.

« Last Edit: April 19, 2018, 02:42:40 AM by TaLaR »
Logged

Megas

  • Admiral
  • *****
  • Posts: 12118
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #6 on: April 19, 2018, 07:12:02 AM »

Some quick comments:

The main thing plasma cannon has over heavy blaster is range.  If it did not have 700 range, I would use Heavy Blaster instead of Plasma Cannon.  Plasma Cannon is too expensive and too inefficient.  For mod ship with large universal, it is filled with Mjolnir, Gauss, or Tachyon Lance (or other overpowered mod weapon).

Mining Blaster is so bad that if Hyperion was not in the game, I would not use it.  Terrible range and terrible flux efficiency, it is really awful.  On Hyperion, it is useful because Hyperion does not have the flux stats to fire two Heavy Blasters more than once after teleporting and have enough flux left to teleport away, but it can get flux stats to fire two Mining Blasters once and have some left to teleport away.  For Hyperion, Mining Blasters are stronger, and two of them are like an AM blaster with a much faster recharge time.  Range is not a problem for Hyperion.

Mining Laser is also bad because it tracks nearly as slow as Tactical Laser.  The low damage is bad enough, although it might have been tolerable if it tracked swiftly.  But it is slow enough that it is not hard for incoming to outspeed mining laser tracking.  If you think about getting Advanced Turret Gyros to fix that, you spend enough OP that you might as well upgrade to PD or LR PD (which are somewhat common).

When player is stuck with Open Market weapons early in the game, Thumper is good because it is fast (and Arbalest and Mortar are slow), and it will destroy armor-stripped ships.  I have been on the receiving end of Thumper kills a few times, at least during early 0.8.

Railgun's main weakness is long windup.  Sometimes, this matters.

Sometimes, weapons with superior range is not as useful because other weapons cannot reach as far to compliment.  For instance, 900 range heavies and 800 range mediums (instead of Mauler/HVD) on Legion, and Legion has fighters to mitigate weapon range.  Of course, extra range can help match ranges if ship has mounts front and back.  For example, Brilliant is a great sniper with Maulers in hardpoints and Gauss in the large universal in the rear.

For PD, all I care about is how quickly and effective it is in stopping missiles and/or fighters.  Range is not useful enough if it cannot stop incoming fast enough.  For small hybrids/universals, Vulcan is almost an auto-pick.  For medium slots, dual flak if I can afford it.
« Last Edit: April 19, 2018, 07:19:13 AM by Megas »
Logged

xenoargh

  • Admiral
  • *****
  • Posts: 5078
  • naively breaking things!
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #7 on: April 19, 2018, 07:33:59 AM »

Just wanted to say that I've already found some problems; I apologize, I got excited and released it a little too early, I think.  That's what happens when I don't get enough sleep over two days, lol.

Mainly, my formula for kill_power is pretty unsatisfying; while it works for spotting the outliers, it's not great at understanding certain weapons properly; I'm sure in the critiques above, that's been noted (hopefully without too much flaming this time?).  I'm really going to have to do something about TTK, also; the lack's skewing stuff for sure.

I'll read over the responses soon, and I welcome reasonable critique; this was a good attempt to build something with clarity as to where I'm getting the numbers.  It's still not perfect, but at least it's cleaner math than the versions I've gradually chipped away at for Rebal, where I was aiming at specific problems I wanted to solve.

Quote
I hope you somehow factored projectile speed into your accuracy scoring.
Definitely; see the final formula for clarity.  Essentially, it's:

+(if(AB7>0,if(AB7/F7=1,0,if(AB7/F7>1,((AB7/F7)-1)*(1/30),(((F7/AB7)-1)*(1/30)))))))

Which, boiled out of Formula Land, basically means, "if the shot's moving faster than average, include a smallish factor to balance at the end of the other stuff that's going on".  It's not a giant factor, but, in a few cases, it's essentially nullifying the effects of accuracy.  Not sure that's perfect, but it was a fudge-factor I brought over from Rebal.

I can do the same for Beams, but frankly, I think that's a binary switch; either Beams hit instantly, or they take a very short period of time.  But because they get corrected so quickly compared to projectiles, it's going to have to be a pretty minor buff.

I'll definitely include windup as a minor nerf, but... bear in mind, most of this has to be done in the context of AI's shooting at one another.  So, it'll be a minor nerf; every once in a while, windup means a miss, but usually no.
« Last Edit: April 19, 2018, 07:40:05 AM by xenoargh »
Logged
Please check out my SS projects :)
Xeno's Mod Pack

xenoargh

  • Admiral
  • *****
  • Posts: 5078
  • naively breaking things!
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #8 on: April 19, 2018, 09:05:29 AM »

@Tartiflette
You do have a point that different range offensive weapons rarely compete for a mount, but shorter ranged ones definitely need to have significant edge at their ranges to make whole approach of a knife-fighter viable (either due to inherent stats or stats of ships that use them).

Then again - take Desdinova (I found no relevant vanilla comparison). I can take long ranged Maulers or shorter ranged, fast projectile HE guns with higher single-shot damage (forgot their name, also BRDY). Both can be viable (because I dictate engagement range anyway) within mostly same variant(in terms of other weapons/hullmods/etc).

There are also some weapons sharing same niche, but having different ranges.
Light AC = Dual AC < Railgun < Light Needler.
- Dual AC: shortest range, OP cheap. Worst one, but serviceable unlike Light AC.
- Railgun: shorter range, can threaten frigate/fighter armor, highest dps.
- Light Needler: max range, best efficiency, worst against armor.
I completely agree on all points. 

The numbers do reflect that, but they don’t currently reflect that the difference in range for the Light Needler makes it best in class, by a little bit.  Range is definitely a god-stat; if your ship can also kite indefinitely or alpha before the enemy can strike back, you win.  It also impacts AI behavior; I made Shepherds work by sticking Tac Lasers on them, not to do damage but to kite properly. 

Short-ranged weapons, in general, have to be buffed somewhere to compensate.  I’ll be tightening the relationship between range and overall efficiency when I do the next pass and put TTK back in, to account for the weapons with great bursts or high overall damage (this will pull the Plasma Cannon and Devastator out of the dumps a little bit).
Logged
Please check out my SS projects :)
Xeno's Mod Pack

intrinsic_parity

  • Admiral
  • *****
  • Posts: 3071
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #9 on: April 19, 2018, 09:08:31 AM »

Some additional Ideas for building on this idea. Rather than using pulse laser as baseline, why not average weapon stats and then normalize by average. It also might be wise to compare weapons within categories (maybe start with mount size). There are some cases where you might downscale a mount (like light needler in medium slot or heavy blaster in large slot) but in general, weapons are competing with other weapons of the same size. I think this goes hand in hand with Tartiflette's comment about range since increasing weapon sizes tend to have increasing ranges. They are generally categorically different.

Also I'm not 100% sure your DPS are all correct. Some of the weapons that fire in bursts seem low? Mark IX is listed as 135.9 hull dps while the wiki has 348 dps listed. You should take a look at that, it is likely influencing some of your other metrics.

You could also account for range non-linearly, maybe exponentially or logarithmically.
Logged

Blothorn

  • Commander
  • ***
  • Posts: 116
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #10 on: April 19, 2018, 10:55:29 AM »

Some thoughts from an operations researcher:

As the saying goes, all models are wrong; some are useful. Weapon balance is deeply multidimensional--DPS, burst, hitstrength, damage type, efficiency, range, accuracy, velocity, OP, mount size, sometimes ammo, scripted effects. This is a good thing-, as this multiplicity of dimensions helps allow distinct weapons that do not dominate each other. However, it does make comparisons rather difficult: there are several approaches, but all are rather limited in some way.

  • Pareto comparison. The *only* possible zero-assumption comparison of vectors with incommensurable dimensions is to asses whether one is superior in every respect. This is important, particularly for modders--if a modded weapon Pareto-dominates a vanilla weapon (or vice-versa) it needs to be changed. But with so many dimensions Pareto dominance is rare, and so usually a comparison requires more assumptions.
  • Strong assumptions. This seems to be the approach of the OP's spreadsheet--assign weights to the various dimensions and collapse dimensionality heavily. Such a number is tidy, but given the variety of situations in SS, I can think of no situation in which I would choose between, say, a heavy mauler and a HMG by comparing their respective "aggregate efficiency" numbers.
  • Limited dimensional reduction. Instead of assigning weights to all dimensions, collapse what dimensions you can with highly-general assumptions and leave multidimensional output. In particular, I think Tartiflette is exactly right that range should usually not be collapsed with other numbers--once I have committed to a short- or long-range build range is either mandatory or irrelevant, and telling me which overall strategy is superior for a given ship/fleet is beyond the capability of a spreadsheet.

    Toward this end, I think DR had good insights in the sheet he shared on Discord--assuming you are neither capped on vents nor flux-stable with all mounts filled and care about sustained DPS, you can trade between OP and flux costs by investing OP saved on a cheaper weapon in vents and calculate efficiency as DPS / (OP + flux/s / 10)--although DPS is itself highly dimensional due to armor reduction/hitstrength interactions. (I have, though, come over to thinking that mount size should not be counted against efficiency--using an oversized weapon is impossible, not merely inefficient, and using an undersized weapon does not return any bonus.)

    Limited dimensional reduction can thus yield highly-generalizable results, but it is still dependent on human intuition to work through the remaining dimensions. Still, whatever dimensional reduction can be achieved with generalizable assumptions makes that intuitive judgment more reliable.
  • Situational assumptions. As noted before, large dimensional reductions require strong assumptions, which rarely generalize. But sometimes you do not want to generalize--sometimes you just want to know what weapon will kill a wing of Talons the fastest. Rather than outright collapsing dimensions this is a dimensional transformation, translating from the dimensions of weapon stats to the dimensions of possible situations.
Logged

Goumindong

  • Admiral
  • *****
  • Posts: 1889
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #11 on: April 19, 2018, 11:03:53 AM »

Pareto dominance is not necessarily bad. If weapons are considered strict upgrades the progression that players experience is valuable.
Logged

Unreal_One

  • Ensign
  • *
  • Posts: 44
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #12 on: April 19, 2018, 07:02:04 PM »

Pareto dominance is not necessarily bad. If weapons are considered strict upgrades the progression that players experience is valuable.

However, in the case of Starsector, you'd have to effectively be doing a rework of the weapons system for that to be the case. Even the weapons that have a strict hierarchy of one better than the other still have at least one way in which they are worse (see: the Autocannons, where the better option takes more OP)
Logged

xenoargh

  • Admiral
  • *****
  • Posts: 5078
  • naively breaking things!
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #13 on: April 19, 2018, 07:40:15 PM »

Quote
Strong assumptions. This seems to be the approach of the OP's spreadsheet--assign weights to the various dimensions and collapse dimensionality heavily. Such a number is tidy, but given the variety of situations in SS, I can think of no situation in which I would choose between, say, a heavy mauler and a HMG by comparing their respective "aggregate efficiency" numbers.
Pareto balance won't tell us anything really useful, honestly, other than, "gosh, that gun sure is OP in every way".  

We need to know why, and to know why, we need weights, because not all variables are equally important.  Guns aren't just some stats; the stats have contextual meaning in the game design.

All other things being equal, range is the most important stat, followed by efficiency of DPS/Flux (whether that's via Kinetic bonus or cheap shooting in general), followed by alpha-strike, followed by DPS, in that order.

In the end, the goal here is that each gun is not an "upgrade"; the "upgrade" is paid for with OPs.  If you pay more OPs, you get a better gun, plain and simple; OPs are the coin of this land.  This is the baseline assumption here.



OK, I've gotten the next version built, should be live at this point.  This adds in TTK again, adds a burst-alpha value (not sure if that's tuned right, will add a visible tuning variable soon).  It's getting a lot closer to reasonably accurate again.

I think it's interesting how the Devastator is just about the bottom tier no matter how we cut it.  Does anybody find that gun useful, in Vanilla?  I never used it there; in Rebal, after tweaking to the par there, it was pretty useful as a Large-scale belly-gun.
Logged
Please check out my SS projects :)
Xeno's Mod Pack

Wyvern

  • Admiral
  • *****
  • Posts: 3786
    • View Profile
Re: Vanilla Weapon Balance Sheet
« Reply #14 on: April 19, 2018, 07:49:28 PM »

I think it's interesting how the Devastator is just about the bottom tier no matter how we cut it.  Does anybody find that gun useful, in Vanilla?  I never used it there; in Rebal, after tweaking to the par there, it was pretty useful as a Large-scale belly-gun.
...I'm not sure whether to laugh or to cry, here; the Devastator is currently very good and one of my go-to options for large ballistic slots.  It's especially good on SO dominator builds, but it works -well- on everything - Legion, Onslaught (usually only on two of the three large turrets, though), etc.  Does a decent job of cracking even heavy armor (especially if you get in close), is very very good at wiping out fighters, does backup missile defense...  The hellbore's better at cracking armor, and is a bit cheaper to mount, but is otherwise quite inferior.
Logged
Wyvern is 100% correct about the math.
Pages: [1] 2 3 ... 7