Chess is not an RPG: The Illusion of Game Balance

[Bedrockgames]

I did the math a long time ago and it moves the average from 3d6's 10.5 to about 12.something. I wrote a program to compute it on a Compaq hand held running Windows CE 2.0. good times.

It was my impression that few people played D&D with straight 3d6, as it was just too swingy. Except for really die-hard old-timers.


My point was really to Celebrim's that nobody actually likes random generation. It's been working for us for over 20 years. I don't hear that much complaining really.

Keep in mind 3d6 is the method used in 2E, but in 1E the default was 4d6 drop the lowest.

 

[Celebrim]

We always use 4d6, keep the best 3. Its worked for us.

I played that way for a long time myself. But, I should point out that even choosing to play that way is itself the beginning of creeping away from random and toward 'above average'. Why are you choosing 4d6 keep the best three (probably rearranging to taste)? Why not if you really like randomness and diversity go straight up 3d6 in order?

If IIRC, the average on 4d6 keep three goes up from like 10.5 to 12.7, but the really big thing is just how much less common it makes truly crippling scores of 7 or less and how much more common it makes truly useful scores of 16 or higher. This means that you probably don't end up with a character that has significant flaws in 1e terms and are more likely to have a character with a real advantage. And in my experience, it was characters like that that got played with. I never actually saw a lot of people trying to play characters with sub par stats.

Though there is that one guy who always rolls bad, though he's more of a goof off than suicider.

Goofs are mentioned in my rant. I'm not entirely opposed to goofs, but well... they aren't really involved with the rules primarily.

Or some variant math of that idea. It would end up that a guy who rolled well would end up with built-in counter-balances.

Again, this is all randomness mitigation. You can't truly claim you like random if you are doing more and more to make sure the results are decidedly non-random and that everyone at the least gets a functional character. Eventually, down that route you are just deceiving yourself. Once I realized that, I gave up on random stat generation.

What I liked out of random generation was that your next PC was inherently different from the last. In point-buy, it was too easy to just sling out Bob 3 when Bob 2 died who had the exact same build-out as Bob 1, just reset back to the beginning.

This is a good example of what I mean. You claim to like random because it gives you diversity. But if you really like diversity, why are you making Bob #3 when Bob #2 dies? Point buy doesn't prevent diversity. It just highlights for you what you really want. And if you are making Bob #3 to replace Bob #2 whenever the rules allow you to.... diversity isn't really your highest priority.

 

[Bedrockgames]

I'm thinking of this in terms of best practices in design, stepping beyond personal preferences, and thinking about what's better, broadly speaking, for games.

I think the issue is we have a fundamental disagreement about how broad the scope is here. I am saying, yes, if you want to design a game like D&D meant for a broad audience, you are absolutely correct: random disparity is not a good choice. But I am also saying in terms of good design what matters is what you are trying to achieve. If your goal is game played by the vast majority of people in the hobby that has broad appeal, by all means, avoid this. But what if you want to make a game specifically for people like me (and people like me do exist, I know a number of gamers who like the random disparity thing, not many but enough to form a group around). There are also players out there, and again I know because I've met them, who might not normally go for random disparity, but once in a while for a certain kind of game, will enjoy it. I really do not see the harm in these instances, in employing random disparity. Why would it be bad design practice if you know the people you are designing for both like it and want it?

 

[Jacob Marley]

If IIRC, the average on 4d6 keep three goes up from like 10.5 to 12.7, but the really big thing is just how much less common it makes truly crippling scores of 7 or less and how much more common it makes truly useful scores of 16 or higher. This means that you probably don't end up with a character that has significant flaws in 1e terms and are more likely to have a character with a real advantage. And in my experience, it was characters like that that got played with. I never actually saw a lot of people trying to play characters with sub par stats.

Average (4d6 drop the lowest): 12.24
Distribution (4d6 drop the lowest): 3 - 0.08%, 4 - 0.31%, 5 - 0.77%, 6 - 1.62%, 7 - 2.93%, 8 - 4.78%, 9 - 7.02%, 10 - 9.41%, 11 - 11.42%, 12 - 12.89%, 13 - 13.27%, 14 - 12.35%, 15 - 10.11%, 16 - 7.25%, 17 - 4.17%, 18 - 1.62%.

 

[prosfilaes]

Average (4d6 drop the lowest): 12.24
Distribution (4d6 drop the lowest): 3 - 0.08%, 4 - 0.31%, 5 - 0.77%, 6 - 1.62%, 7 - 2.93%, 8 - 4.78%, 9 - 7.02%, 10 - 9.41%, 11 - 11.42%, 12 - 12.89%, 13 - 13.27%, 14 - 12.35%, 15 - 10.11%, 16 - 7.25%, 17 - 4.17%, 18 - 1.62%.

Or http://anydice.com/ with "output [highest 3 of 4d6]"

 

[Umbran]

Why are you choosing 4d6 keep the best three (probably rearranging to taste)? Why not if you really like randomness and diversity go straight up 3d6 in order?

Answer this:

So, assume you want some pie. Why not eat the whole pie in one sitting? Why not go and buy all the pies of all flavors the grocery has in stock right now? If you aren't going to do that, why have any pie at all? The answer is very similar - it isn't an all-or-nothing thing.

It is not nonsensical to want *some* randomness, and to have that randomness somewhat tailored so that we can have it without some of the more egregious failure cases cropping up often enough to be an issue.

 

[Bedrockgames]

I played that way for a long time myself. But, I should point out that even choosing to play that way is itself the beginning of creeping away from random and toward 'above average'. Why are you choosing 4d6 keep the best three (probably rearranging to taste)? Why not if you really like randomness and diversity go straight up 3d6 in order?
.

I think here it is simply a matter of where on the spectrum of randomness one is. 4d6 drop is more about creating some randomness but generally rigging it so things tend to fall within a higher range. It worked pretty well in my experience in AD&D. And groups varied this a lot. I think the fact that they had like five different stat generation methods shows there is incredible diversity of taste here and I think we can accommodate everyone by having that many options for stats in a game.

When 3d6 became the default in 2E, that was when I was doing most of my gaming (I started younger, but really began playing regularly when the 2E PHB came out). It might simply be because I grew up playing this way, but I rather enjoyed 3d6. In fact as a player I preferred 3d6 straight down the line, no assigning stats. There were a few reasons for this:

1) It is exciting. It brings a certain thrill to character creation to have each roll have so much weight and significance. So I enjoy bracing for each result, and yes when you do happen to get three sixes, it is glorious and feels great but...

2) I learned to really like some of those lower numbers. It got me to think outside the box in terms of character and in unexpected ways. Suddenly I am trying to figure out what a character with a 14 Intelligence but 7 Wisdom is like in terms of personality. For me that was a good deal of fun. We also were a role-play heavy group, so this worked for us. I quite like the rolls leading to the character concept rather than the other way around.

3) When you do get an 18, or if you are super lucky and get an 18/00 it is all that much more significant. The few times that actually happens, it is a nice feeling. Somehow I like that I can't just say I want to make a character who is the strongest guy in town, I actually have to roll and get it.

All that said, I think 4d6 drop the lowest is the best fit for how the vast majority of players approach D&D. Making it the default makes sense. Including an option for 3d6, also makes sense as does including optional point buy or stat arrays.

 

[prosfilaes]

I think the fact that they had like five different stat generation methods shows there is incredible diversity of taste here and I think we can accommodate everyone by having that many options for stats in a game.

Or the fact that people were fundamentally unhappy with the dice rolling system, and were trying variations on it. As the history of RPGs shows, there was a lot of tweaking of what D&D gave us before people really managed to break out of the box.

 

[Bedrockgames]

Or the fact that people were fundamentally unhappy with the dice rolling system, and were trying variations on it. As the history of RPGs shows, there was a lot of tweaking of what D&D gave us before people really managed to break out of the box.

But they keep going back to 4d6 drop the lowest. AD&D started out with 4d6 drop the lowest and that is the now the default for 5E. I don't think the options reflected an unhappiness with dice rolling, it just reflected that some people wanted other approaches. But if there was something fundamentally flawed with the old 4d6 drop the lowest method, you'd expect it wouldn't keep coming back like this.

 

[Janx]

Answer this:

So, assume you want some pie. Why not eat the whole pie in one sitting? Why not go and buy all the pies of all flavors the grocery has in stock right now? If you aren't going to do that, why have any pie at all? The answer is very similar - it isn't an all-or-nothing thing.

It is not nonsensical to want *some* randomness, and to have that randomness somewhat tailored so that we can have it without some of the more egregious failure cases cropping up often enough to be an issue.

Umbran nailed it.

We don't want suck characters with all stats less than 7. We don't want to be forced to play a fighter because we didn't roll even a 10 for INT.

But we also don't want cookie-cutter PCs that were just using the exact same point buy as the last PC.

Celebrim's misinterpretation was that I wanted wildly random results with low numbers. I want random results in a certain desirable range.