Tuesday, February 25, 2014

On random character generation


On random character generation

I like it random*.  I like being surprised by what the numbers mean –or finding out how to make sense of them. 
* shut up.

I also like point buy systems – I like being able to hit the ground with the character I feel like playing.

At one end we have the roll 3d6 in order & suck it up (OD&D)
On the other we have complete customization down to the last .1%; nothing is randomized (Champions/TFT, GURPS).

I've seen it argued that Random vs allocation systems are at the heart of what makes one a true roleplayer, and worthwhile and self assured person; It works both ways, too: random players are innumerate boobs role who play to play a loser, allocation players are "compensationg for personal powerlessness".
For me, the problem is thatwhile I like playing what i am handed,  I cannot help but be a munchkin, given the chance. , while I love getting to play exactly the character I want, I know myself.  I can’t help but look at allocating numbers as an optimization problem.
All the supervised generation characters tend to revert to the basic type I like to play –all combat guys end up as Thrudd bigaxemann , the VERY strong of the strong hills of strongaria  ; all wizards are Bored-Flak the pyromaniac genius, living platoon support weapon; all thieves as Batman.  Period.
Also, it doesn’t have to be numerical optimization per se: I had to have a GM limit my character choices in a 3.5 game to only base classes and feats because I kept locking up with options, and I could never quite finish the damn character. 

So, I tend to want to design games with random generation; but, with rules lite gaming (which I prefer) there is a stats problem I’ve never seen mentioned:
 if you only have two or three stats, it is very easy to get screwed by one low probability roll;
and the likelihood of getting a low and a high to balance out are not excellent with three possibilities.  And yes, I know one can get gifted by a single outlier roll which is good –or even two.  But, one, no one complains about that, and two, it can screw the other players in terms of fun or participation  if one guy can handle all the mental tasks.  But the real problem that messes with both is the character who cannot function.  So that’s what we worry about, essentially falling off the left side of the bell curve, which for only a few rolls is surprisingly steep.

Less stats is always at the heart of Lite RPG design, but more stats at least means more hope of an acceptable distribution.  So rules lite typically are point buy/allocation systems, with or without numbers. Let’s call them Supervised.  And, to even things out, we could call the random ones Stochastic, but that is pretentiousness of a level barely tolerable after using “Supervised” so let’s stick to calling it  Random.

The first version of BAGS used a pair of D6 to generate 0-5, skewed ((roll 2 d6, subtract the high from the low.  Most frequent roll is a 1 with 5 being least frequent (1/36).  Good, right ?  Nope.  At least one character could not roll above a 1, and indeed had a 0,1,0 character.  With the original BAGS rules, that meant that for any action or interaction test, he needed a 6 to succeed. Hes a good roleplayer, so he sucked it up and played thru until I changed to 7+ for success –then he complained.  At which point, the party (and me) realized that despite his best efforts, he couldn’t do jack , so any situation which was complicated by more of the players failing a roll pretty much was always the case. 

But crap.  If I give players 5 points, some dork invariably goes for 0,0,5.  If I require at least a 1, the number of different characters is vastly reduced.  More than 5 means you need to limit the max, so basically all you get is Mr average, and miss onestat wonder.  If I change the value of a 0, whatever number it becomes is going to be the most frequent one, and there is no granularity for low stats. 

IIRC, I wussed by allowing players to pick a profile instead of rolling.  Which is okay, but balance in small numbers of permutations equals (for me) boredom.

Okay, here is my new thought.

Mongoose traveller (see, a motion to discuss traveller is always in order) originally included ironman rules.  This harked back to the old “you can die in chargen” rules for Traveller classic/original.  Normally,  in MGT, a failed survival roll left your character alive, but physically messed up –missing an arm, one stat massively reduced, permanent high tech tattoo of LOSER or THE THIRD IMPERIUM SUCKS or ALL SPACE MARINES ARE FILTHY COWARDS on your forehead kind of thing. In the draft of MGT, you could opt to have survival rolls kill your character, and in exchange you got some kind of a reward.

I think it was essentially a level of Ironman per term you rolled for, and in play, you could always use your ironman skill instead of any other skill. Or maybe it was a number of rerolls.  Whatever, it got left out of the final version (boo hiss) because I think the stodgyest members hated it worse than death because it made no sense.  Or was unrealistic, scientifically.  (As decreed by a which  qualified him to have authoritative say on social issues, probability and trade economics, as well as play balance.  )

Anyway, rant over, the point is that it allowed you to be more random and get rewarded, just like one gets a reward for supervised chargen.  Which I like.

So, guess what is next up on BAGS revision planning ?  This:  for each stat you roll randomly, you get a free point of GRIT.  

Wasn’t that a lot to say to justify a simple design choice ?

Yep.  It was.

2 comments:

2eDM said...

at least in my stripped down AD&D 2e games, I kinda solved this issue. I found that I like playing & DMing in the 61-65 ability score total bracket, but I HATE just letting people spend their points. I ended up creating a file, one table for each class(Fighter, Mage, Cleric, Thief) with 100 pre-rolled characters in that range. When someone needed a character, I'd just have them roll 1d100. Once a set of rolls was taken, I'd reroll that slot after the game that way it was unlikely any 2 characters would be exactly the same.

Ed Allen said...

Back around the time we met I was still hacking a bit on my Forty Cities D30 homebrew RPG. Ran it a bit at the end of my time at Steam Tunnel U but I never ran it here.

Here's the approach I took in that game. In a 6 stat system, players allocated 2/3 of the points, IIRC bumped up a bit over a random Joe, so instead of 42 that would be average, they got like 45 or 48. They put 2 to 12 from that pool in each attribute. Then, after allocating, they rolled the last die for each stat, no swapping, no rerolling, and added it on. So your mighty Thrud might be 13-18 STR and DEX and 3-8 WIS and INT.

I still like this system a lot in how it splits the difference. You are guaranteed not to suck at any stat that's important to you, but not guaranteed to shine at it. It is a bit swingy since you only actually roll 6 dice. Even if rolling badly, you are unlikely to be a worse than average guy overall because of the extra points on the allocated "rolls". But even minimax dumpstatting a couple of stats to concentrate points you are only guaranteed to be good, not necessarily great, at the other stats.

In Traveller it would have to be allocate half, roll the second die, or allocate 2/3 and roll a D4 (heresy, I know).

You recall seeing something like it published anywhere?