Nested Risk and Noita

Been a while since posting, but that's because I've been cooking up a fairly in-depth one. 

Randomness as it works in games generally, and in particular in ttrpgs, has an interesting feature to do with buying into different layers of the gameplay. This basically means that you construct games within a game. This has implications not just for good game design, but also for perceived fairness. The old-school style (if it's coherent enough to call it that) is often criticised for being unfair or confrontational, or at least overly blasé about killing off characters. Hopefully this will explain how the approach is a) justified and b) actually a valuable thing to bring to games.

🕷

Allow me to explain. At length, I'm afraid.

Recently, I’ve been playing the excellent indie videogame Noita. Noita is a roguelike, meaning the environment you explore is generated randomly each playthrough. At each level, you get to a shrine, where you can heal up. If you somehow bring a monster into the shrine, the gods, in their anger, will manifest a spirit to fight you, which is to all intents and purposes unkillable.

On one playthrough, I found myself doing really well by the end of level 1, when I got to the shrine and discovered a worm had burrowed into it, angering the gods, and effectively ending my run there.

Initially, I was annoyed. The randomness of the game had created a situation where, regardless of how skilfully I played (although let’s not make any assumptions on that score), I was going to die one level in, and lose the haul of good items and so on that I’d collected – forever, since the game re-randomises each playthrough. But as I thought about it, I came to a realisation that dispelled my frustration.

The thing is, I was caught up in the moment-to-moment gameplay of taking risks for the chance of reward – choosing whether to fight or flee, whether to search around for just a moment longer, despite my wavering hitpoints. I was winning that game. But I was also playing another game, within which that game takes place. This other game was betting what the environment would be like when I got down there. That is, how tough the fights would be, how good the items are and whether they’re easily obtainable, and, yes, whether the game was even survivable in the first place. The skill-based game takes place within that other, antecedent game of betting what the challenge is going to be like, and whether it’s even beatable. That game can override the skill-based one, but it’s not unfair when it does so; part of the buy in for the skill-based game is the possibility that things won’t go your way. And far from being an undesirable feature, that’s part of the charm of the roguelike genre: You never know quite what’s down there, including whether there’ll be an unbeatable challenge around the next corner.

This all keys into interesting features of rpgs specifically, and even more specifically, OSR design theory.

🕷

Let’s break it down with some analogies. When you play a slot machine, you have one choice point, which leads, via randomness, to one outcome being disclosed. Your choice is whether you put in a coin and pull the lever – basically, whether you ante up. You don’t know (probably) the exact odds of winning or losing, but you have some idea (hint: it’s low). You might play the game several times, but you’re repeating the same choice each time – you’re not progressing the game. It’s a new game, so to speak, each time you play.

Now think of blackjack. This is all going somewhere I promise. In blackjack, your choice is whether to hit, taking a card, or stand, staying with what cards you have. Again, at each point, you take a decision on whether to ask for another random outcome, taking a gamble on its being a favourable one. But blackjack is one step more complex than slots – there’s an antecedent bet you make. That is, like slots, you choose to ante up to enter the round. And, like slots, you do this knowing you’re your money might effectively be lost at this point, before the opportunity to make any further choices, and without the opportunity to claw back the round through skilful play. That’s because the cards in a given round of blackjack can fall so that it’s impossible for you to win, given the limited control you have over the game (i.e.: where standing will leave you with a lower total than the dealer, but hitting will have you exceed 21 and bust out).

In the moment, it seems unfair when this happens. But that perception of unfairness quickly vanishes with the simple reflection that you accepted that this might happen when you ante’d up. At that point you rolled a die, figuratively speaking, that determined whether it was even possible for you to win. Everything after that was contingent on that one die roll, which you bought into with your eyes wide open. Contingent, that is, in the sense that the die roll could override your skilful play, not in that the game comes entirely down to that one choice point.

You might be able to see where I’m going with this. We can see the same phenomenon – call it “nested risk” – in ttrpgs, specifically those set up to work like old school D&D. A campaign where players have the opportunity to pursue genuinely risky activities, but also the opportunity to refuse or abstain from so doing, or even to pick between them, presents a scenario like blackjack.

Dungeons offer a good and extremely central example: upon entering a dungeon, you take a bet that the dungeon, or the route you take through it, will at least be survivable (for a party of your level and makeup), antecedently to getting into the actual nitty-gritty of the game itself. The random factor here is that the route the party takes as they explore will be characterised by a lack of information – it’s therefore not a test of skill to navigate around the certain-death encounters, but a roll of the dice in the knowledge that there may be such down there. We find another layer (pun intended) in the convention of placing tougher enemies on deeper levels. Not only do you ante up by entering the dungeon in the first place, you place a further bet by descending to any particular dungeon level.

In a sense, this justifies several aspects of old school design that may seem unfair at first glance – things like save or die poisons, traps with no immediate clues that require a shot-in-the-dark skill roll to spot, or monsters that can only be harmed by magical weapons. Obviously this is to be taken with a pinch of salt, and I’ll explain some exact caveats a little later.

Let’s focus for a moment on the magical weapons thing, as something we can use this schema to clear up. It always used to puzzle me – how exactly does it improve gameplay to have your monsters be immune to non-magical attacks? If your party has magical weapons, it’s just like a normal monster, but if they don’t, there’s nothing most of them can do to harm it. Either way, the requirement doesn’t make the actual encounter more interesting, seemingly, so why have it? I think this logic has been applied in later editions, which is why such features have largely been removed (I don’t think there’s anything in the current Monster Manual that’s straight up immune to non-magical attacks - Edit: There actually are some monsters that have this, but the prevalence of full immunity appears greatly reduced).

But it actually does serve a purpose, specifically, at the level of buy in. That is to say, if you have no magical weapons and you enter a dungeon (or a certain level thereof), you do so on the understanding that there’s a chance you’ll face an insurmountable challenge. The rewards, which should be consistent with the general challenge of the dungeon or level (not of the specific encounter) are the pot, and your life is the ante. And the DM doesn’t have to provide an opportunity to get a magic weapon so they can say, when you die, “Well it’s your fault for not following up this lead…” Nope – you chose to gamble, and you knew the stakes, if not the specific odds. After the point of buy-in, the game doesn’t have to be winnable by a sufficiently skilful player in order to be fair.

This explains the appearance of unfairness: You’re caught up in playing the moment-to-moment game of making choices and taking risks while within the dungeon, i.e.: the decisions and risks you take from encounter to encounter, or in individual encounters. According to that game, it feels unfair to come across a trap you couldn’t defuse, for instance – one feels as though it should be figure-outable. But this is the same contextual phenomenon as in blackjack, or Noita. It feels unfair in terms of the game of hitting or standing, or of exploration and pushing your luck, when it’s revealed you couldn’t have won – but you actually made a prior bet on whether or not you could win in the first place.

🕷

Now, there are a few caveats to this, and I’ll address some of those before giving some more detail on why this is actually a valuable feature. First up, this schema doesn’t work if you force the players into a dungeon (or encounter area). In that scenario, they aren’t choosing to ante up, so it’s not their fault when they get toasted – you’ve just chosen to put them in an environment where there’s a chance they’ll get a random result of “you’re dead”. I imagine this might hit some spot between dramatic irony and gambling for the DM as they watch the PCs bumble around their death maze (with maybe a touch of schadenfreude thrown in), but this is because it was the DM’s decision to put the PCs in danger, so it’s the DM gambling with their lives, not them. You could get the same kind of kick (with maybe less schadenfreude) playing a solo game with simulated players.

Opting into a dungeon is like your character opting into Russian roulette – exciting and meaningful, in that you’re taking your life into your own hands. Being forced into a dungeon is like having your character forced into a game of Russian roulette – at best, a game you have little control over, taking a risk you haven’t chosen, and at worst, well, torture. The Deer Hunter is a tense film to watch, but it wouldn’t be fun to play through at the gaming table; “Roll this d6, and on a 1 you get shot, and if you don’t roll, you get shot” isn’t gambling, because it doesn’t involve a buy in – it’s essentially just execution by random number generator. Perhaps the best way to put it is that it has the tension of potential character death, but lacks the thrill of betting your character’s life.

Now, this isn’t to say megadungeon or focused dungeon-delving campaigns are dysfunctional – they just have to give the players the ability to buy in or choose not to in other ways. This is one reason it’s advisable to have multiple dungeon entrances (something I’ll cop to being particularly bad at remembering to do); if you don’t the players are basically forced to experience the first area of your (mega)dungeon, meaning any insurmountable obstacles, or perhaps a lucky crit resulting in an insta-kill, feel unfair – it’s not a risk the players chose to take. This is also why it’s a good idea to have safe havens in the megadungeon, perhaps maintained by friendly (or befriendable) factions – anything that means that when the players descend to the depths, or descend to these depths as opposed to those depths, it’s of their own volition. The point is that the forced, un-variegated dungeon lacks a buy-in, so when things go wrong, it’s not a result of the players’ risk-taking. Compare the megadungeon in Ultima: Underworld, which is mostly very carefully balanced, and which the player is stuck in for the whole game, to the many megadungeons of TES: Daggerfall, which are decidedly not balanced, but which you enter at your own risk.

The other thing that’s a bad idea – and I think many will agree with me on this one – according to the gambling formula is the linear dungeon, or linear encounter sequence, which is the same thing. When the players enter a dungeon that’s effectively a straight line, or a thinly-veiled set sequence of encounters, there’s no element, on the scene-to-scene level, of them having brought any bad consequences upon themselves. Now I have to be careful here, because technically the linear dungeon falls under the roulette model – you take a gamble at the point where you embark upon it. And that’s why I said “on the scene-to-scene level”; at the whole-dungeon level (in, I suppose, a different sense to whole-dungeon thinking), the dungeon represents one big gamble. That’s why these are often presented as gauntlets, where the gamble is to see if you can endure the whole thing, since presenting it in this way emphasises the choice to enter, and reminds you that you are here of your own free will (and could have chosen to wait until you were stronger to try and complete it, for instance).

So gauntlet dungeons and linear encounter sequences do actually work, provided they emphasise that initial choice point (and provide it in the first place). But their internal structure leaves much to be desired – they don’t satisfy on the “nested risk” front of playing a sub-game of taking smaller risks minute-to-minute, like blackjack does. In this, they’re a bit like a slot machine that takes several hours of focused concentration to return a result. Of course, it doesn’t help that this format is often used as a stretch of a larger railroad, but that’s beside the point here.

Still from Gauntlet

🕷

So to recap, a nested risk scenario offers several levels of investment. The players are essentially playing two games – one where they bet they can survive a foray into the dungeon, trusting to their luck to a certain degree, and a second where they do all the stuff we usually think of as constituting a game session, i.e.: making decisions such as which route to take and whether to fight or run, etc. Game 2 offers by far the most choices-per-minute, but Game 1 can be an important addition. One reason is that it’s a way of having your cake and eating it: You can provide a compelling challenge for the players’ skills, while at the same time maintaining the idea that the world doesn’t revolve around the player characters. The moment-to-moment gameplay is fun, but they could get in over their heads or get unlucky, which is exactly how you’d expect adventuring life to go. That possibility keeps it risky, and reinforces that the PCs are genuinely daring. A more modern style of rarely if ever including unbeatable situations feels, by comparison, somehow tame or safe.

But this is also important because it frees up what counts as (and perhaps more importantly, what feels like) a fair game. To a certain extent this is only possible within an old school game, in that it only works if the characters a) have genuine freedom to take on whatever challenges they choose, incentivised of course by suitable prospects of reward, and b) are live to the possibility that they might choose to take on a challenge they can’t overcome, or will otherwise get unlucky and end up as monster mulch.

If nested risk scenarios are important for old school play, they’re especially important from the point of view of the referee. That’s because they allow for the “no balanced encounters” ethos attributed to the OSR. If the referee controls which encounters the party faces, then it’s the ref who’s responsible for any mishaps that occur. If the party ends up in an unwinnable situation, the referee essentially pushed them towards it, so it’s their fault. Even in less extreme cases, the onus would seem to be on the ref to balance encounters for the party if they’ve determined which encounters the party will happen across – or, more minimally, that the party has to happen across some encounters, as when they’re placed at the entrance to a dungeon.

And that responsibility, of being the arbiter of balance and fairness, is a heavy burden. I think that one of the unspoken appeals of the OSR, that finds partial expression in slogans like “don’t focus on balance” and “divest yourself of their [the characters’] fate” (from Principia Apocrypha), is that the referee doesn’t have the burden of being responsible for making sure that each encounter or challenge is in the goldilocks zone of being tough enough to be fun but not too tough. The point is that the ref doesn’t have to think in terms of difficulty at all – their job is to populate the world with whatever they find interesting or cool. The players are then responsible for investigating and discovering all this stuff, getting into scrapes along the way, on the understanding that this process of rattling around the world may risk them getting in over their heads.

🕷

I think that grappling with nested risk as a concept is an attempt to show how this ethos of un-balance and seemingly unfair encounters can be consistent with fairness generally. I could go on to talk about the further caveats – how this works with west marches games, or the true old style of playing the same character at Gary’s table one week and Dave’s the next. I think it even carves out a niche for an interesting little idea of the “dungeon check” (one roll to see how you did exploring the dungeon as a whole, or on your first reconnaissance mission). But I feel like I’ve said enough (and probably more) for now. I’ll certainly be using these thoughts in designing my games, if only to help me “divest myself of their fate”. Until next time.