Power levels and scaling

[Greywander]

I want to design my own TTRPG, but there's one glaring issue that I'm having trouble resolving. And that issue is power levels, and how they influence attributes and core resolution mechanics.

Most games are built around a specific power level, and pushing the bounds of the system beyond that power level can lead to problems. A game designed for comic book super heroes probably isn't going to handle normal humans without super powers very well, for example. However, while a system is typically built around the PCs being a specific power level, it is usually inevitable that other power levels will be included in the rules, often as antagonists.

It gets more complicated, too, because there are actually two distinct "types" of power levels that we can talk about. First is creature power levels. A god is more powerful than a dragon, which is more powerful than a knight, which is more powerful than a cat, which is more powerful than an ant. This doesn't mean that a knight can't hope to defeat a dragon, only that the odds are stacked heavily against him. In the World of Darkness, there's a wide gulf in power between vampires and mortal humans, and yet vampire hunters still exist, they just have to employ specialized tools and tactics that prey on a vampire's innate weaknesses. Second is experience power levels. A legendary hero is more powerful than a greenhorn fresh out of Adventurer School, even though both are fundamentally human.

So in the end we're left with the question of how to model creatures that are innately stronger than others, and people who are more skilled or experienced than others, and we have to handle interactions between stronger and weaker creatures and between more skilled and less skilled people and define how skill relates to innate strength. Can a hero train so much that they simply become stronger than the dragon? Or will no amount of training ever bridge that gap? High fantasy says yes, low fantasy says no, so there isn't really a right answer, it all depends on what kind of game you want to play.

I'd like for my system to be generic and universal; you can run any kind of game you want in this system. This means that the system can't be built around one specific power level, but rather needs to be able to accommodate them all. The system should be just as capable of handling PCs that are gods as it is PCs that are ants, and should even be able to handle both in the same game, and allow them to meaningfully interact with one another.

My thinking is that the best way to accomplish this is to divide such power levels into tiers of a sort. One simple example of a tier type is size; a bigger creature is stronger and tougher than a smaller creature. Attributes can then modeled around whatever is the average for a typical creature of a given size. A giant with a Strength of 6 is not necessarily weaker than a human with a Strength of 15, rather the giant is just weak compared to other giants while the human is strong compared to other humans, but the size difference means the giant still gets enough of an advantage against the human. Another good candidate for tiers is intelligence; it always struck me as odd that in D&D you can have humans with as low as 3 INT, the same as most beasts. Likewise, it always seemed rather restrictive for beasts to be mostly relegated to an INT of 2 or 3, making it difficult to define one beast as being smarter or dumber than their kin. If humans and beasts exist in entirely different tiers of intelligence, then both of them can use the full range of possible attribute scores.

With all this in mind, perhaps instead of defining a few tiered qualities, such as size and intelligence, it might make more sense to define the core attributes and then allow each attribute to be scored separately on their own tier. Tiers can then scale infinitely up or down, allowing a creature of any power level to be defined. What I'm not certain of is how either the attribute or the tier should interact with the core resolution mechanic. I can pretend that tiers would only come into play in contested checks where the two contestants had different tiers, but even uncontested checks would have different DCs depending on the creature attempting it, so there should be a way to take a DC for one tier and convert it to a DC for a different tier, meaning that we could just as easily establish an absolute DC value for a task and then treat the tier as a modifier to the roll. Bleh.

As for skill, I think the best option is to define a finite number of skill ranks and set some of them to be within normal human (or otherwise) limits while ranks higher than that are more mythical (or anime, take your pick), allowing DMs to set their own cap depending on whether they want heroic high fantasy or gritty realism where anyone can die. I did kind of want some kind of infinite skill progression with diminishing returns, but there's issues with that that make it both difficult to implement as well as difficult to properly adjust the algorithms for how quickly diminishing returns set in and what practical limit a skill can reach. Perhaps I can meet in the middle with some kind of infinite recursion, e.g. a skill can be rank 5.4.3.2, which means that the skill is rank 5 against anyone that isn't rank 5, and against someone who is also rank 5 then it is subrank 4 (against the other person's subrank), and against someone who is also subrank 4 it is sub-subrank 3, etc. Basically, you can max your skill out, then keep growing by gaining subranks that determine if you are stronger or weaker than someone else who is also max rank, and so on.

Lastly, this brings me to the question of what the core resolution mechanic should be, and how attributes, tiers, and skills influence it. I know different methods of rolling dice and manipulating the roll with modifiers will produce wildly different outcomes that fit different styles of games. Then again, it's also true that almost anything "can" work, and it's mostly down to preference. d20 + modifiers vs. d100 roll under vs. 2d6 + modifiers vs. attribute die + skill die etc., each one will produce a result, and maybe it's not worth sweating over finding the "perfect" mechanic. For tiers, I think it makes sense that being a certain number of tiers above another creature makes it almost impossible to interact meaningfully, e.g. you don't need to roll to crush an ant, nor does the ant get a chance to roll to crush you since it is deemed impossible. If, for example, each tier up gives a +5 and the core resolution is a d20, then it becomes basically impossible to beat someone four or more tiers higher than you. But I'm not sure that's how I want to do it. Maybe something more like 5e's advantage mechanic, where you'd roll an extra die for every tier higher, and after a certain point there would be a cutoff where success is automatic.

I would spend more time refining this post, but I have to be somewhere so this will have to do.

[NichG]

There's a lot of places I could go with this, which would each be different systems, so there's not like one simple right answer...

But the one that happens to catch my fancy right now as I write is that you could have the conceit that 'skillfulness determines whether something happens or does not happen, whereas intrinsic attributes determine the impact of that which happens or does not happen'. So rather than the 'add your stat and skill' kind of system, its more like 'compare skill to determine if you have a chance to apply an effect, then use stat to determine the scale of the effect'.

So for example, lets say humans have a viable attribute range of Strength and Endurance in the 20-60 range, whereas for dragons its more like 500-2000. But dragons (here) are basically beasts, and do not gain in skill, whereas humans are able to become arbitrarily highly skilled. If a dragon attacks a knight, as the knight gets more and more skilled the chance of the dragon actually landing a hit sufficient to injure becomes smaller and smaller. But if the dragon does land a hit, they will almost certainly instantly kill the knight. Similarly, the knight may get close to a 100% chance of landing a hit on the dragon, but they need to land 10-20 hits to actually slay the dragon.

In a system like this, something like called shots or intentionally taking penalties to your skill usage in order to get some benefit would be important to prevent things from turning into a slog, but those should be soft-capped somehow. Like, maybe you can divide your effective skill level by 2 in order to take two actions in a round instead of one, by 3 to take 3 actions, by 4 to take 4 actions, and so on (though you would have to be careful to have a system where having 0 skill means success is impossible). So a very skilled knight could somewhat make up for the fact that they're still 20 times weaker than the dragon is hardy, but it would be very hard to scale that to something say 2000 times more hardy than the knight is strong.

One way to implement that organically would be to say that each skill is like a per-round pool of points you can spend. Specific tasks or actions then have a difficulty rating in dice (which could be something like 3d4 * 5, so you'd roll 3d4 and multiply the result by 5). You can attempt the action, but then you have to pay the result from your skill pool to succeed or otherwise fail or at least don't complete the action (you could even have multi-round actions this way, like casting a certain spell requires you to invest 3d4 * 500 skill into it). Once you've picked a certain course of action in a round, that's what you're doing actively (but you could do it multiple times if its attacks, casting a certain spell, whatever); however you can use any of your skill pools reactively if a situation demands it, such as paying to defend against attacks. I'm not sure exactly what would set the dice pool for attacking or defending in this case - maybe its just a fixed thing based on the weapon? This would get a bit tedious at high levels of skill though, as you might have to roll 20 times to use up your pool (maybe you roll once and pay N times the cost to act N times?).

[Greywander]

It sounds interesting, but I'm not sure that would be practical.

I could imagine generalizing this idea to skill checks where, in addition to the task's DC, the task also had "hit points", and you'd roll a skill check against the DC and deal "damage" to the task according to your attribute. Once the task runs out of HP, you would have completed it successfully. For contested checks, instead of rolling against a DC both creatures would roll opposed skill checks, and the winner would deal "damage" according to their offensive attribute to the loser's "HP", which depends on their defensive attribute, and once someone runs out of "HP" then the contest is finished.

What this essentially boils down to is a method of determining if a task should require multiple skill checks to pass or not. If you're trying to pick a lock with an HP of 15 and you have a DEX of 5 to 7, you'll need three successful checks in order to pick the lock, but only two if you have a DEX of 8, and only one if you can get your DEX to 15 or higher. Like I said, it's interesting, but making extra rolls will slow the game down. This does have some interesting potential for things like combat, though, where it will influence the number of actions/turns you have to spend to perform some kind of task, such as grappling or casting a spell.

I'm not sure I'd want to use this directly for combat, i.e. for attacking and damaging enemies. If we assume that two humans have roughly the same strength as they do toughness, then that means they're probably going down in one hit, maybe two if they put a few extra points in toughness or their opponent took a few points out of strength. This applies to pretty much every creature facing another creature with comparable stats, though some creatures will have a disproportionate amount of toughness compared to their strength. Furthermore, having more strength becomes pretty much irrelevant, so long as it exceeds your opponent's toughness, as they're going down in one hit regardless.

Again, it's interesting, but like a lot of interesting ideas I don't know that it works will in actual execution. It can be easy to get caught up in a cool mechanic and all the interesting things it can do that we neglect to consider what it will look like in actual play. I think there's something to this idea, though, and that you could refine it into something that would work well, but I'm not sure it's quite what I'm looking for. That said, I do like the asymmetry between attributes and skill (i.e. that they do entirely different things instead of both just adding together to create a skill modifier).

[Yakk]

First, YAGNI.

You ain't gonna need it. When designing something, it is really tempting to bolt on yet another feature. But YAGNI is universally true.

Second, a system built in someone's head isn't going to be all that good. So you *need* to actually play your system for it to get good.

And that means you have to optimize it around ... the actual game you are playing. And make that experience great.

The 1990s had piles of "universal systems". They all imposed their own implicit genre on the supposedly distinct games they supported.

Write an actual game that solves an actual play scenario you want to actually do, then use the game to do it.

And, I mean, scaling could be a fun problem to solve. But both solving scaling *and* being a universal game engine *and* making it fun is too many balls to juggle at once.

Create a concrete setting. Solve scaling by theorycrafting. Then actually play and work out how to make it more fun. Here, we are solving 1 novel problem at a time, not 3.

Then iterate. The actual play lessons about fun can cause you to replace your scaling mechanics with different ones.

...

I can talk about how various games handled various kinds of scaling. Everything from RIFTS to Starcraft 2 to every edition of D&D to video games to GURPS to 4X video games to Greg's Reign ORE game to RISK to Axis and Allies to a metric tonne of different heartbreaker theorycrafted games I've written.

But if your goal is to solve 4 novel difficult problems at once, ya just gonna fail mate.

[Anymage]

Everything Yakk said.

Also, the resolution mechanic can absolutely influence the sort of game you'll be playing. How skilled you can get vs. the variability of the dice as well as how well curved the probability distribution is will both influence the sort of actions players are likely to take. As will the inherent lethality of combat. Any system truly capable of emulating a wide variety of genres will either be excessively complex with multiple resolution mechanic options, or super simple and expect the group to self-narrate into the desired genre. Pick one thing to focus on. If you have a cool idea that might work elsewhere write it down (GW does have a bunch of interesting ideas), and if it doesn't work for system X it might be an interesting mechanic to play around with later for something else.