To RNG or not to RNG

[Tynan]

Just a design discussion.

This isn't really new but as we approach 1.0 I've been thinking about it more. Basically I've seen a variety of variants on messages like this:

I just started a colony with no reloadable saves

I dont know how I'll like it, because when I reloaded it was for BS unlucky stuff

for instance: a melee guy I had (decked out in plasteel plate armor and advanced helmet) was one shot KO'd by a friendly fire bullet to the back of the neck

instantly died, wearing full plasteel, pure BS

It concerns me because this really goes to the heart of what RW has always been designed to be.

The original design intent of the game is that sometimes unlucky things will happen (especially if you take risks like with friendly fire), but the game is also designed so you can recover from bad situations. So it's up-and-down. Mostly you'll succeed, but sometimes you'll lose things, even if you play well. And that's part of the game. Kind of like poker.

This is as opposed to making good play always lead to an unbroken string of triumphs (like many games). The unbroken-triumph thing works fine but I thought it was also worth trying to do something different. Specifically focusing the game on story generation kind of requires that it can't just be unbroken perfect triumphs; a story of continuous victory isn't that interesting as a story. Every good story has setbacks in it.

Traditional games handle this by putting the failures in cutscenes and making the gameplay 100% victory. But if you generate the whole story as RW does, the failures have to be in gameplay too.

But very few games are trying to generate stories. Most games are skill test/reward pumps. We test your skill, and reward you in accurate proportion to the skill you showed. This is what people are used to, to a large degree, which may explain why they emotionally reject outcomes that fall outside these boundaries.

Is the anger reaction a consequence of caring about characters, or them being not easy to replace? Nobody minds that much when a marine dies in StarCraft because they're replacable. But either for story or gameplay reasons, people do tend to at least care about when people in RW die. Which sounds like a design success. But it also creates this anger reaction in some circumstances.

So basically there's some kind of complex psychological/design conflict going on between the traditional skill-test/reward pump game design and the story generator goals, and it creates this friction. Because RW is to some degree trying to mix both, but the goals of these two conflict strongly.

I'm thinking about how to handle this sort of thing. Thought it may be worth discussing. Questions to consider:

1. Should the game have a such a thing as bad luck outcomes that's not induced by some obvious, non-pressured, voluntary player decision? Or should I make a universal design standard that nothing bad ever happens unless the player actively induces it or makes some clearly-traceable mistake to cause it?

2. Should I just ignore some classes of player feedback as simply not linking up with what RW is? Are some players worth leaving alone to try to make a game that's different from the usual assumptions? Even if it leaves them pissed off because they intepreted a story generator as if it were a skill test?

3. Should players be able to consistently avoid losing people/resources even at high difficulty? At any difficulty?

4. Is there a way to set expectations (relative to the whole game, or relative to a given difficulty level) to encourage players to accept some degree of randomness to game outcomes? Or will they always reject this randomness and demand to be rewarded in accurate proportion to their skill/effort?

[DubskiDude]

Can't really speak for the community at large, but I personally savescum any time a pawn dies. There's a certain randomness already to Rimworld that can lead to pawn death, like friendly fire, diseases, infections, or plain old bad luck. And when a pawn dies, the colony suffers, from a little to a lot, depending on the pawn. It irritates me when the numbers are so far from being in my favor that a character dies, permanently.

Good pawns aren't easily replaceable in my experience. The beginning of the game is the most crucial when sorting out the "bad pawns", i.e. Pyromaniac, Volatile, Body Purist, Lazy/Slothful, Slowpoke, Depressive, etc. Pawns with these traits aren't worth it or are just too risky to use, because keeping them requires a bed, food, micromanaging, their mental breaks, etc. Why bother if the pawn doesn't stack up with the others? This is why I outright banish a pawn if they join and have even one of those traits. These traits are just a net negative. I suggested in the past, though, that these "mostly negative" traits should have an upside to them. Maybe not a big enough upside to where it makes the negative trait worth it, but just something so that you're not just staring at fire-starting sprees or -18% break chance. For example, maybe Slothful can mean less work speed, but the pawn requires slightly less sleep. Regardless these negative effects have pushed me to being very selective with pawns, and that will likely persist forever with my play style, because I like to min/max, and a pawn with a big negative trait that can't be fixed just isn't worth it to me.

I have noticed that I'm more forgiving of Rimworld if my pawn loses a limb or an eye, because now that bionics are possible, they'll eventually be replaced. Not sure how that could be emulated to fix pawn death, but maybe it could go somewhere, somehow.

[Kalre]

Some people like me, love to see the game as an RTS with some social interaction mechanics and i guess thats why its pisses us soo much, wich is probably not the way the game is meant to be enyojed.

Maybe add a NOT soo RNG/Story intensive mode ?

Love the game but 1.0 has taken it too far on frustration, i know its been tested and open to change, but some changes are just very questionable, Drugs per example are soo high risk for "some" reward thats not really a good idea to use them taking a huge part of the game out.

Not to mention some of the mental breaks where you have little or non counterplay (i literally roll eyes when i see Run Wild Lol).

Raid variation is also a huge issue thats been only getting worst and worst with each Update.

And yes, with proper preparation and mechanics we should be able to see some kind of consistency in our bases, colonist mood. Maybe just openly suggest to the player to play Randy for the "get fun with rng" mode and let other storytellers for more consistent/combat focus gameplay.

Sorry for bad english

[bbqftw]

RNG can actually enhance the skill-testing element of a game. I think Xcom franchise is game that pulls it off very well - it tests your ability to plan and react to uncertainties. If there was less RNG, then many decision trees become simplified, and arguably the gap between a master and a mediocre player narrows.

Lets look at tribal or CL start year 1 to understand how "good" (in my subjective term) RNG exists in the game.

You could face month 1 cold snap / heat wave, year 1 toxic fallout, year 1 siege with sniper rifles (on b18 at least), psychdrone in various combinations, early psychship, tribal sappers - or you might not. But if you want to increase your win%, you need to be able to prioritize how you will prepare in the face of the possibility of these outcomes. So even small things like deciding whether to prioritize growing / fridge, or tailoring / mass hunt, or any combination of these opening prioritizations even in the first 5 days of the game can have a fairly large effect on how well you deal with early game events.

Contrast to some naked brutality scenarios.

You could get plagued on week 2. Unless you rolled a 10++ med skill pawn, that means you die. It has practically zero skill testing element to it.

Of course most scenarios are not as ridiculously skill-independent as this, and many players do complain about RNG dominating their outcomes when they themselves made critical mistakes that were as influential towards putting them into the position they were in. But I think subjectively there's an  valid complaint that things like lancers fall closer to the NB-plague scenario than the CL/tribal scenarios I described.

I'll talk about what my mindset is when I fight a lancer, which the game can deploy a half dozen even in early year 1 / early year 2.

At full cover and high range, you're looking at 3-5% chance to hit with a highly armor penetrating weapon. Should one of these lances hit, I then roll 100-sided dice, with rolling less than 6-10 (typical year1/year2 armor) being a kill. That's all the story is to me when someone dies. But my victories in such a shootout feel hollow, and so do my failures - if I get neck shot, I shrug - for my only fault is not finding a way to engage with the lancer that doesn't involve it shooting my pawns at all.

In a similar vein, in situations where I played horrendously bad but lived due to coinflip RNG, I have restarted, for 'winning' there feels empty, and I know I did not deserve to win.

(of course, there are various micro tricks we can use to make the shootouts much more asymmetric, abusing aspects of the AI pathing, etc. But I thought you disliked cheese, tynan~)

This is not to say that combat should be riskless, or that OHKOs should not necessarily exist in the game. However If you tune the chances of OHKOs high enough, the lesson to players interested in improving is that you should avoid interacting with this wonderful combat system, which in many ways is deeper and more interesting than even dedicated strategy games. I think that's a shame, but we may have different intentions, and ultimately it is your choice. Just don't be surprised when someone new tries to learn the game and instead of some instructions on the cover system he is given a diagram about how to simplify every raid to various trap/cheese corridors.

2. Should I just ignore some classes of player feedback as simply not linking up with what RW is? Are some players worth leaving alone to try to make a game that's different from the usual assumptions? Even if it leaves them pissed off because they intepreted a story generator as if it were a skill test?

This is a choice only you can truly make.

But when I play b18, and even look at the 1.0 mood management and general resource management aspects, I see a game that was (accidentally?) designed as an excellent and pretty deep test of skill with lots of replayability.

3. Should players be able to consistently avoid losing people/resources even at high difficulty? At any difficulty?

This is very hard to assess - because frankly, I doubt that the strongest players even stream or post at all.

1. Should the game have a such a thing as bad luck outcomes that's not induced by some obvious, non-pressured, voluntary player decision? Or should I make a universal design standard that nothing bad ever happens unless the player actively induces it or makes some clearly-traceable mistake to cause it?

I think this is impossible. The game is so deep that actions committed several months before, or fundamental mechanical ignorance, are the dominant factors in a failed run. And unless you are looking closely, you might often not notice said factor, or misattribute it (to RNG or whatever)

[Tynan]

I have noticed that I'm more forgiving of Rimworld if my pawn loses a limb or an eye, because now that bionics are possible, they'll eventually be replaced. Not sure how that could be emulated to fix pawn death, but maybe it could go somewhere, somehow.

Resurrector mech serum?

[Greep]

Mostly it comes down to game length and randomness I think.  If you lose a game in FTL because the RNG gods decided the flagship was going to have every drone hit, people can be annoyed at first but you can just pick up and try again, and usually get a completely new experience on your next try.

In Rimworld, the game is very long and the differences in gameplay are not as much (in fact, the storyteller even tries to force every game to be the same in the background!), so I think efforts to make the game forcefully punishing are always going to be a failure.  People just see a loss coming and think, "well I've got a job and want to relax, and I'll just be playing more or less the same experience if I have to start over, so screw that I'm savescumming/using turrets/ insert attempt at not putting pawns in danger"

The other part is something you just can't really understand till you play it yourself, is just how grueling straight up combat is on any difficulty, to the point where some form of tricking the A.I. in a dumb way is necessary if you want to continue playing.  If you have 70 large threats in a game before launching your ship, and you recruit like 20 pawns in total, you have to pretty much win almost all of the fights unscathed.

Think how weird it is to expect players to fight 50 v 20 battles (or 20 v20 on rough), over and over, and not lose a pawn more than once or twice?  If you're fighting in any remotely fair way, you're going to lose every single one of your pawns in a single fight, let alone a single pawn once every 5 fights.  So when people try to play the way you would imagine they should, and only see that it is in fact impossible, it probably annoys them.

[erdrik]

Can't really speak for the community at large, but I personally savescum any time ...
...

I mostly do this too, but for me the triggers for save scumming is:
-Does my colony end with this? Yes.
-Was this even possible for me to prevent? No.
My general feelings or thoughts when this happens is, "Well thats a loss. Lets keep going and see how things would turn out if that didn't happen."
In those situations where the colony doesn't survive I see it as having "finished the story", but then becoming interested in the "alternate timeline story" where the colony survived.

I actually think part of the problem is that I don't care enough about my colonists or their story. If I did, Id be interested enough to see it through to the end and then accept that ending and closure. But because I don't the story feels like it ended prematurely, and I become more interested in what would have happened.

Its kind of like when Im watching an anime and get real excited about whats to come, only to find out it got canceled after season 1 on a cliff hanger.

[5thHorseman]

I personally only reload a save in 2 instances:

1) I made the save specifically to test something that I wasn't sure the mechanics of, reloading once I know the mechanics so I can "do it right." Examples of this stretch from mining that last pole under a mountain to see how cave-ins happen, to sending out a caravan to fight bandits. The second may be cheesy save scumming but if the literal alternative is colony death, so be it.
2) Something so bad happens that enough colonists die that I don't want to play that colony anymore. This generally means more than 2 people die, or maybe 1/4 or so of the colony dies. If it was preventable, I go back and prevent it.

So if one of my guys gets shot through the back of the head (it's happened though he wasn't decked out in power armor) I'm like "oh well" and move on. If him getting shot causes a fault in my front line that allows the raider with the rocket launcher that he was fighting to get a shot off that takes out 2 more of my people and then the raiders decide to grab another one and run off... yeah I'm reloading because the colony needs an amount of luck that this game doens't give, to survive. And I'd rather spend the next 5 hours starting a new colony than limping along waiting for the death blow.

If the game was a bit more lenient after a loss (and I think you're doing that, with the fun points and ramping and whatnot though I can't tell for sure, I've not tried it) then I'd be more willing to roll with the punches.

Note, a lot of this is from before you made that development graph public. I've not played a TON since then, and what I have played, I've so far not had a massive defeat. I did lose 2 guys to one-shots from Scythers (an hour before you toned them down ) and kept going and have so far done okay.

I for sure would like it, if you added a "no actual killing people" gameplay mechanic, if it was an option.

[Disnof]

I came to RimWorld because I love the aspect of do x and y or z will kill you, with RNG thrown into it.

Ex. try building more out of stone or a raider with molotovs will come burn everything, but maybe molotovs never show up so a wooden base could survive sometimes but not usually.
Ex. try your best to get a sniper rifle/mortars early or a mech ship will be hell.

The game currently has a few things in it (being one-shot in the best armor, because RNG) that the player can do little to nothing about. When you get destroyed by a raid that was too strong you can try to learn from the situation and improve. But, when someone just dies because RNG, then most people are just left with savescum. The game really does need ways to hurt the player for those up-and-down, but the more uncontrollable it is the less enjoyable the whole experience is. I would think that a system of risk management would be better than better luck next time/reload.

Although no mater how detailed a game is people will try to min/max it, and will find the best way to win. Having RNG come in and just say NO hes dead goodluck seems to be the best way to spices things up. So you shouldn't abandon a good part of what rim is just try to balance it with frustrating the player.

[patr]

Well its a fine line between RNG to enhance the story and making the player feel like they are no longer a part of the story obviously. If the players own decisions don't mitigate outcomes then you aren't really playing the game at all. I don't think people are typically arguing that RNG should not exist in the game at all, but recently some of the latest patches have had some very strange outcomes in relations to breaks and combat that could reasonably make a player wonder if the things they are doing are properly mitigating outcomes.

As far as your specific questions
1: I think in the context of rimworld this is an appeal to extremes. I personally feel like bad random outcomes is the meat of the game, I just also feel like if I put someone in armor it should do something. For me the difference between 1.0 and previous iterations is that it tends to not recognize the players efforts to mitigate things and the outcomes are more extreme.

2. Thats really your call.

3. No.

4. The players mitigating actions should make bad outcomes FEEL less frequent. Thats where 1.0 falls short IMO it doesn't feel like I am effecting the outcome in certain situations. If a guy in a power armor helmet gets his brains scrambled every once and a while I can live with it, but if it happens every other raid it makes the player questions whether their actions have effect.

[Kalre]

Storytelling as an "excuse" for shitstorms is hard to accept because they ultimately always end in someone dies, theres no such thing as "Colonist joins as he knew of his fathers dead and wants to help you out of this situation", so you dont get the feeling of rewarding or interesting storytelling at all, just punishment after punishment, whereas in high difficultys plain means that your 10 hour colony dies right there. Theres no real motivation as its right now to continue with an injured colony as you certainly will not expect a huge come back or anything. Once again this feel changes for each player and each playstyle wich is why i would implement different storytellers for each kind of person, wich is possibly a lot of work.

[ZaPhobos]

I came into the game expecting to get my ass beat, especially when I put it to Cassanda Extreme permadeath.  Since rimworld is a story generator, and not a test of skill, I have no issues with the RNG.  Much like how in XCOM you'll get unlucky and miss a 90% shot, you can get unlucky here and end up having your pawns killed.

The issue with getting pawns killed, however, is that oftentimes when one dies it's either a nonessental pawn, or a VERY essential pawn.  Each person eventually ends up a specialist, for the most part.  Your dedicated grower or researcher and then you end up with haulers and cleaners, which don't set the colony back much, if at all.  Often times it's getting beaten down over and over and just praying you keep everyone alive.

The second your expert builder or grower dies your colony is set back a TON, since they were the fastest at sowing, or the fastest at building.  Time is critical when you're getting raided.  But that comes with the territory, as I've come to learn.  If someone dies, I'll try to bounce back.  At best, I can manipulate RNG to keep things together by walling up or prioritizing defensive gear.  I do not want to be on the receiving end of a colonist death until I'm fairly far it.

1) Luck is random, and I'd rather be lucky on the rim than skilled because I know luck will screw me over eventually.  At the same time I know Cassandra is trying to kill me, Pheobe will do nothing for months but is trying to kill me, and Randy is truly random.  All I can do is reduce the chances of something terrible happening, such as with sickness prevention drugs and armor, but until I have enough resources to actually get these things and keep up their production, I'm at the mercy of the storyteller.

2) Rimworld is a story generator as opposed to a normal game and should be played and expected as such.  Winning isn't the true goal, it's the story it makes for me.  If all my people get wiped out sure it sucks but I have the epic tale of how they warded off the mechanoid menace but ultimately fell after countless raids.  And that's perfectly fine, rimworld has it's own niche, and people are very much enjoying it, permadeath or not.

3) Certain events are nonsensical, resource wise.  When someone goes on a tantrum and destroys my medicince, my components, or some other very important or rare items before I can get my wardens to beat them down it does mess with me.  That type of resource loss is available at all difficulties and are really, REALLY inconsistent.  They could end up destroying a "good" quality weapon or they could destroy the x50 components that I was planning to use which just... isn't proportional. 
Losing people on the other hand, I try to mitigate as much as possible and I try to avoid it at all costs.  If someone must die I prefer they be the hauler or cleaner, because some people are just more useful than others.  At higher difficulties I expect this to be incredibly difficult, but on medium to rough I only expect it to happen every so often since raids don't occur as often nor are they as large.  Sure it's frustrating when you're attacked again right after a raid, but I've come to accept that, and mitigate my losses, but mostly mitigating colonist losses.  If my stuff gets stolen that's fine, but I can't replace a good colonist.  Getting back together after large raids, especially in the early game, is simply a chore.  People are hurt and angry, constantly, but once you're back to speed you're fine as long as you didn't lose anybody.  Colonists are the #1 resource, weapons and food be damned.

4) Expectations really should be geared to training the player that the game isn't going to be fair, especially at higher difficulties, but that you can prepare for it.  At the same time it's really demoralizing at times when you can really do nothing and you need to take a step back.  In the early game it's especially terrible if you have back luck since you've only got so many people to spare, and even 1 colonist can make the difference.  Cassandra really does try to outright kill you though, and I have my expectations dead set if I play with her.  With my luck, I always have my expecations for the worst with her.  Pheobe is like Cassandra but with lots of space between.  Randy on the other hand?  He'll more often help the colony with random item drops and then suddenly send something ridiculous once your wealth spikes.  He's the only storyteller that I can expect nothing from, and that's perfectly fine because I already know what the other two do.

[bbqftw]

Two related thoughts -

1) Is there a perception that fights don't feel impactful enough that drove recent balance changes lowering vital organ HPs, increasing one-hit-kill chances across a wide range of weapon scenarios? I would suggest an alternative with adjusting healing rates down, so that there's less of an 'all-or-nothing' granularity of fights.

2) If you're interested in storytelling mechanics, there's really nothing in the game that overtly rewards people for sticking through adversity. The only outcomes of winning a closely hard fought fight is your survival ...and host of only negative outcomes (you could say shooting xp, but that can also be gained with approaching fights in a very cheesy manner). Maybe some guy with injured eye can learn to perceive better in time (so eventually restoring full vision would give him elevated sight relative to normal unhandicapped pawns). Or guy that kicks an addiction gains better tolerance to mental breaks. Maybe some permanent bonuses that can only be obtained through managing auto-reject trait pawns. So that you feel like there's actual accomplishment for sticking with crappy pawns and crappy circumstances, instead of doing so as some masochistic handicap.

[Wanderer_joins]

Resurrector mech serum?

But it's precisely RNG. Rez mech serum could be rare, very valuable but still have a rare and semi reliable source like orbital exotic good traders, say 0-1 mech serum per traders.

I've had colonies with 0 opportunity for rez mech serum.

Or it could be a special quest you'd have to ask for (same for healer mech serum), idk.

Otherwise i think most people haven't yet realized how losses help your colony with fun points, it's just been implemented.

[Fleurs]

Is the anger reaction a consequence of caring about characters, or them being not easy to replace? Nobody minds that much when a marine dies in StarCraft because they're replacable. But either for story or gameplay reasons, people do tend to at least care about when people in RW die. Which sounds like a design success. But it also creates this anger reaction in some circumstances.

I haven't played since the update before the road/river added. But i think perhaps if less pirate would just die randomly instead of being incapacited, it would make people more easily replacable. Sure, it could allow a lot more potential recruit to remove the auto-die instead of incapacited mechanic completely, but between the really mutilated incapacited (i won't recruit an incapacited pirate if he has lost a arm), the bad trait pawn, the pawn that have skills that are already covered by the remaining people i have, the ones that have 1% recruit chance, and the ones who really died, it won't matter that much. Actually, if 10% of the raider really are incapacited, the decent enough ones that aren't mutilated are very limited, thus, make loss hard to recover.

Perhaps tying this mechanic of auto-dieing with difficulty level harder than the default one (rough?).

Perhaps allowing much more recruitable prisonners would allow the loss of a pawn to be recoverable more easily, while still inflicting a penalty, since you will lose time to recruit him, and he won't be as skilled/leveled as the people you lost in the battle. If you remove that auto-die function.

Additionnaly, i've read here that now, with the new miss feature on shooting, the bullet can randomly hit ANY square in the field of vision, including pawn on a tile next to the shooter, that doesn't help on the rng.

(Btw, if you read this, i don't know if it has already been fixed, but it seems tomb and sarcophagus doesn't display the name of the pawn anymore, that was a most liked feature)