hmmm. I'm not convinced either way yet. Think you could come up with four concrete examples from four seperate game systems to support your points? For extra points, none of them can be D&D or SR of any iteration (yes, that includes all forms of D20).

Okay, I can do that. Games are fun, anyway.

Here's an example from Ars Magica, 3rd edition. At character creation time, you have 150 points to purchase levels of magical arts, using a triangle progression. (Basically, something like 2*art + d10 determines how powerful your spell is, and people resist with a number + d10.) They suggest a few ways to spend the 150 points.

• Specialist. 16 4 2 1. You are a world expert in your specialty. Sages ask you for advice, and woe unto anyone who tries to best your magical power in that area.
• Dualist. 12 11 3. Taken as a form/technique combo (magic is divided into both forms and techniques which are complementary), this provides awesome power in a narrow area.
• Concentrated. 10 10 8 2 1. A specialized character with a secondary area of competency and a couple of minor tricks.
• Widely adept. 7 7 7 7 7 4. Pretty good at a range of magic. You can hold your own in many situations.
• Generalist. 6 5 5 5 5 5 5 4 4 3 2 2 2 2 1. You know all magic, just not very well. You know more tricks than the proverbial Fox, but nothing you do is that potent.

To know how these fare against each other, one would need to know the system better, but the basic idea is that you get power from flexibility or power from brute force, but you don't get both with one build. For example, a 16 2 1 1 1 1 1 1 1 1 1 1 1 build is possible, and gives a good deal of power, but the flexibility is so puny as to be effectively useless. A flat cost system just can't do this. (See my d20 examples--same idea here, except you spread things out wider.)

In Nomine uses a test mechanic that is basically a target number that you must roll lower than with 2d6. Generally, you must roll lower than your attribute (a value that ranges from 1 to 12). As a celestial (these are powerful beings) gets 9 "forces" to divide between three areas each of which has two sub-categories: strength/agility, intelligence/precision, and will/perception. For each "force" point that you apply to a category, you get four points to devote to a sub-category. The template build assigns three forces to each area, and divides the resulting 12 points evenly to get stats of 6 6 6 6 6 6. All linear. Here's the trick, though. Even though the point allocation is linear, the benefit is not, due to the probability distribution of 2d6. If instead, 1d12 was used, a build like 12 6 6 4 4 4 would provide the amazing power of never failing on tasks related to one attribute with only a modest decrease in success rate in other areas. But with 2d6, a 12 is only slightly better than an 11, while success rates fall more than a factor of two from 6 to 4. So the limited number of points plus the nonlinearity in the test conspire to make balanced characters a good idea. Of course, if you gain the ability to have quite a few points to throw around, buying 12s at linear cost then becomes a very good idea since you don't have to worry so much about failures once you get everything critical up to around 6 or so (the difference between 6 and 7 is often not so critical, but the guarantee of going from 11 to 12 can make a big difference for frequently-performed tasks).

Earthdawn uses a complicated system of "steps" to determine how many dice you roll. The basic idea is that tests related to a stat increase the average roll of your dice using a rather unique and not exactly correct progression of dice rolls. The basic mechanic for success tests is to roll those dice, with a linearly increasing average, against a target number (kind of like the d20 system but with a lot of twists that avoid the stale problems of unreachable TNs and stuff--the ED dice explode like SR3 dice, for example). Well, now you have all the potential problems with the d20 system if you buy stats linearly. Their purchase system is really weird, where it costs one point to buy each level up to 11, two points to buy 11 to 12 or 12 to 13, three points until you reach 16, and then two points thereafter. The values are capped at 18 and 2. This system is so confusing that when playing ED, I never noticed whether it worked or was broken--the range of three points followed by two tended to lead to scores of either 18 or 11, IIRC. This is an example of a nonlinearity that I think adds a lot of complexity and doesn't benefit the game any. It was probably an attempt to keep people from having too many stats at 18 due to the linear advancement of dice rolled.

Torg uses an unusual variant of d20 where the die explodes both on 10 (which adds 10, and you roll again) and 20 (which adds 20 and you roll again). This is then converted to an approximately logarithmic system by means of a table called the "bonus chart" which ranges from -12 (for a roll of 1) to 0 (for a roll of 11 or 12) to 7 (for a roll of 20) to 11 (for a roll of 40)...rather weird and not uniformly distributed, but the idea was to generate a logarithmic distribution off of an exploding d20. They sort of did. Everything else is measured on a more accurate logarithmic scale, where there are five steps for each factor of ten. Attributes add linearly to this system--a skill of 11 allows you to accomplish things that are ten times harder than you can accomplish with a skill of 6. Since tasks are converted to a log scale, and your abilities are linear, you get exponential advancement with each increase in skill. Therefore, amusingly, you're limited to 13 skill points for all your skills at creation time, and can't apply more than three to any one area, and attributes range from 7 to 14 and must be taken from pre-generated templates (most are around 11, so that after rolling d20 and using the table, you get scores from 0 to 18 or so, representing an astonishing 4000-fold range in outcomes). Torg is designed this way (kind of weirdly) because it's supposed to be an ultra-generic system (requiring decent performance over an incredible range of situations) and tends towards a superhero feel during gameplay. The core book doesn't even have advancement rules. There's no way at all do this with a linear system. With the logarithmic ranking of difficulty and linear skills, it sort of works. I wouldn't use the Torg rules for most games--really, an infant can out-wrestle an olympic lifting champion about one time in 400--but if the goal was universality, they got it. (And then didn't use it to its full potential, sadly enough.) The cool thing about the system that they use is that it does work just as well for superheros throwing aircraft carriers into an enemy city to destroy it, as for a cricket wrestling with a spider. If you spend the time to learn the rules, and want sorta passable rules to use in any imaginable situation, they fit the bill.

Anyway, all these systems use nonlinearities, with varying degrees of success. In almost every case, one can figure out why they did so, and in some cases there doesn't seem to be a linear method that isn't badly broken. I'm not familiar with any rigorously linear games (I tend to get into rulebooks that start seeming overly linear, identify an absolutely critical flaw, and then not buy the book...so I don't even know how linear of a system exists). I certainly wouldn't use all these systems for the same setting (they're setting-specific), but they each do something that one can't do with a linear system. And in some cases (e.g. Ars Magica) the nonlinearity is used specifically to generate tradeoffs between flexibility and specialization that one can't get with linearity.

A linear relationship can be expressed by the relationship a*x + d*y = c. Well, okay, that's affine. But normally we call this linear, and typically write it as y = m*x + b.

If you raise x by one point, you should raise y by m points. That's it. That's linear.

If you raise x by one point, and y doubles that's not linear. (This is exponential.)

If you raise x by y points, and y goes up one point, that's not linear either. (This is a triangle progression.)

In ED, the dice system actually is linear in step. The variance fluctuates a fair bit, but the mean is linear. It's linear and complicated. (Well, it was supposed to be linear; the way that the dice explode makes it slightly nonlinear; maybe they did the math wrong.)

And if you broke In Nomine with its mild nonlinearity (granted, it's not that hard to do), it'd be even more broken if the system were linear, because the slight weaknesses you might have had would've been even more slight.

I think I'm done with this conversation. I don't have enough confidence that we're even talking about the same thing to make it worth continuing. I've said multiple times already that I want systems to be as simple as will work. But you keep referring to broken systems that are broken because they are linear and would be less broken if they weren't linear and saying that you like that kind of simplicity because it's easy to understand, and anyway, anything can be broken (without providing examples of breaking something a little more complicated).

Well, I like systems that don't break so easily. Sometimes that means using somewhat more complicated mathematics, like it costing X+1 to advance a skill to X+1 from X.

This is called a triangle progression because the total cost looks like this:
# < - skill 1
## < - raise to 2
### < - and 3
####
#####
######
#######
See the triangle?

The total cost to raise to skill n from zero is n(n+1)/2. SR3's extra steps depending on attribute complicate matters without adding much balance, IMO. But the basic cost is still more complicated than "5 karma buys another skill point".

With 18 18 6, you have +4 to hit, and a -1 to armor class.

With 14 14 14, you have +2 to hit, and +2 to armor class.

Net effect: Mr. 18 18 6 has a +2 net bonus to hit Mr. 14 14 14.
But Mr. 14 14 14 has a +3 net bonus to hit Mr. 18 18 6.

So Mr. 14 14 14 actually lands a hit 10% more often than Mr. 18 18 6. (Mr. 18 18 6 does, however, do quite a bit more damage, and has more hps, so he does come out on top, but not by as much as he would if he didn't get hit so much because of his low dex.)

No matter what gaming system you use. There WILL always be a optimal build of a character. It is impossible to get away from that. The key is making or choosing a system where changes to a character "work". So that if you choose poorly in designing your character it will be under powered. So that the game system means something. But at the other end of the spectrum, making 1 or 2 small changes to your character from the optimum build does not have a real hit in power.

Its a balance

Well, you have very different players than I do. I recommend GURPS to your group--you can have all those character types coexisting and more.

I'd thought that the difficulty was min/maxing, but clearly it's not. The difficulty is that people think Wolverine is cool, and therefore they have to have regeneration, adaman...er...orichalcum bone lacing, and dikoted cyberspurs (never mind the mix of impossible and useless things they're putting together). Oh, and since he's Wolverine, he has to be badass, too. I have, rarely, encountered this type of player, and here I agree: no amount of carefully tuned rules (or even bold text in caps in the rule book saying specifically that they can't do something) will really dissuade such a player. That's up to the GM, because the player isn't playing Shadowrun. He's playing X-Men.

cain, i really fail to see how your argument applies to linear vs. non-linear creation/advancement systems. someone who wants to play wolverine will attempt to play wolverine whether the system is linear, non-linear, or completely diceless. having a linear system with hard caps isn't going to stop them any more than having a non-linear system with soft caps will. it won't even change the outcome.

That's rather the point. No matter how difficult you make it, they're going to try and break the system at any cost. So, for the rest of us, we might as well have a simple, straightforward, and easy system.

Since the munchkins will ignore any limits you set-- and good players will tend to follow them no matter what-- there's really no need for a complicated system, is there?

except for those of us who find that a certain amount of complexity allows for finer control of the character's statistics, making it easier for us to create character sheets which match our mental image of the character we have in mind. i mean, seriously, by your argument--and i'm not trying to be insulting at all, i really mean this--why not just play tri-stat, or something similarly rule-light? it's not like you'd have a lot of work doing conversions, or anything.

edit: non-linear creation/advancement also makes characters with high stats special, something to be proud of, because high stats are hard to get. this is a roleplaying game; there's nothing wrong with deriving pleasure from the fact that your character's a badass, especially when you have to work very hard to get there. with a linear system, high stats are run-of-the-mill. everybody's got 'em, who cares.

you assume way too much GM tinkering, Cain. let's just acknowledge that the GM can change the game any way he likes, and then ignore it for the moment because we're discussing systems, not game masters. yes, the GM can just hand out more points. but the system has a default number of points to hand out, and a default averagish rate of karma awards. if a GM hands out more BP and/or karma than is recommended, he cannot legitimately complain that the system is broken when his players build more powerful characters than they'd have otherwise been able to do.

as for Tri-Stat v GURPS, i really don't know what to say. judging from that statement, the distance between our different understandings of gaming and characters is so vast as to be measured only in AUs.

I understand what you're trying to say; but I've discovered that the simpler systems allows for better fine-tuning than exceedingly complex ones. Tri Stat, being purely effect-based, seems to be better than GURPS for this sort of thing-- GURPS being mechanic-based. (Every GURPS sourcebook comes out with a ton of new special-case rules for a particular situation; the Tri-Stat books tend to not need as much.)

At any event, I've discovered that quicker and simpler systems make for easier and smoother gameplay. The SR3 rules, for example-- if you strip out the major problems, such as Rigging, Decking, and Open Tests, it becomes a perfectly simple and straightforward system, easy to run and smooth in play.

i agree that overcomplexity is bad. that's why i gave up on the rigger rules after several attempts. the decking rules i can handle and enjoy, but i recognize that i'm one of a very few, and am not opposed to reduction in complexity there.

you may have discovered that simpler systems allow for better fine-tuning for you. i have discovered that more complex systems allow me much more freedom. i honestly can't imagine the style of game you run/prefer to play in. when i want to create a character whose statistics reflect the character i've imagined, it's easier if i've got lots and lots of options to work with. more options, if they're balanced and realistic, generally means more complexity.

and as well, there's the fact that i enjoy mastering a system. for me, the mechanics are not just something i use as a vehicle for roleplaying, they're a game. i like deep roleplaying, don't get me wrong--but i also like winning games. it's easy to "win" a straightforward game, because challenges progress linearly. if the opponent is X levels (or whatever) higher than you, you can't beat him; if he's Y levels (or whatever) lower than you, you can't lose. you can skew that by introducing critical hits and the like, but that just means that the game requires more luck than it otherwise would. with a complex game, though, challenges are much more non-linear. you have to think and plan in order to win, and you can take on opponents that would normally be out of your league if you're careful enough.

okay. at this point, i'm just going to have to say "nuh-uh" and leave it at that. i've never had any of the problems you've mentioned, the time it takes me to generate characters is far shorter, and i've never seen anything in a simple game like what you describe. i really can't imagine how you game, and i think that's where 99% of the differences between our view lie. we apparently do very different things with the systems we use.