Effective range of firearms still too short.

[Firestorm01X2]

Issue description

Players still reporting that ranged weapons has low accuracy:
https://discourse.cataclysmdda.org/t/firearm-accuracy/14675/17

Current situation

Short range shooting accessible but as long as we start talking about range further that 15 tiles shooting become pure gamble with low chances.
It happens even with tuned sniper rifle that has maximum range on 60. Player can't utilize even half of this range despite of his skills.

Current ranged balance based on Coolthlhu design:
#21244 (comment)

Or it should be based on that design. Current ranged balance covered by ranged tests pass but surpisingly problem still there.

I expect that tests itself has problems:
#22211

Also design may also has his own problems.

Conclusion

1. Ranged wepons overall has too low accuracy.
2. Ranged wepons hardly usable on distances higher than 15 tiles despite of player character skills.
3. At the same time enemies can get hit on players outside 20 tiles.
4. Design idea of wepons ranges should be checked.
5. Skill has almost no effect on player accuracy in most cases.

[kevingranade]

Here are the parameters of the problem as I understand them:

1. The game engine imposes a maximum engagement range of 60 squares on the player.*
2. We want guns to have various ranges relative to each other** to make them act differently from each other.
3. If we give guns anything close to ranges representative of irl performance, accuracy above that of a pistol is roughly pointless, the majority of guns would have acceptable accuracy anywhere within sight range, and more accurate weapons would show little to no benefit. Basically this is a problem of saturation, nothing more accurate than some threshold will show any further benefits.

Changes we can make:

1. Use the entire range available to us, perhaps with some saturation.***
   Right now the maximum effective range for any gun is 30 squares, that doesn't leave a ton of wiggle room for other guns or characters with suboptimal stats.
   Maximum effective range for rifles can be extended out to full sight range (~85 squares, considering the diagonal in trig distance mode, or maybe just call it 90). Other weapons can have their ranges extended similarly. This does mean that probaby every serious bolt-action rifle is going to be able to shoot anything the player can see, with no significant differentiation between them, and assault rifles are going to be in nearly the same boat, but so be it.
   This also stops very far short of "realistic" ranges, but maybe it will be good enough to be able to hit what feels like the right proportion of the map with various guns.
   A caveat to this is that at best, monsters with ranged attacks can equal the player's effective range. This isn't a departure from the norm, the player is almost always granted an edge in sight range anyway.
   What it also means is that engagements at extreme range will be about aim speed and toughness, not outranging your opponent.
2. Allow distributions of gun ranges to overlap.
   The effective ranges outlined by Coolthulu had no overlap, the intent of this was to keep the weapon groups as distinct from each other as possible, but this had the side effect of compressing the real effective ranges (characters with less than max stats and skills, and suboptimal guns) to be extremely short. If we relax this kind of sharp seperation and allow the gun types to significantly overlap, it will allow for significantly increasing accuracy at the low end in particular. This has the side effect of making guns less distinct from each other, but will gain us a better minimum usable range for most all guns (though things like unrifled zip guns are still going to be really absurdly terrible).

What actually needs to happen to make this go:

• Create a new set of tables for range constraints based on these guidelines (or whatever guidelines they get shifted to).
• Update unit tests to match new tables.
• Coarsely adjust the variables established in [READY] Ranged balance unit test #21468 to make the new test parameters pass.
• (Later issue or PR) Place all the firearms on a relative scale based on IRL stats when possible.
• (Later PR) Tweak individual gun accuracy values to place them in their correct places in the scale.

* I have some ideas for adressing this, but they're long term. Also nothing I can imagine doing is going to allow the approximately 1,000 squares of range necessary to make gun ranges "act like real life", so we need some kind of compromise even if we can increase max engagement range. Hopefully this compromise can simply be scaled up as new features are implemented without reinventing the wheel every time.
** The de-facto way to assign relative ranges is to base them on the real guns they are based on, both in a coarse pistol < shotgun < SMG < assault rifle < rifle way, and in a "X gun is significantly more accurate than Y gun IRL, so it should also be so in the game" way. #22756 is attempting to enumerate relative ranges.
*** This isn't going to be a popular statement, but gun accuracy is working as intended right now.
Coolthulu specified that a "god character wielding a perfect rifle" has a 50%+ chance of a good hit at 30 squares, and that is what the tests establish. Spawn a character, give them a BLR, ammo and a rifle scope, give them "godlike" stats and skills (20/10), if they use precise aim, they will hit a medium target at 30 squares distance at a rate well above 50%. This is what Coolthulu suggested, and no one put forth any other suggestions, so that's what I went with. In fact, I weakened several of his constraints by not limiting accuracy as much as he asked for. Even after the rebalance was completed, one of his main complaints with it was that long range sniping is too good. "endgame" characters can easily reliably snipe much farther than they can see

[AMurkin]

What about learning accuracy curve?

[kevingranade]

Which one? Label your axes please or I won't know what you're talking about >_<
https://flowingdata.com/2012/06/07/always-label-your-axes/

[AMurkin]

Huh. Just want to say that the maximum achievable level is defined, but how is the result determined under the worst conditions? Will it be rapid growth at the beginning and a slow improvement in the end? Or vice versa? I know that this is a function of many variables, but somehow one has to come to a consensus (if it possible).

[Firestorm01X2]

Create a new set of tables for range constraints based on these guidelines (or whatever guidelines they get shifted to).

While looking on #22756
I recommend to utilize max weapon range in range tables. And do not target exclusively to effective range.
I mean, if Cataclysm is a game and guns are not suppose dto be as IRL then lets drop this 50% of good hit.
Lets use something more relible percentages. Like 70% of good hit.

It is better to overbuff guns and then cover the edges than underbuff it again.

To compensate that I recommend to icnrease aiming time and effect of skill on aming time. So even underskilled player could hit but he will do it very slowly that additionaly may force him to shoot without proper aiming in critical situation.

[kevingranade]

@AMurkin Have you read #21244 and #21468? Those outline a lot of the scaling as well, a lot of that needs to be adjusted, but those establish the general parameters we're working with. The existing tests cover the expected progression of a character, with improving skills and stats as well as higher quality weapons, but we also need the edge case tests like "highly skilled character with terrible weapon", and "beginner character with excellent weapons" to round things out.

[aeoo]

I think a character with 20 perception is kind of unrealistic and/or broken. It would require destroying all other status just for perception. I don't think the game should be balanced around a per 20 character at the upper end. (Normally I agree with virtually anything @Coolthulhu says, but I think I have to take an exception here just this once.)

Also right now the scopes are absurdly rare.

The situation right now is that there is virtually no effective range difference between any firearms. That, to me, is absurd. Also as bad as the iron sights are compared to glass, the iron sights were still used for sniping by Simo during WW2:

https://www.simohayha.com/

So sniping is not a matter of optics, but a matter of skill. Simo makes this obvious.

I think the way the game is right now, it's all kinds of broken. Right now I cannot tell what the "marksmanship" skill even does, other than unlocking a few recipies.

Yes, guns should be differentiated much more inside the game. Iron sights with high skill should allow for relatively long range sniping without first having to create a broken per 20 character.

Of course sniping with optics should be even more effective, but sniping the Simo way should still be possible.

Just my 2c.

[Firestorm01X2]

need the edge case tests like "highly skilled character with terrible weapon", and "beginner character with excellent weapons" to round things out.

Or at least character with default stats (8/8/8/8). Soldier does not need to be superhuman mutant to fire efficiently.

And probalby it is time to finally add this issue to 0.D milestones.

[Firestorm01X2]

I have an idea. Looks like proper solution of current issue will take at huge amount of time( I expect at least 6 months). So let's merge this: #22731
As temporary solution. After proper solution mod may be marked as obsolated mod.

[kevingranade]

You already have feedback for #22731
#22731 (comment)
Repeating yourself without changing anything does not help your argument.

[Firestorm01X2]

Why is this still not in 0.D milestones but at the same time #20414 there? Both are non minor balance problems.

[aeoo]

Well, I finally was able to find both an A7 rifle and a rifle scope and my skills were 8 for marksmanship and 5 for rifles, and yet with all that, I still cannot hit the broad side of a barn. I also have a targeting bionic! Accurized rifle. The whole shabang.

The effective range of firearms even with the best conditions is idiotically short. What is the point of bullets or energy bursts going to 40 or 60 range if the max effective range is something like 9 squares? It's insanely bad. Horrible. Broken.

There is a PR that extends the ranges of all the firearms and it has a spreadsheet attached to it and everything. And even kevin "himself" (barf) has spoken positively about it. And yet it still isn't getting merged. Kevin, sorry bud, but the way you're going, no one will want to contribute anything. All the useless junk gets merged but a lot of great PRs are just sitting there because kevin has issues.

Kevin will talk about realism and then firearms are accurate only to 9 squares and suddenly "what realism?" It's so idiotic. Such bullshit. All this talk about "realism" is pure nonsense. Why talk about it so much? Just admit realism isn't important. Cause if it were, no way would effective ranges ever get that short to begin with.

Also that thing about skills in real life having virtually no influence on accuracy with a firearm is beyond idiotic. There is like a day and night difference between high skill and low skills when it comes to marksmanship, no matter what. Anything else is not real life. In real life skills actually do matter and not just equipment.

[Zireael07]

@aeoo: Which PR are you talking about?

[aeoo]

@Zireael07 #22756

[Zireael07]

From what I can see, the holdup is a merge conflict the PR author is dismissing as 'dates got changed in descriptions'. Doesn't change the fact he should fix it.

[aeoo]

I've been watching that PR and I've seen its author or maybe someone else resolve those conflicts many times. So many times it was ready to get merged without conflicts. But it was ignored. As it continued being ignored, conflicts arose. They got fixed but still the PR was just sitting there until more conflicts arose. I think the original author got tired of this nonsense and said "screw it." What's the point of fixing conflicts if it will just sit there for a month until inevitably more conflicts will creep up?

[aeoo]

You want to hear something really insane? Bows now are more accurate than the rifles. :) I mean a bow with all the easily available mods outdoing a rifle in both accuracy and damage, and it's quiet, easy to use, infinite ammo. I just want to punch my own face. The bows are fine, btw. But the rifles are simply broken.

I mean, if from a game design standpoint it's not desirable to dispatch zombies from 60 squares, fine, but please come up with a different game mechanic and reason other than "you just cannot be accurate to 3 house lengths." I mean, seriously? Effective ranges of most modern carbines are something like 300-500 meters right? Do any of you know how long of a distance that is? A football field is roughly 100 meters long. A yard is a bit shorter than a meter. AR-15 is rated to 500 meters effective range (~600 yards). Get real people. Or visit a rifle range one of these days. 100 meters is nothing for a rifle. Even a garbage rifle with a less than world class skill can hit at that range.

Basically bullets easily and often hit zombies, period. Maybe say they hit, but don't do much damage. Maybe only critical hits which take the head clean off actually put down zeds. If realism matters, that's how it would be resolved. So if you don't want people to finish off all the zombies before they can become dangerous, then find some other reason than "you just can't him them." Cause yes you can. If realism matters, all the rifles will absolutely hit all the Zs at a long range.

Maybe let Zs shoot back with their own rifles. I don't know what the problem is, but it's absurd that pistols and rifles are effective to almost the same range while the bows are better than most rifles. Just what is going on here.

[Zireael07]

The PR sitting for a month is not inevitable - big PRs sit longer, but that month was a month with most merge-rights people seemingly busy with IRL matters.

[Coolthulhu]

#22756

That PR alone wouldn't do anything to fix the problem as it doesn't have actual mechanics, just data prepared for later. It is also huge and "monolithic" (unfragmentable, "atomic") and PRs of this kind are notoriously hard to get through.
The idea isn't bad, but it's not like it's "press button - receive fixes". It's an idea more than a fix.

I mean, if from a game design standpoint it's not desirable to dispatch zombies from 60 squares, fine, but please come up with a different game mechanic and reason other than "you just cannot be accurate to 3 house lengths."

One of the huge problems with the accuracy is the way errors are summed. Not as (angle) vectors, but as sum of errors, most of which are uniformly distributed.
Few months ago I wasn't able to convince Kevin to switch to Gaussian distribution, which not only solves the problem of small errors mattering way too much, but also gives distribution more like ones in real life. And is much more well-behaved mathematically, which would help UI and AI.
Maybe real life data could help?
http://www.the-long-family.com/group_size_analysis.htm

[aeoo]

@Zireael07

The PR sitting for a month is not inevitable - big PRs sit longer, but that month was a month with most merge-rights people seemingly busy with IRL matters.

That's understandable. I think one way to approach this is to schedule a time for when the conflicts should be fixed. It should be convenient for both parties, right? Otherwise a PR submitter has to be fixing conflicts all the time while only hoping for a merge without any guarantees?

I get that people are busy, but then it makes no sense to demand conflict resolution if you cannot promise a merge, right? Scheduling a PR submitter and someone who could merge it to meet in IRC might work. That way they'd fix the conflicts once and be assured of a merge.

[aeoo]

@Coolthulhu If I understand you correctly, the various sources of errors should not all be additive, because in many cases the errors from the various sources should wash out much of the time. So if the scope gives me a "veer left" error, and the barrel is giving me a "veer right" error, they wash out in this case, instead of being added together. It's only in the worst case scenario, which should be rare, that all the error sources err in the same exact direction.

If so, then it makes sense.

[Firestorm01X2]

So plan B means actually huge range for everything. At least for now. That is I haven't understood right.
In this case plan A is better because B will get proper balance in very very distant future. However a lot of players may like more "realistic" approach.

It is definitely needs more opinions.

Also time to balance in plan A may be decreased after putting balance adjustments in json and properly preaparing whole infrastructure to quick rebalancing. In best case scenario via editing json files only.

About jsonizing: #23226 . It will allow to load hidden options from separete json directly with various types (not bool only).

Also important note. Balance itself is huge task but putting dispersion adjustments in json can be completed gradually. And it will not break tests if completed correctly. In this case it may be first task of rebalancing.

[Firestorm01X2]

Ok. Another approach.
Let's pretend that

*** This isn't going to be a popular statement, but gun accuracy is working as intended right now.

true. It almost may be. Scaling between [0,30] tiles may be ok with some tweaking (like more relation on character skills).

Another side of that are shooting enemies. They also balanced on maximum range 30. Different approach is to to tweak enemy stats to match current ranged balance for player closely. Maybe tweaking stats of fake_npc created on shooting or tweak enemy weapon ranges?

Problem that if we are going either A or B it ca lead us to same situation - enemies will have much more more ablillities to hit players while players need excelent weapon and stats to do that with enemies. But it will be on ranges [0,60] instead of [0,30].

As long as enemies will use similar outlines - there are will be much less complains. With that #22756 can help.

[Coolthulhu]

With plan A, all guns stay in their designated niches at all times (pistols very short range, shotguns and SMGs at intermediate range, rifles at long range), but the scale of those ranges shift over time as new range-extending features are added.

The problem is, at the moment there is no intermediate range niche.
There are guns that shoot accurately (rifles), guns that aim fast (pistols) and guns that do neither of those things (SMGs and shotguns).

The problem with ranged combat is not lack of ability to attack things outside reality bubble, it's the fact that any tiny error absolutely ruins the ability to shoot accurately. Not maxed out marksmanship, lack of god tier scope, not playing on easy mode (single pool+lab/VBD), any gloves etc.
This is unfixable with sum of (absolute) errors model of dispersion. Unless there was some non-linear stepdown function applied at the end.

[Firestorm01X2]

The problem is, at the moment there is no intermediate range niche.

Completely agree on that.

Not maxed out marksmanship, lack of god tier scope, not playing on easy mode (single pool+lab/VBD), any gloves etc.

Exactly. Also enemies given that good stats by default. Monsters uses to shoot fake npc with 8 marksmanship and 8 weapon skill. Player most of time the lacks of that. Also that fake npc has above average stats.

This is unfixable with sum of (absolute) errors model of dispersion. Unless there was some non-linear stepdown function applied at the end.

If it helps to even things up - this should be tried firts before any kind of A or B. Let't not reinvent the wheel that may lead to same trap.

Especially if it is simpler than moving thresholds on different ranges.

[aeoo]

@Coolthulhu

There are guns that shoot accurately (rifles), guns that aim fast (pistols) and guns that do neither of those things (SMGs and shotguns).

Based on my experience, even if you eliminate every possible source of optional error, there is still that mandatory dispersion minimum on a barrel, 45, and that's still huge. As a result, to my own sensibility the rifles are not in fact accurate. Even the most accurate rifle is not accurate enough. Even if I could get the scope side to 0, with a barrel at 45 I still feel like I am spraying way too much.

On the other hand, the dispersion levels on handguns don't matter as much at that range, so the irony here is that handguns are somewhat accurate in a relative sense considering the range of engagement.

The way guns work right now, I am strongly incentivized by the game engine to let the zombie get as close as possible before shooting, even if I am using a rifle, because I want to get one or two headshots in more than anything. It wastes fewer bullets and it kills faster. Long range engagement is very discouraged right now by the engine. The optimal way to use a rifle in-game is not how it is in real life. -1 for realism.

I still very much agree with the rest of what you'd said, namely about how the sources of error are added up together, and how this makes the early game shooting much harder than it needs to be.

---

(optional reading below)

The big thing that's missing in the game is the concept of cover. I think Kevin is concerned that the baddies will have too much of an advantage against the player if the range of engagement were to be increased and I think that's because there is essentially no cover, no camouflage, and no sneaking in the game. Right now the only way to defend is to wear thick armor (or to drive an armored vehicle, which isn't always possible). Shooting from behind a corner is not an option as far as I know. There is a "peek" command which only allows looking.

If two game entities engage at long rifle ranges, cover or camouflage should become vital if realism is important. In real life what will determine who lives and who dies is not barrels or scopes like in the game, but skills, the element of surprise (who shoots first), and cover plus camouflage.

In general I am not concerned about being shot from a far distance. I turn my NPCs down to 0.2 spawn rate, because I don't like too many NPCs running around, and even hostile NPCs do not preemptively shoot at me, but rather, they first try to get close, make some threats, and then they may start shooting (and annoyingly, stealing my stuff too).

The only things which do shoot without warning are turrets and other automated gun systems. These things are already silly brutal. Even a lame Uzi turret is something that demands respect: either a very thick armor or cloaking with a lot of bionic energy to get close enough to throw a scrambler grenade at it. I don't know of any cheap way to get rid of it. Maybe smoke + mask + scrambler? I haven't tried it yet. Even with the heavy survivor armor I sometimes get scary 10-20% HP shaved off if I try to just "hang out" inside the firing range without cloaking.

So in a lab the engagement range is very short for the turrets, and increasing the overall gun engagement range will not make those turrets more dangerous. It will only affect the turrets above ground, and if that's a concern, I'd rather above-ground turrets get removed from the game anyway. If the only thing that's blocking a more realistic rifle engagement range is the above-ground turret, I don't want to let the above-ground turret win. I don't think above-ground turrets add enough fun to the game to make them "must keep no matter what."

Also, turrets are pretty OP when they're taken over. I always grab 2-4 scrambled turrets, and they provide infinite light for crafting and amazing automated defence for my car, so that no fool even thinks of getting close to it. All that turret evilness gets completely turned around once the player commandeers one.

[darwithe]

To me the current problem is attempting to add realism, by way of the current aim dispersal curves, onto arbitrary game limitations, the maximum ranges. One or the other has got to go, and Kevin's plan A and plan B are largely attempts to fix the problem by dealing with the arbitrary side of things. In my opinion both plans are a bandage, likely a good bandage and possibly even a good enough bandage but they retain problems.

Plan B basically kills variance by gun, pretty much forever, other than in certain edge cases where it is gun vs gun; Kevin acknowledges this problem, but I think is underestimating it because in this setup once you get a pistol you are set against anything that isn't made for gun ranges. There isn't any feeling of "Ok I've got a pistol but what I really want is a rifle so I won't have to get so close to that horde of terrifying shocker zombies".

Plan A is actually better in that it doesn't make anything worse. It is basically just "The same as things are now but with more breathing room." But I'm not sure how much it will fix things before bubble extensions, and I don't think it will fix the feeling that the game is lying to me about a gun's range.

If I were stuck between those two choices I'd go for Plan A, but with the preamble out of the way my suggestion would be to go more gamey not less. Most players don't have a problem with arbitrary range limits, with the exception of the "But muh realism" set that probably won't be satisfied regardless. The big complaint I have is that even within that short range I never feel that the gun was useful compared to the risk. So take Coolthulhu's well reasoned well derived accuracy limits and toss them out the window for something more arbitrary to match the arbitrary ranges.

Something like an 8's an 0 character hitting 50% of the time at maximum range, while the godlike 20s an 10 character pretty much always hits(He's godlike yah?) This may be over-correcting but balance issues arising from this can be better fixed by time penalties and rarity and the like, rather than the current situation where you tell a player he can use a weapon at 10 but can actually only hit something at 3.

And yes this suggestion could easily be combo'd with Plan A, but frankly the interface already kinda sucks for things offscreen so I personally would prefer something that doesn't increase maximum ranges beyond what they currently are.

EDIT: I should note that I haven't touched bothered to touch guns in months as they've been too useless to bother picking up so it's possible some of the earlier attempts to balance partially address my problems.

[Firestorm01X2]

And yes this suggestion could easily be combo'd with Plan A, but frankly the interface already kinda sucks for things offscreen so I personally would prefer something that doesn't increase maximum ranges beyond what they currently are.

Make sense. Current balancing between ranges [0,30] and considering ranges more than 30- very long range are not bad actually. Enemies already using maximum range 30 (sniper turret) as far as I know. It is Ok.

In this case using that outline may be good: #22756
because it is balanced around distance [0,30]
But we must make sure that average player can hit something on that ranges reliable. Average mean not endgame mutant.

[Firestorm01X2]

Let'se try that kind of outlines:
m14ebr has max range of 23 based on that PR: #22756
Let's take tweaked m14ebr.
So player with skills Marksmanship 3, Rifles 3 should be able to hit target on distance 23 with at least 30 %
Marksmanship 6, Rifles 6 distance 23 chance to hit at least 60%
Marksmanship 9, Rifles 9 distance 23 chance to hit at least 90%
Marksmanship 10, Rifles 10 distance 23 chance to hit around 99%.

Also looks like Kevin do not fully understand problem. Both plans A and B is about increasing engagment ranges. Problem is not in the ranges itself. Problem that on current engagmnent range around that balance was built (0-30) real effective ranges for player are too short.

So balance may be fixed on current engagment ranges without moving them first. What the point of increasing ranges to 60 if old problems just will be scaled to ranges [0,60] instead of [0,30]? Ok player will be able to somehow shoot at 20-25 tiles, but the same time enemies will shoot him from 30-60 tiles and so what?

[kevingranade]

because in this setup once you get a pistol you are set against anything that isn't made for gun ranges. There isn't any feeling of "Ok I've got a pistol but what I really want is a rifle so I won't have to get so close to that horde of terrifying shocker zombies".

This also assumes you're sitting at max skills etc. I feel like this is a big part of the problem with the previous balance numbers, it refused to push maximum effective gun ranges to the edge of the map, therefore everything else had to fit within that ultimate range, and in particular if you had a suboptimal gear, skills, stats, etc, you couldn't hit anything with a pistol. Allowing "ideal" pistol users to hit anything on the map doesn't particularly break the game, but we're getting a lot of feedback that not allowing average shooters to hit at much more than arm's length does.
In other words, as long as you don't have ideal stats and skills, you DO still want that rifle to hit things before they get too close, especially targets that take several shots to take down.

the feeling that the game is lying to me about a gun's range.

Can you expand on this? What is the game saying that makes you think guns are more effective than they actually are?

[darwithe]

Can you expand on this? What is the game saying that makes you think guns are more effective than they actually are?

This seems to have been fixed since I last bothered to even look at a gun by giving loads more info. So great job there, mind the ?:m help documentation is still overplaying guns compared to their actual effectiveness.

In other words, as long as you don't have ideal stats and skills, you DO still want that rifle to hit things before they get too close, especially targets that take several shots to take down.

Very much this. That's why I suggested throwing out the previous accuracy balance and replacing it with one that is dramatically more in favor of the novice but loading down the novice with time penalties so that it becomes a simple distinction: A novice should be able to reliably kill a single simple zombie with a gun if it doesn't get the drop on him. A competent shooter taking advantage of terrain should be able to handle small hordes if they have enough ammo, while someone near the peak of human ability should find handling a small horde trivial. This is basically the way melee skills are currently handled, and guns have the added drawbacks of ammo and noise.
Or TL;DR: First few levels of gun skills should be about decreasing time penalties while last few should be about increasing head-shots.

[Coolthulhu]

Allowing "ideal" pistol users to hit anything on the map doesn't particularly break the game, but we're getting a lot of feedback that not allowing average shooters to hit at much more than arm's length does.

True, but the problem is that dependency on equipment, stats, and skills scales pretty horribly.
For many guns, going from "terrible" to "very good" increases the effective range as much as going from "very good" to "perfect".
Near the "perfect", every tiny aspect has huge effect. Wearing boxing gloves at 0 skill isn't very noticeable, wearing thin gloves and lightly shaded glasses when godly will noticeably impair effective range, but also headshot rate at close range.

The "godly" shooters from my examples weren't supposed to be the cap, something that you can reach. Rather, they were there only to make it less likely that the formulas blow up as described above.
If the ranged math was rebalanced to be usable at low levels without care for high levels, middle level balance would most likely suffer too. I guarantee that it would happen if they were balanced just to pass the (not very broad) tests, rather than with great concern for things that are NOT covered by the tests.

[kevingranade]

The "godly" shooters from my examples weren't supposed to be the cap, something that you can reach.

So the "unreachable" character loadout had the following range goals:

Character	>50% good	<10% good
God pistol	10	15
God SMG	20	30
God rifle	30	60

This is the problem. It's not using the available range, instead reserving the outer half of the map as visible but too distant to hit reliably for any possible character. The problem is then compounded by limiting "best SMG ever" to 20 yards and "Best pistol ever" to 10 yards. I'm not blaming you for this, I'm on the hook for thinking this was a reasonable starting point as well, but it's totally unproductive to throw around a bunch of theoretical issues when there's a HUGE GLARING issue of "the best sniper ever can't hit reliably at 60 yards".

Under plan A, the ">50% good" range for rifles would be moved to ~85 squares.
Under plan B, the ">50% good" range for rifles would be moved to ~1,000 squares.

In more detail, plan A

Character	>50% good	<10% good
God pistol	40	off map
God SMG	60	off map
God rifle	90	off map

And plan B

Character	>50% good	<10% good
God pistol	off map	off map
God SMG	off map	off map
God rifle	off map	off map

i.e. with plan B we stop caring about god characters, they might have trouble hitting at the edge of the map with crappy pistols guns, which we probably want to quantify, but otherwise they just hit all the time as long as they bother to aim.

I guarantee that it would happen if they were balanced just to pass the (not very broad) tests, rather than with great concern for things that are NOT covered by the tests.

If you have other constraints in mind, list them, I'm not interested in hand-waving about it being "hard" or "impossible" to balance if you aren't going to outline the scenarios where the problems occur.

These tests are completely agnostic of the dispersion system by design, if you can provide sensible tests that the existing system can't pass, that provides evidence that there's a problem, but no amount of repetition is going to make the case for you.

[Coolthulhu]

This is the problem. It's not using the available range, instead reserving the outer half of the map as visible but too distant to hit reliably for any possible character.

Intended. I wanted a system that keeps the guns different in 60 tile reality bubble system.
Without that, we're about to get a system where guns have ranges that are different, but only in ways that do not matter.
No one cares about the difference between sniping things from 2 overmaps and sniping them from the other end of the bubble - it's sniping from safety either way. Meanwhile the difference between sniping shockers before they zap or having to deal with the zap IS meaningful.

HUGE GLARING issue of "the best sniper ever can't hit reliably at 60 yards"

Not yards, unless you want to argue that no human can see farther than 60 yards.
A tile can easily be 29 yards.
The issue exists, but don't misrepresent it with tile = real life const, it will not help it the slightest bit.

If you have other constraints in mind, list them, I'm not interested in hand-waving about it being "hard" or "impossible" to balance if you aren't going to outline the scenarios where the problems occur.

At the moment most problems with sniping are stopped by the same thing that is the main cause of this very issue: the 45 dispersion low cap.
If you want to get rid of that, we're back to the problem that already existed before it:
At high levels, expected value of dispersion gets close to 0, where it just plain breaks. Because trigonometry.

range ~= 1/tg(to_rad(dispersion/2))

Then there is the horrible linear addition of errors, which greatly favors going from say, 30+30 dispersion to 60 dispersion.
The errors are added as in:
http://anydice.com/program/f4ff (not sure if anydice remembers well. In case it doesn't: output 3d100 > 30)
As opposed to a less "nitpicky" linear addition, which could look like:
http://anydice.com/program/f500 (output 1d300+2 > 30)
Or a magnitude of sum of 2D vectors, which I can't easily present here, but which (for vectors of comparable magnitude) has expected value somewhat close to square root of the sum of magnitudes.
Or a sum of normal distributions, which is pretty much perfect in every way.

This will not be shown on the UI, because of simplified math it uses (another problem with the whole system).

[kevingranade]

Intended. I wanted a system that keeps the guns different in 60 tile reality bubble system.

If you think the current range limits are reasonable, there's nothing to discuss.

No one cares about the difference between sniping things from 2 overmaps and sniping them from the other end of the bubble - it's sniping from safety either way.

If you read the many posts here, on the forums, on reddit and discord, you'll see that yes, people do care. Game balance is not the only consideration, verisimilitude is a major issue, and the current range limits are breaking the hell out of it.

[terribleperson]

There's a post on the front page of the subreddit right now about ranged weapons not feeling worth using. It's not the only one I've seen. I'm not going to try to suggest a system, but I don't agree with the premise the current system is built around.

I much prefer Plan B. I don't think the realistic mechanical MOA accuracy of various firearms unmodified by skill, ergonomics, and physiology would be much of a concern to players, even if we could massively expand the reality bubble. Sure, it's important when you're at war against other people using firearms in a large open space, but that's not a big part of Cataclysm. I don't see why it's important to clearly differentiate weapon classes in terms of accuracy and aiming speed. Coolthulhu's original balance breakdown entirely discounts volume, weight, and draw time as advantages, and I disagree. Those were always important considerations for me in my games. That sort of simplified balance is common in multiplayer FPS games, but this is Cataclysm. The ability to simulate more aspects and in greater detail is the great advantage of the top-down, turn-based format. If we find that realistic firearm accuracy results in weapons seeming too similar, we can ask what makes them different in real life and seek to simulate that.

[Coolthulhu]

If you think the current range limits are reasonable, there's nothing to discuss.

Current effective ranges are bad, but not because they don't allow sniping things on the other end of the bubble. They're bad because they don't allow sniping at all until you become a cyborg master with a scope worth more than its weight in meth.
They're also bad because they grow very slowly early on, then suddenly explode by the end.

That scaling behavior is not my fault.
When I stated my numbers, no one complained about them. In fact, they seemed to be really well received.

It seems that the problem isn't with my numbers, but rather:

• Changing the breakpoints. In my comment, novices used 90% aim, which would noticeably change their ranges compared to current ones.
• Changes to stat balancing making it harder on actual characters (stat spread)
• Scaling of the range being vulnerable to trigonometry (slow growth early on)
• My descriptions of good/bad equipment being interpreted very liberally, especially the parts about gun mods being comparable to ones a given character could find
• Other, more minor shortcuts

verisimilitude is a major issue, and the current range limits are breaking the hell out of it

Does not follow. Current ranges aren't reality bubble sized, but smaller than one overmap tile (ie. <24 tiles).
Unless people are complaining specifically that they can't shoot things 40+ tiles away, my point stands.

[Coolthulhu]

@aeoo

I don't think the game should be balanced around a per 20 character at the upper end. (Normally I agree with virtually anything @Coolthulhu says, but I think I have to take an exception here just this once.)

I didn't say that 20 perception should be used as a balancing point. In fact, in my "godly character" description I put "only" 14. The only thing about 20 perception that I wanted there was that there was some reward for going over 14 rather than strict cap making any points above 14 useless for ranged combat.
So, depending on what you mean by "game should be balanced around a per 20 character", you may actually be agreeing with me there.

So sniping is not a matter of optics, but a matter of skill. Simo makes this obvious.

Fun fact: before the last change, high perception allowed high perception characters to forego using a scope and go all Simo on zombies. I wanted to keep that.
Nowadays you can only get a penalty for low perception or bad optics and even a 20 perception character benefits from a scope (at all ranges).

[Firestorm01X2]

@kevingranade
I understand why you are closing this: #23343
But at the same time it is very very bad, bacause contributors may interpetate it as "Don't mess with gun balance. It is forbidden".

This way you are basically locking this issue on yourself. Or in fact you already did it.
Now no one may even attempt to do something.

[Firestorm01X2]

@kevingranade
Anyway. Issue exist for two months but the problem itself almost for 8 months. I guess it is time to at least decide, what to do.

Plan A, B. Or Coolthulhu's ideas (let's call it Plan C).
You have to decide this bacause problem are basically "locked up" until your decision.

Personally I am for easier approach. If Coolthulhu know how to correct current range balance without reinventing it- let him do so. It will be some progress at least.

So, waiting for your decision.
A, B or C?

[pipehat]

For what it's worth, lost a game the other day because I couldn't drop a shocker with 8 point-blank shots with a remington. I don't care much about realistic dispersion. I do care about it being satisfying to shotty a zombie in a zombie apocalypse game. I recommend ratcheting down the 'realism' concerns and make guns fun and useful - even if imperfect.

[Firestorm01X2]

For what it's worth, lost a game the other day because I couldn't drop a shocker with 8 point-blank shots with a remington. I don't care much about realistic dispersion. I do care about it being satisfying to shotty a zombie in a zombie apocalypse game. I recommend ratcheting down the 'realism' concerns and make guns fun and useful - even if imperfect.

It is strange. Shotguns usally not that bad.

[smolbird]

I recommend ratcheting down the 'realism' concerns and make guns fun and useful - even if imperfect.

Shotguns not being trash at point-blank range would also be realistic though.

[Firestorm01X2]

Problem looks more and more abandoned. There are no attempts to fix it. I've even seen mentions that is will be dealt after 0.D. At this rate it may take years easily.

At the same time I've already provide solution:
https://discourse.cataclysmdda.org/t/experimental-builds-with-improved-gun-accuracy-updated-7-05-2018/15131

Download link:
https://github.com/Firestorm01X2/Cataclysm-DDA/releases

Video showcase:
https://www.youtube.com/watch?v=yvG5B2UO2QE

It solves issue even without reworking balance from scratch.

Changes:
master...Firestorm01X2:firestorm

I have around 400 downloads:
https://www.somsubhra.com/github-release-stats/?username=Firestorm01X2&repository=Cataclysm-DDA

And overall positive feedback (mostly on discord). Sadly not so much people post constructive feedback.

Maybe it is time to finally allow it to exist in mainline as mod? Because I don't see any other solutions around here.

[smolbird]

What exactly is the issue with not just implementing that in mainline? Are there any edges cases or oddball math preventing that, or has no one just bothered to examine it yet?

[Firestorm01X2]

Worth noting that it: #22756
was closed by Kevin also.

What exactly is the issue with not just implementing that in mainline? Are there any edges cases or oddball math preventing that, or has no one just bothered to examine it yet?

I can create PR with mod, with fully JSONized effect of reducing dipsersion. It is not problem at all.
Maybe it is not perfect for mainline but it is good as mod. After Kevin's rebalance mod can be marked as "obsolated" or even put to mainline "oficially" with changed unit tests eventially. Everything can be done safe and steady with no harm at all. Mod can be done for evening.

But are there guarantees that it would not be closed?

[Night-Pryanik]

Should be closed after #23720?