Community The New World Design Poll #2: RNG

[Rinslin Merwind]

One way of approaching it would be to make every attack into multiple smaller attacks. You attack once with 70% thc, character strikes three times each with a 70% chance. You're gonna stay much closer to the average. Of course, this changes the balance quite significantly in cases of flat damage bonuses/reductions.

That's ultimately what it comes down to though, the importance of any single hit/miss. The more important it is, the more annoying a bad roll gets. If hitting once wins you the match, then your to-hit chance == your chance to win. And vice versa with taking hits and losing. On the other hand, if you just lower damage(or just increase hp) so much that it averages out anyway, combat becomes long and boring.

Except it would be weird for non automatic weapons.

 

[Raapys]

Nah, it would just turn the attacks into combo attacks. You already have it to some degree in AoD, e.g. Flurry.

 

[Tigranes]

If I miss, I miss, and if the system is transparent and consistent, then I know that any rage is my own failing...

But it's not your failing because you aren't the one rolling the dice. If you missed because your skill is low and/or your character build is weak, that's on you. If you did everything 'right', raised your THC to 80%, then missed 3-4 times in a row, that's not your own failing.

I do see the point about different ways to design the said consistent and fair system. I don't think it should be driven by "how many times in a row can I miss", that's just how probability rolls, and if I found out that every first hit, or every hit after 3 misses, is guaranteed to hit, that would set up some degenerate metagaming as well. I think it suffices to, say, get a 70% from rolling 10 times, not 100 or 1000.

First, if your skill is low, you can miss as much as you want and then some. We're talking about adjusting streaks only for high THC rolls. Second, I don't see how you can metagame it. Let's say after 3 misses you know that you'll hit. If your THC is 70% or higher, you'd better fucking hit something after 3 misses. How are you going to exploit? You can't attack air. While you're missing left and right, you're taking damage. This 4th hit isn't some freebie to be exploited but a long overdue hit not because you're special but because your THC is high enough.

I mean that raging "how did I miss 3 times in a row at 70%, the game is rigged" is my own failing. I think there's nothing wrong with missing 3 times in a row at 70%, and that the rare occasions in which that happens can serve to create great moments.

What I mean is that I don't want to play TNW, miss 3 times, then think, "OK, now I'll hit it for sure." That metaknowledge is an exploit. However, this isn't such a big deal because only random Codexers like me will even know about it.

The big point for me is to have a system that isn't rearranging things behind the scenes based on what just happened or what the player expects, but to have a system that is already in place & does its thing.

For example, would this work the other way round? Let's say I have a 10% chance to hit, and hit twice in a row. Will the game then adjust the pool of possibilities that the next hit will definitively miss, using up one of the nine 'misses' in that bag?

 

[IHaveHugeNick]

But that's just the thing. Understanding probability doesn't make it any less rage inducing when the dice rolls against you. Shooting mechanics are abstractions and as such they need to remain believable. Missing a point-blank shot with a shotgun just ruins your suspension of disbelief, regardless if you understand the math under the hood or not.

Ie, people still play roulette and the slot machines, even though it's public knowledge that both are rigged and house always wins in the long term. It's on the house (or in this case, the devs) to come up with a way to rig the dice that will be entertaining.

 

[trais]

Fudging RNG and lying to the player via tooltips is bad. Really bad.
If you want to cater to crybabies, do it some other way, e.g. when some sufficient level of bad luck in combat is reached, give them a free re-roll.

Or better, some extra XP or SP for being capable of handling a streak of bad luck like a champ.

 

[FeelTheRads]

If I miss, I miss, and if the system is transparent and consistent, then I know that any rage is my own failing...

But it's not your failing because you aren't the one rolling the dice. If you missed because your skill is low and/or your character build is weak, that's on you. If you did everything 'right', raised your THC to 80%, then missed 3-4 times in a row, that's not your own failing.

I do see the point about different ways to design the said consistent and fair system. I don't think it should be driven by "how many times in a row can I miss", that's just how probability rolls, and if I found out that every first hit, or every hit after 3 misses, is guaranteed to hit, that would set up some degenerate metagaming as well. I think it suffices to, say, get a 70% from rolling 10 times, not 100 or 1000.

First, if your skill is low, you can miss as much as you want and then some. We're talking about adjusting streaks only for high THC rolls. Second, I don't see how you can metagame it. Let's say after 3 misses you know that you'll hit. If your THC is 70% or higher, you'd better fucking hit something after 3 misses. How are you going to exploit? You can't attack air. While you're missing left and right, you're taking damage. This 4th hit isn't some freebie to be exploited but a long overdue hit not because you're special but because your THC is high enough.

What a load of shit.

Can never actually state the real reasons, can you?
It's always "loooooooooooololol rpgs were wrong always we'll be ones to fix them with this one easy trick" with you "developers".

Sure thing, bro. Nobody ever figured out these easy tricks and thought of implementing them, thank fuck for this new breed of "developers".

You're doing it because of newfags and retards who can't stand missing and kids who go "loool it looks stupid, how can i miss if i'm standing right in front of him looalalalool". Period.

 

[Rinslin Merwind]

I mean that raging "how did I miss 3 times in a row at 70%, the game is rigged" is my own failing. I think there's nothing wrong with missing 3 times in a row at 70%, and that the rare occasions in which that happens can serve to create great moments.

There a little story for ya:

Spoiler

A group of professional policeman staying in cover and shooting at 100 year old lady staying (she holding a tiny plastic bag) in doorway of exit from one big building without windows (company owner used private distributive of Linux), after 3 minute of intense fire lady pulled from her tiny plastic bag revolver Python (she can't learn javascript, poor lady) shot to death all policemen and herself by accident.

After 3 days journalist asked a question to director of Police Department about such tragic incident, and he answered: "It''s ok that we wasn't able to kill weak pedophile without military experience during 3 min intense fire, you see there a possibility for bullet to magically disappear in air because God with name RNG wanted to do this. But it's our fault there nothing wrong with this, next time our policemen would praying harder to RNG for better luck. Her life was ended by unfortunate critical fail during last shot and she had no fate points." "But how lady without experience in weapons can kill 5 policemen?" - asked journalist. Director said: "Random critz!"
The End.

 

[Trashos]

Just pointing out that missing a 70% attack three times in a row has a 2.7% probability of occurrence. So in a game where all your attacks are 70% THC, you are bound to see such a thing once every 35-40 attacks. Which is not very rare, and should be expected.

Carry on.

 

[Murk]

I don't care strongly one way or the other, but if you're asking for opinions then I'd rather pure RNG than any degree of meta-rigging.

I can appreciate that I may be frustrated at le 3x miss @ 70% THC* but such is life, and frustration is part of it.

*3x@70%! is a fitting codeword since it doubles as a frustrated growlix exclamation.

 

[Sensuki]

You're making the same error as those who complain about 3 consecutive misses by focusing on individual attacks rather than your average. The only question here is how many times you have to attack to get to that average. 100? 1000?

Since it's not an ARPG or a game where you attack hundreds of times per hour, a lower amount is probably better IF you're going to go down that route - that said I suppose it's something that could easily be changed through testing and feedback. I agree that 20-30 would be a good start.

If it's about making the combat easier, then no, and go fuck yourself.

If anything it would make it easier to tune the difficulty to the desired state (probably on the quite hard side) because of the more consistent results of rolls from both players and enemies.

 

[Mr. Magniloquent]

You don't need to rig dice, or create exceptions to your system's rules. You simply need to design a more robust system. Rolling multiple dice and taking the highest is an excellent one. Quantity of dice for reliability, scope of dice for magnitude.

https://anydice.com/program/10259

Follow the link. View 'Graph' and 'at least' for Data. I've created a simple comparison between 1d10 (Laymen), 2d10 (Trained), 3d10 (Specialized). Each of them has the full range of possibility. A scrub can get lucky and knock it out of the park, and a master can trip on their shoelaces. The probability of them doing so, are highly skewed such that the greater training will trend to higher values (reliability). You can also get into interesting match-ups between someone who is ordinary but trained vs. someone exceptional but untrained. So many ways to manipulate these dice to simulate most anything you want. For example, where you have your XCOM commando point-blank to an alien, give them a bonus of 2dx to their roll. Taking the highest from rolling 4d6 instead of 2d6 will effectively guarantee a good result (for an easy shot) without eliminating flubs "artificially" or changing the threat of the weapon.

 

[Mr. Hiver]

Just pointing out that missing a 70% attack three times in a row has a 2.7% probability of occurrence. So in a game where all your attacks are 70% THC, you are bound to see such a thing once every 35-40 attacks. Which is not very rare, and should be expected.

Carry on.

Thats all very nice but:
1. Games where you can attack three times are very rare.
2. In the few that provide such opportunity i played three consecutive misses - and repeated consecutive triple misses happen much more often.
3. It never happens on lower THC... or extremely rarely.

I guess it could be because by some quirk of xp distribution i end up playing with 70-ish % THC more time then with lower ones, but its hard to take that idea as evidence or a fact just like that, and im not the only one noticing this discrepancy.

You don't need to rig dice, or create exceptions to your system's rules. You simply need to design a more robust system. Rolling multiple dice and taking the highest is an excellent one. Quantity of dice for reliability, scope of dice for magnitude.

https://anydice.com/program/10259

Follow the link. View 'Graph' and 'at least' for Data. I've created a simple comparison between 1d10 (Laymen), 2d10 (Trained), 3d10 (Specialized). Each of them has the full range of possibility. A scrub can get lucky and knock it out of the park, and a master can trip on their shoelaces. The probability of them doing so, are highly skewed such that the greater training will trend to higher values (reliability). You can also get into interesting match-ups between someone who is ordinary but trained vs. someone exceptional but untrained. So many ways to manipulate these dice to simulate most anything you want. For example, where you have your XCOM commando point-blank to an alien, give them a bonus of 2dx to their roll. Taking the highest from rolling 4d6 instead of 2d6 will effectively guarantee a good result (for an easy shot) without eliminating flubs "artificially" or changing the threat of the weapon.

This is something i would be much more in favor for.
I dont think trying to force in a quick and simple solution is the best option. As the current mechanic is already a gross simplification - and obviously provides undesirable effects.

Trying to remove it as Larian did is the opposite extreme and shouldnt be in RPGs at all.

A more nuanced approach would be optimal.

 

[Fenix]

Needless to say, if a player misses even 20 times in a row with a 70% THC, he will rage quit and will never ever believe that the system worked as advertised (i.e. 70% THC).

You see, if the problem is to make believe player it works, you can go the way Underrail did - simply there is shooting range in game, where you can check probabilities.
This thing help greatly to mitigate all these stupid talks - you can simply go there and shot a session of any number of shots.
I did that few times and now know it works.
I think it's a good idea and demonstrative evidence of how everything works.

 

[deuxhero]

Why not have an option to write all random numbers (assuming the same range), even branches where the player reloads, generated to a text file to show the RNG is accurate?

 

[Daedalos]

The best possible outcome of this is always give the player what they expect.

After a certain skill/level/degree/percentage, make the attack always hit or ALMOST always hit.

Is it really fun to differentiate between 75 and 85 % hit rate? How important is it really for the overall gameplay?

Wouldn't it be better to just say, well, you reached THIS level of skill in guns, and hence, you should always hit with weapons. What does a To-Hit percentage really add to the game in terms of gameplay and fun?

You could also instead have something like hidden scores with titles instead, just as novice, trained, skilled, excellent, master etc.

Excellent and master skilllevels would ALWAYS hit no matter what, but there could be other benefits between excellent and master, whereas novice, trained and skilled might have actual differences in to-hit chance and other negatives.

tldr: It isn't FUN or engagning to miss a target with 85 - 95 % hit rate, and it sort of betrays player agency and expectations because of it. If I leveled the gun skill THIS much, well then I expect to always hit.

Making it harder to achieve the "always hit" skill-level is another discussion in terms of balance, but it should definitely be there.

I mean would a master sniper miss his target? Probably not, obviously it depends on alot of factors, but most people would expect an expert marksman sniper soldier dude to know his shit and always hit the target even in shitty conditions.

 

[Van-d-all]

I think the whole issue comes from misconceived representation of randomness in terms of probability distribution. This boils down to the old D&D d20 vs GURPS 3d6. Linear probability leads to those kinds of ridiculous outcomes, because it models borderline cases as equally possible as common ones; and rigging it to player favor is equally unnatural. IMO applying band aids to a broken system fixes nothing, so just adapt a proper bell curved model...

 

[Fenix]

I wonder how many people who ask for truly random rolls would be ok with missing 3-5 times in a row in-game. This isn't challenge, just bad luck.

If they are ok with hitting 5-6 times in a row, they should be fine with such misses. Otherwise don't know what to say.

Devs should rig RNG ofc, because in another case there can be situation like in XCOM

I heard it doesn't display real numbers, that's the problem of the game.

First, you've earned this card with 3 misses despite your high skill, meaning investment and planning that didn't bear any fruits.

What if different kind of attacks cost different in AP? Could you use some cheap attack, target farthest enemy then when you know that your hit-card is near - you can make shot fro sertain, with devastating effect. ???

Second, we'll use different decks for different attacks.

Well, looks liek the answer...

As much as I can understand (If understood devs intentions incorrect, fix me later) it's about making combat more logical for people

As much as I can understand, it's about appealing to more popamole audience which is increasingly more and more stupid with time, as society degrades.
While question "what's can change the nature of Random" 30-40 years ago didn't generated such backlash as today because people were evenly educated - at least those who played such games, today is different - there are many of those who intellectually can't cross the street on the green light, while they can press a buttons on gamepads and even keyboards, can connect to internet and other techy stuff.
The gamer's society has changed...

Also, this your screenshot is sophism - this number represent a number of factors - state of soldier, state of enemy, weapon etc.
Do you really think that if have knife you can hit everyone on the onther end of the knife? Even those who was trained to dodge such attacks?
I know this is a fun pic, I laughed too, but the logic behind it is wrong.

 

[Fenix]

I have a question - how this problem was solved in f.ex. Jagged Alliance 2?
I ask those who played - I think JA2 has no such manipulations with random, so it's interesting how players solved their problem.

For example you need to kill enemy whithin next turn no matter what. Your THC is infamous 70%.
What will you do?

Also, my overall conclusion - ideally, and I mean ideally I would be glad to have BOTH systems in game.
Thus I can compare what is best for what.
If it takes too much resources then... do as you want, I prefer old crystal clear random...

 

[agentorange]

The people who whine about this stuff fit the character of a person who will never be satisfied. You can deconstruct the entire system of AoD and New World and remake it into some kind of deterministic system and they would still find some way to complain about it not being fair. So I don't see any reason to attempt to appeal to them at even the most minor level.

 

[janjetina]

You could abandon modelling attacks as independent events and use a simple 1st order Markov chain, like below, or some other Markov chain.

x denotes the state after i consecutive misses. x[0] denotes the initial state and the state after the last hit (each hit resets the current state to x[0]). So, only two transitions are possible from each state x: one to a state x[i+1] (a miss) and another to a state 0 (a hit).

p(x[0]) = p : probability to hit after 0 misses (fixed)
p(x[0] | x) = p(x) + a*(1-p(x)), where a <= min ( p(x), 1 - p(x) ) : probability to hit after i misses
p(x[i+1] | x) = 1 - p(x[0] | x) : probability to miss after i misses


For example, for a = min(p(x), 1-p(x)) starting with p=0.6, you get the chain p[X] = {0.6, 0.76, 0.82,...}
Starting with p=0.4 you get p[X] = {0.4, 0.64, 0.77,...}

A more conservative option would be p(x[0] | x) = p(x) + a^2.
Both options boost mid-range probabilities the most, while very low and very high probabilities are not as affected.
If one wants to avoid a ridiculous situation like a point blank miss, one could simply implement a hard probability threshold to transition into x[0] (i.e. to hit) when p(x) >= T).

In addition to probability p, only current probability p(x) and equations above need to be memorized. Implementation is a piece of cake. Then use non-popamole testers for proper model selection between the independent attacks and a few Markov-like models.

 

[Trashos]

janjetina, how are you going to let players strategize over this?

I guess it could be because by some quirk of xp distribution i end up playing with 70-ish % THC more time then with lower ones, but its hard to take that idea as evidence or a fact just like that, and im not the only one noticing this discrepancy.

Yes, that's what I think is happening. As you can see here (and probably in other games' forums too), developers are trying to alleviate players' frustration, not backstab them. Wouldn't make any sense.

 

[Mr. Hiver]

Excellent suggestion janjetina

how are you going to let players strategize over this?

By presenting such system in gameplay, lore and setting terms. Of course.

Its not like it drastically changes how things work, in fact it simulates real world better and so it would be naturally understandable to players as is, through effects they can experience and see in the gameplay.

Yes, that's what I think is happening.

Thats completely irrelevant, as my actual experience is completely different. And especially because such gross simplifications as that form of THC distort how things actually work.

I never said anything about developers trying to backstab anyone... try not to address me with strawman arguments, however unintentional they may be.
I think this mechanic can be and should be improved. NOT made easier for players, NOT removed. Improved.

 

[Tigranes]

The best possible outcome of this is always give the player what they expect.

After a certain skill/level/degree/percentage, make the attack always hit or ALMOST always hit.

Is it really fun to differentiate between 75 and 85 % hit rate? How important is it really for the overall gameplay?

This is a good example of what I don't want to see - design being driven by some idea of player expectation that dilutes system consistency.

If you are going to have a percentage-driven THC system at all, then you are saying that (1) the many vagaries of combat will be abstracted into a possibility of missing; (2) that every action carries a certain risk, and there is a strategic/tactical element in learning to mitigate those risks and also responding to when things go astray. Given this, there is simply no good gameplay reason to say "above 80% everybody hits". There is only one reason for such a thing: "people who don't understand probability get annoyed and we should make them not get frustrated." I really don't see why the above systemic benefits of a THC system have to be sacrificed for that. (You might as well just bump all THCs up by 20% across the board, and make it easy for people to get to 100%, instead of lying about 80%.)

Now, if we were to say, THC is silly anyway and we should have a different system, e.g. one where attacks always hit but can be set up for guaranteed dodges/mitigations, etc., then fine, put up such a system and argue for its strategic benefits.

 

[Daedalos]

The best possible outcome of this is always give the player what they expect.

After a certain skill/level/degree/percentage, make the attack always hit or ALMOST always hit.

Is it really fun to differentiate between 75 and 85 % hit rate? How important is it really for the overall gameplay?

This is a good example of what I don't want to see - design being driven by some idea of player expectation that dilutes system consistency.

If you are going to have a percentage-driven THC system at all, then you are saying that (1) the many vagaries of combat will be abstracted into a possibility of missing; (2) that every action carries a certain risk, and there is a strategic/tactical element in learning to mitigate those risks and also responding to when things go astray. Given this, there is simply no good gameplay reason to say "above 80% everybody hits". There is only one reason for such a thing: "people who don't understand probability get annoyed and we should make them not get frustrated." I really don't see why the above systemic benefits of a THC system have to be sacrificed for that. (You might as well just bump all THCs up by 20% across the board, and make it easy for people to get to 100%, instead of lying about 80%.)

Now, if we were to say, THC is silly anyway and we should have a different system, e.g. one where attacks always hit but can be set up for guaranteed dodges/mitigations, etc., then fine, put up such a system and argue for its strategic benefits.

Yea I'm probably leaning more towards not having a THC system at all, because it is quite silly in terms of combat, it renders it very shallow imo, because it gets to dictate combat depth and strategy.

Even in a THC system, I think there's a perfectly good gameplay reason to say, well at THIS certain level of skill, THC becomes a non-issue, because of how skilled you are, effectively ensureing that missing is no longer an issue, because of how specialized you are. You could even add new bonuses on top of that. So only the really specialized builds, get an always to hit chance, it's a choice, it's a strategic advantage of having specialized yourself so much, I don't understand what the problem with it is.

If you don't allow the player to have any means or forms of mitigating frustrating situations, then its really bad game design imo. You can mitigate THC and the frustrations thereof, by allowing people to ALWAYS hit beyond a certain point, but they sacrifice much doing so. Give the player the OPTION to always be able to hit, directly through specializations, but instate a more complex combat system in conjunction with THC.

If you don't give people the option to mitigate missing a shot at 95 % THC with every combat specialization, and it only encourages save scumming and otherwise cheesey behaviour not intended. I want a gaurentee, that if I specialize heavily into weaponry, then I won't encounter unexpected dumb RNG situations, that's basically what it boils down to.

 

[Tigranes]

In that case, your THC should be 100%, or possibly 95% as a cap. i.e. the skill gains should get you there. I don't think there's any benefit to getting it to 80% then fudging the rolls so the player doesn't cry when they miss.

THC-based games have, in general, always had plenty of ways to mitigate the risks, anyway. You often have many ways to raise your THC, temporary raises, lower enemies' AC, so on and so forth. Rarely are you playing the entire game missing more often than you hit. So this isn't a new point either.

'Encouraging save scumming' is not really a major problem for me either. If people want to waste their lives reloading a video game every milisecond, so be it.

I just don't see any great reason to lie about the THC here, only a possible argument for making it easier to achieve very high THC or to get rid of THC altogether.