• The Infractions Forum is available for public view. Please note that if you have been suspended you will need to open a private/incognito browser window to view it.

Down the rabbit hole: design for high crunch, high complexity

Ormologun

Registered User
Validated User
I'm trying to find a home for this unpopular, kind of grognard discussion that REALLY goes against the grain. So I need a forum large enough to find someone interested in this kind of dialogue, and at the same time very niche and specific. The plan here is long term.

Disclaimer: no product is being made. While this long term project has its ideal form, the goal here is purely the fun and passion about messing with the rules and delving into this type of nitty gritty game design. It's an experiment. In the remote possibility it will coalesce into something more solid, it will all stay open and free.

The idea begins with a project for an ASCII roguelike computer game that is built on top of an explicit pen and paper-style ruleset. Being on a computer doesn't affect the development of the system outside a core area: bookkeeping. The primary problem with very crunchy systems is bookkeeping, in this case being built as a computer game (eventually) simply allows me to get away with more complexity in a few areas, and so I can dare going a step further than would otherwise be tolerable (for example right now the initiative system I'm using is built on top of the relatively complex "phases" coming from the "Combat & Tactics" supplement of AD&D 2nd edition).

The goal is a misshapen, monstrous HYBRID where I hammer together pretty much everything that looks interesting, with particular attention to the feel of the old-school in the 70s and 80s. Not the simplified minimalism on the OSR side, but instead the grognard wargame tradition that was the true origin and that faded almost completely away, overshadowed by the narrative, rules-light systems (I appreciate these too, I just have opposite goals here).

I've got already a fairly long list of systems that I'm going to analyze and, eventually, incorporate in the Frankenstein monster of an hybrid. The ruleset was meant to be structured around Harnmaster (percentile dice & skills, no classes, no levels). From there, I intend to warp and incorporate rules from:

- The Riddle of the Steel (and its wicked spawn of three: Blade of the Iron Throne, Band of Bastards/Sword & Scoundrel, Song of Swords)
- Sword's Path Glory / Phoenix Command (OH YES)
- Chivalry & Sorcery
- Swordbearer
- Ysgarth
- RuneQuest 6
- Dangerous Journey
- RoleMaster (2nd & supplements)
- The Fantasy Trip (Advanced Melee)
- Zweihander (whenever I figure out what's the difference with WHFRP)

(sometimes I also find inspiration from systems that seem distant from the same concept. It happened recently with "Fading Suns 2e")

But I'll also try to learn from stuff you don't expect to see here, like Gloomhaven, Mage Knight, and Kingdom Death Monster. (but this is also part of a side project that goes in a different direction) Or that removed part of history that produced a very particular flavor of extinct games, as a broad category of which "Valkenburg Castle" is a good example.

...What I'd like to do, here or elsewhere, is discuss the design of those games, imagine new mechanics that spawn from that analysis. Experiment at all levels. And forgive me because my math skills aren't good and neither I'm an expert of this type of design. A lot of this WILL be blunder, but I do it because I enjoy the PROCESS. And maybe out there I can find someone who enjoys this process as well and help me with some ideas.

So I'll try to do two different things. On one side I'll read and analyze very old and sometime obscure systems, down in their quirks. And on the other side take the inspiration to transform and twist those rules into something else.

But because this is a twisted process, and I'm already well into it... I'll start with something more unusual. I'll take two new "indie" systems I just found out, freely available, and discuss what I don't quite understand after skimming the rules. The two systems are:
- Knights of the Black Lily
- Against the Darkmaster

I'll try to keep it relatively short as the introduction was already kind of meandering.

The first system got my attention because it was publicized as "bell-curved d100 as a core mechanic". And... the d100 isn't bell-curved. Lately I've spent some time to study linear systems versus bell-curves, and I considered switching the Harnmaster foundation on the d100 with something entirely different. So this piqued immediately my curiosity because maybe they tried to do something that was new to me.

I'm not sure that this is the case, though. In my previous analysis I figured out that there's one aspect of a bell-curved system that can easily be mapped onto a linear system: skill progression. You can create custom rules to "pace" the progression any way you want. So you can "map" the shape of a bell curve on top of a d100, simply by adjusting the "steps", or the cost of every step. In fact, most systems already do this. For example Elric! (Harnmaster too? Can't remember) lets you improve the skill only when you roll OVER it. So the higher the skill, the rarer the improvement. It's not a bell curve, but is the bell curve really a good choice to map skill progress?

That spawned an analysis of what a bell curve "does" (I eventually want to study dice pools mechanics too, and once for all write an ideal, complete conclusion to that debate. I've already done one for THAC0 that I'm happy with). In general it seems the main feature, the one that is relevant, is the "skill vs randomness". In the longer term, if the dragon has a million hit points, it doesn't matter if your sword hits for 1d12 or 2d6, because the damage will settle around the middle point anyway. But within a single combat, the individual swings of a sword, the linear system produces more "swingy" results, whereas the bell curve is somewhat more realistic and more predictable. Well behaved (unless you attach to it exploding dice, like Earthdawn).

All this to say: the bell curve has its use, but its use seems to be a case of performance. A skill progression system is not directly part of this, because you can map progression in any shape and curve you prefer. The "bell-curve VS linear" theme instead is built as a DICE mechanic. The choice is about how you roll those dice and the shape of randomness in your game.

Am I wrong?

That's why the choice in Knights of the Black Lily started to not make a lot of sense to me... What is going on here was already clear by glancing at a table, and noticing this table was exactly the same for every pregen character.

As you can see (below), what they have done is to chunk the 100 steps of a % skill into 10 tiers, and then modifying the SIZE of each step, to conform to a bell curve. So, going from tier 0 to tier 1 you gain +5%, but going from tier 5 to tier 6 you gain +9%. A fairly shallow curve, actually. There is something non trivial going on with critical rolls, but the distinction between success and failure is fixed at a certain %. That means that the moment you roll to make an attack with a sword the system is structurally identical to Harnmaster. The actual performance is just as "swingy".

Thankfully we also get a design note:



This is where I understand even less. It says the strength of a d100 is granularity, but wasn't that granularity removed by merging the system into 10 fixed tiers (and chopping the extremities)? If anything it's even less granular than D20. The results are on a matrix of four possible outcomes. It doesn't seem the system uses that granularity in any meaningful way?

But indeed there is more to it. Because they explain that the "modifiers" to the roll are tiered in the exact same way, so you move up and down the same table, and in this case this bell curve has an impact even in the performance.

I'm now trying to figure out if the implications are more substantial than how they appear. Because when it comes to combat it's the result of the attacks that matters directly, and all this juggling of tables and U-curves behind the scenes doesn't seem to produce an outcome that is substantially distinct from Harnmaster, for example. Let's consider the basic resolution matrix of both games:





As it can be seen in the first chart the combat "matrix" is still built by four columns, and they are functionally identical to Harnmaster:
- Let's say your "sword" skill is at 59% (to fall in the 6th tier for the table of the first game). The concrete possible outcome is the same as in D&D: there's a binary possibility about a hit or miss, and then two other cases to cover a fumble or a critical hit (rolling a natural 1 or 20 on the d20). In a d100 system you have your more granular skill values, and a more varied distribution of those criticals, but it works in the exact same way. You usually just have to roll under the target number to succeed, instead of over.

If you look at the numbers in that 6th column of the TRC table you can see it mirrors the Harnmaster behavior. The skill value is 59%, so you roll 1-59 it's a success, roll 60-100 it's failure. The chart on the side isn't that different either, it's just a different presentation. If both roll a success (2) the outcome would be 2 - 2 = 0, so a stand-off/block. If the attacker succeeds, then in Harnmaster the defender can either successfully block or take damage. In KotBL instead the defender could roll a +3, and end up with a -1, opening the way for a counter-attack. But all these cases are just arbitrary design choices about the flow of the combat, they aren't BUILT on a different mechanic. In both these games we have a matrix of four cases, and in KotBL it's just a little less explicit, but you could rebuild it exactly as it is on the Harnmaster table.

So my question is: do we really have different mechanics at work here, or just the appearance? What is KotBL *really* trying to do here with its bell curve and d100?

Let's say there's a duel between a character at tier 2 and another at tier 8. The bell curve makes it so if both are wounded, the penalties are going to be bound to those tiers. You cannot lose half a tier, or get a -15% to the roll. Instead the actual numbers would go 25%->18% (-7) for the first character, and 75%->67% (-8) for the second. Since the bell curve is similar, they roughly receive a similar penalty, but if a character was at tier 5 instead, he would go 50%->41% (-9). What I'm trying to say is that the force of this bell curve seems rather trivial...

If we go back to the concept of the bell curve, its strengths are on one side the progress of a skill. But that I've said can easily be mapped on the d100 any way you want. And on the other, the performance, in that "skill vs randomness" debate. The less swingy results. In KotBL the only active part of the bell curve, once combat begins, is the application of the modifiers through the skill tiers. But are they contributing meaningfully? This is my question. What is this thing doing? If I reset all the numbers so that all the 10 tiers go linearly from 1 to 10, then 11 to 20, and so on, are the results of a fight really going to be different in a tangible way?

I'd also contend the other claim of that design note. Is this system really preserving the granularity of a d100? The strict 10 tiers seem quite the opposite, especially because this systems puts the human average in the middle, so that low values will be really rare in a game, and the 10 tier is considered beyond human. So I suppose most fights will happen within the 4 to 9? That's six steps practically available. Is there any concrete need to map this on a d100? Because it might be that even a d20 offers more than enough space to accommodate 10 tiers. And at that point if you really like what a bell curve does you could as well go to GURPS 3d6...

I'll keep the discussion of the other ruleset for later, but it's interesting because in that case we also have skill tiers, but more than 30 available, and built so that they just produce an increasingly slower progression the higher you go. No bell curve.

Whenever I read and criticize a ruleset it doesn't stop there. Is it possible to produce a better solution? Is there a way to retain the mechanical advantages of a bell curve, those that truly matter, and port them over to a d100 system? This question is also open.

--

All this as an example of the type of discussion I create with myself and maybe hopefully I will eventually find someone else to help me along. Because the whole point is that I'm no design expert, my mathematical competency is rather bad, and so no matter how much I write, and deep I delve, I often end up with catastrophic blind spots and make big mistakes. I often find out, much later, that I entirely missed the point, and a system I thought was bad was instead really clever and well built. So I go for another spin of the wheel...

And so, I'm searching for someone who enjoys these types of design discussions, so that I can bounce back and forth some ideas...
 

LuciusAlexander

PalindromedaryRider
Validated User
All this to say: the bell curve has its use, but its use seems to be a case of performance. A skill progression system is not directly part of this, because you can map progression in any shape and curve you prefer. The "bell-curve VS linear" theme instead is built as a DICE mechanic.
.
If I understand what you're saying here, I can't quite agree with you that "performance" is divorced from "progression."

It's the nature of a bell curve that not all +1s are created equal. If your skill is such that you succeed on an 10 or less on 3d6 and then with experience you buy a +1 to that skill, that's a pretty big step. If your skill is at 16 or less and you buy a +1 that is only a tiny step. Now yes, theoretically you could "smooth that out" by creating a complex formula or table that changes how much XP you pay for each +1 depending on how good you already are, but I don't think that would be easy.

Lucius Alexander

The palindromedary says I probably didn't even understand you in the first place and am just shooting my mouth off.
 

Ormologun

Registered User
Validated User
It's the nature of a bell curve that not all +1s are created equal. If your skill is such that you succeed on an 10 or less on 3d6 and then with experience you buy a +1 to that skill, that's a pretty big step. If your skill is at 16 or less and you buy a +1 that is only a tiny step. Now yes, theoretically you could "smooth that out" by creating a complex formula or table that changes how much XP you pay for each +1 depending on how good you already are, but I don't think that would be easy.
I'm not sure.

Let's take the d20, since it works pretty much the same as the d100, just with 5% steps. Just because the d20 is linear doesn't mean that character progress is going to be linear. Your attack bonus, or TACH0, grows by looking at a table (generally), and that table is usually class-based. And maybe the Fighter class progresses faster than a Wizard class, when it comes to swing in melee. There is nothing mandatory that links skill progression to actions performed. You can link them, but it's a choice, not a feature.

When it comes to skill progression (or level progression) we are dealing about what happens from one adventure to the next. So this mechanic of progress works outside the game of dice rolling, like in the middle of combat.

For example nothing stops me from taking D&D and make the attack bonus grow by following a bell curve. If that's the "shape" of the progression I want, then I can have it. I just write down the numbers on that table through a bell curved algorithm (another theme I'd like to discuss is "tables VS formulas"). That's exactly what they've done in that ruleset I examined.

If the game is based on tests done with 3d6, then you have a linear progression (+1, +2, +3 ...) but that becomes bell-shaped when you roll the dice, as you say, because every +1 has a different weight during the performance of dice rolling. Instead this other system pre-calculates that aspect. The dice rolling is linear (like a d20 or d100), but the progression is bell-shaped, so you grow faster at the beginning, speed up in the middle, and slow down toward the latter part.

My point is that the ideal "shape" of skill progress isn't necessarily bell-curved because that's an area of game design where the bell-curve isn't showing its strengths. And I think instead the strengths of the bell-curve, compared to linear systems, is on the more realistic, skill-based, less swingy outcomes (this will lead to a large discussion later on... maybe).

(for example: an idea I'm currently juggling is that in skill-based games the basic statistics, like Strength, Dexterity and so on, become almost irrelevant because once you produce the skill values that's what you work with. The relevance of the statistics gets pushed in the background. I wanted to tweak the system so that the statistics pull more weight through the whole game, but without directly crippling a character. The idea was to structure the skill progress on the basis of 3-5 stats that the skill relies on. There would be a formula, for example let's say a Sword skill is based 60% Dex, 25% Strength, 15% Perception, in order to progress with that skill you'd have to buy "tiers" whose cost is based on the stats values. It means that the statistics represent your natural "talents", if you are gifted with Dexterity then your "Bow" skill tiers would be cheaper to acquire compared to another character with low Dex. You'd progress proportionately faster. At the same time a character that isn't gifted could still eventually acquire high skills, as long he focuses and pays the higher cost. It would be based on choice. You could develop the character toward its talents, and so progress faster. Or go against the grain and develop it in the direction you want, but needing more hard work and time.)
 

tunglashr

Registered User
Validated User
"If the game is based on tests done with 3d6, then you have a linear progression (+1, +2, +3 ...) but that becomes bell-shaped when you roll the dice, as you say, because every +1 has a different weight during the performance of dice rolling. Instead this other system pre-calculates that aspect. The dice rolling is linear (like a d20 or d100), but the progression is bell-shaped, so you grow faster at the beginning, speed up in the middle, and slow down toward the latter part. "

I think you are conflating things here that are not equivalent. The linear progression is in the skill levels, but that is not related in a mechanical way to the results the resolution system produces. The skill levels are to make the progression human readable. The resolution is strictly a bell curve. The way we write them down on a character sheet is completely separate.

"Let's take the d20, since it works pretty much the same as the d100, just with 5% steps. Just because the d20 is linear doesn't mean that character progress is going to be linear. Your attack bonus, or TACH0, grows by looking at a table (generally), and that table is usually class-based. And maybe the Fighter class progresses faster than a Wizard class, when it comes to swing in melee. "

Again, I think you have missed the point here. Every bonus is in a 5% chunk. Its linear. Just because the wizard and fighter dont get the same linear bonus, its still linear. There is a difference between identical growth and linear/non-linear growth.

"My point is that the ideal "shape" of skill progress isn't necessarily bell-curved because that's an area of game design where the bell-curve isn't showing its strengths. "

Bell curved progress works fantastically for skills, especially when they are compared against variable difficulty numbers. Take Hero for example. If you have a standard 13- skill roll, you know about how well that works. Adding another +1 doesnt have much effect for that guy, normally, at least not as much as it does for the guy who has 11-. However, what it does is makes taking penalties more palatable. The 11- guy loses more from a penalty than the 13- guy does, when the magnitude of the penalty is the same. The bell curve makes all of this simple. There is a point of diminishing returns when buying more bonuses, which is great from a character design perspective. But also, a hazard hinders the novice more than the expert, and this works right out of the box with no other adornment necessary.

As an aside, I am also working on a game system that is pretty complex (to some, it is actually just a series of simple things), and I have found resistance from a lot of online reviewers. My best feedback has come here, on rpg.net. I took one of your ideas, that you can shove in all that you want to make the best results, but instead of doing a computer game, I explored an analog method of automation: information dense cards. Many people find them intimidating in print, but when people try them, they find them simple enough.

If you know of another place where people discuss simulationist type system design, let me know. I need that audience as well.
 

Ormologun

Registered User
Validated User
Again, I think you have missed the point here. Every bonus is in a 5% chunk. Its linear.
Nope, it's linear if you make it linear. Those 5% chunks are just what you work with. You can make it so you gain a +1 every three levels, then in the middle it might go to +3 every level (15%), and then it would slow again. This produces a bell curve applied to the progression.

It's just that this type of progress on its own doesn't seem meaningful.

Let's say you want to map 3d6 onto a d20. 3-18 are 16 values. On the d20 you narrow it down to the closest value, so you go:
1-3 = 1,
4 = 2,
5 = 3,
6 = 5,
7 = 7,
8 = 10,
9 = 12 ...
On the 3d6 the progression goes: 3, 4, 5, 6, 7 ...
On the d20 the progression goes: 1, 1, 1, 2, 3, 5, 7, 10, 12 ...

So we obtain the 3d6 type of skill *progress* on the linear d20. But in this case we cut out both the performance of the single roll, and also the modifiers, since you'd have to apply the modifiers to the tiers.

There is a point of diminishing returns when buying more bonuses, which is great from a character design perspective. But also, a hazard hinders the novice more than the expert, and this works right out of the box with no other adornment necessary.
This is a side-feature of the bell curve that matters in gameplay, so it's interesting to consider. The skilled guy gets diminished returns, but is also more resilient to modifiers (though it also applies to bonuses).

But this effect is not bound to the bell curve.

That's what I implied with the "Against the Darkmaster" rules. Here we have 30+ skill tiers. Instead of using a bell curve, here the ranks are built so:
- First 10 ranks: +5% each. (up to 50%)
- 11-20: +2% each. (up to 70%)
- 21+: +1% each.

If you get a -3 rank penalty to the roll, and you are rank 10 (50%), you lose 15%. If you are rank 30 (80%), you lose 3%.

You can then juggle the numbers how you see fit. It's an even better example: a novice is greatly hindered here, and wasn't in your system, because a hindered novice at the beginning of the bell curve sees the same minimal impact as the expert. And that's not "realistic".

But I wonder, if you go to such lengths to map the bell curve on a linear system, why you don't simply adopt the bell curve entirely... And I can see (maybe) what KotBL was trying to do. GURPS 3d6 has its broad range of 10 central values that have at least 10% or so. 3 to 5, and 16 to 18 are so small to be almost irrelevant. So this other system clamped the values to have 10 broader steps in the middle, all relatively meaningful.. But still 10 of them. The granularity of GURPS is even slightly better.

As an aside, I am also working on a game system that is pretty complex (to some, it is actually just a series of simple things), and I have found resistance from a lot of online reviewers. My best feedback has come here, on rpg.net. I took one of your ideas, that you can shove in all that you want to make the best results, but instead of doing a computer game, I explored an analog method of automation: information dense cards. Many people find them intimidating in print, but when people try them, they find them simple enough.
There are probably many different ways. My plan is to complicate things without restraint, as long I try to achieve some goals of what the system can model. Then, after it's complete, I work to simplify and present it. But I'm still too deep rewriting core ideas.

I think I've understood the good things about the bell curves, but I've still to study dice pools and see what I can take out of that.
 

Arioch

Registered User
Validated User
Alexander Kalinowski Alexander Kalinowski , the author of Knights of the Black Lily is here on the forums, I think he'll be happy to help you understand what he means in that passage!

And if you have questions about Against the Darkmaster, just ask away! ;)
 

SignoreDellaGuerra

Audii alteram partem
Validated User
Interesting discussion.

I ve went trough various iteration of this kind of complex problem, and event implemented an hybrid program with Harnamster resolution matrix, criss-cross a Rolemaster damage table, both recoded from scratch to address some problems.

Unfortunatelu the OP is way too long to read trough.
 

Aganauton

Registered User
Validated User
I apologize, I wanted to answer earlier but I was dealing with a sick dog.

I love the idea of 'crunching it up', even just as a thought exercise. So, if you were looking for another voice to the discussion, count me in. I love crunch. Although I not familiar with some of the systems you mention, some I am, notably, Rolemaster, and Harn. Loved both systems back in the day, but unfortunately had a tough time finding players to share the love and I cut my RPG teeth on D&D blue book.

I may have missed the statement in you're OP, but is there a reason you appear to be married to traditional dice mechanics? Don't get me wrong, the bell curve is a beautiful thing, and your ideas to play with it are intriguing, but again, I wonder if it's necessary. If one were to design a crunchy game in this day and age, with smart phones, tablets, laptops and desktops, why do we really need dice at the table (aside from the fun of rolling)?

I do really like the idea of using a characters stats to determine how much it costs a player to skill up the character. That is brilliant. Sadly, it only makes me wonder how to incorporate some other ideas I've had regarding the relationship between skills and stats.

I look forward to seeing more from this thread.
 

kingSpaceLizard

Registered User
Validated User
O Ormologun , if you wait long enough in these forums you'll find people who think the same way ;) I'm doing basically what you are thinking about but I am writing the software for it. My goal, like yours, is to hide the crunchiness away within a software layer. I've been spending some time on modelling this in Excel while considering how to handle increases in skill and how existing skill and attributes affect that. Quantity of time spent important but so is quality of time spent.

I also love the d100 games and a number of 0-100 scores is the outcome of my process. My focus is really on the character generation side, with a system of packages representing periods of time that are translated into increases or decreases in various metrics including natural attributes, skills, relationships and reputation, wealth and so on. The packages provides a timepath for the character career which hides nitty-gritty choices while allowing a player to create a character they want.

Keen to see where this discussion goes.
 

Pladoh's Ghost

Registered User
Validated User
I'm trying to find a home for this unpopular, kind of grognard discussion that REALLY goes against the grain. So I need a forum large enough to find someone interested in this kind of dialogue, and at the same time very niche and specific. ...
Did somebody say "grognard?"

Not the simplified minimalism on the OSR side, but instead the grognard wargame tradition that was the true origin and that faded almost completely away...
I wouldn't say OSR is just the retroclones and minimalist designs.

I've got already a fairly long list of systems that I'm going to analyze and, eventually, incorporate in the Frankenstein monster of an hybrid. The ruleset was meant to be structured around Harnmaster (percentile dice & skills, no classes, no levels). From there, I intend to warp and incorporate rules from:

- The Riddle of the Steel (and its wicked spawn of three: Blade of the Iron Throne, Band of Bastards/Sword & Scoundrel, Song of Swords)
- Sword's Path Glory / Phoenix Command (OH YES)
- Chivalry & Sorcery
- Swordbearer
- Ysgarth
- RuneQuest 6
- Dangerous Journey
- RoleMaster (2nd & supplements)
- The Fantasy Trip (Advanced Melee)
- Zweihander (whenever I figure out what's the difference with WHFRP)
Now you're talking dirty...I love it when you talk dirty!

I still have my SP:G, C&S, and DJ rules. The others have gone by the wayside over the years. This list includes some of my absolute favorites--and games that have heavily influenced me. As I came to RPGs from miniatures play and board wargames--with Squad Leader a favorite--then crunchy games interest me a great deal.
[/QUOTE]

As to plotting bell curves over percentile rolls, I find that it's often done. Most any system that involves a table to plot results seems to have a curve overlay of some sort. A very low roll will be a better result than a very high roll, and the range of each of those will be much narrower than the range of rolls that result in less extreme results found on the rest of the table. The question I always have when confronted with such is whether the results ranges make sense to me, whether they provide the degree of simulation I prefer.

The TRC table you post does put me off because the modifiers also follow a curve. What, then, is precisely the point of using the d100? It seems that all that's left is the ease of seeing the chance of success as a percentile chance, which is easier for most players to wrap their heads around than any system where they have to convert to percentile. If all of the chances are formed to a curve and all of the modifiers also conform to the curve, then why not use a mechanism that naturally conforms to the curve and be done with it?

Now, if the rules supported the GM *also* being able to assign flat modifiers in some instances--and discussed the sorts of situations where those would apply--then I can see keeping the percentile dice mechanism. Short of that, though, and I just don't find it elegant enough to use.

I suppose the curve they wanted to model wasn't obtainable with a particular combination of polyhedrals, though, (I've not checked) and this workaround was the only way they figured they could get that particular curve of results.
 
Top Bottom