Caterpillar Game Engine … someday

Cat 11


:: image_preview pdf version of this method: Link

I saw this post Dice are Statblocks on the Tarsos Theorem blog. I thought it was super neat and it made me wonder about other possible uses.

Basically – in this method you roll 3D6 and gather them up into a sort of dice caterpillar shape (see blue cubes above).

What is really neat here, is that more than just the top faces of the cubes are being used. The order of the cubes and the orientation of the cubes with respect to each other are also being used. Basically, there is a whole bunch of information being generated when the cubes are simply gathered into a ‘dice caterpillar’. Using all this extra information is a neat idea! Tarsos levels up.

PC stat generating method

Tarsos’ blog post prompted me to hastily dash this blog post off: Caterpillar Method’ for Character Stat Generation (i.e. a way to make the ‘standard’ 6 stats for a D&D type character with one roll of 3D6 arranged into the caterpillar shape).

That is, in my method you roll 3D6 once and use a ‘Rule Set’ to generate the 6 PC stats. Because, I wanted any PC generated by this method to be ‘reasonable’, I modelled the method against the ‘standard’ 6x3D6 method and got pretty good agreement, on average, over a big set (there is a graph in my post mentioned above).

Lumpy is good

But, this is a ‘Take 2’ blog post, as I think there is more to say. Specifically, Reddit had more to say (links to follow). I liked what Reddit had to say, so I dug deeper into the system. This system on average gives results that smooth-ish-ly modelled the standard 6x3D6 system … but on an individual basis, the results are lumpy.

But, lumpy in an interesting way. At least I think it is interesting.

Lumpy probabilities

Here is a breakdown of the probabilities of each stat, using ‘Rule Set 1’ (see my first post if you want to revisit the Rule Sets):


(dashed lines above are actually not possible, but included to help visulaize)

So, on balance, you should get a fairly playable if not somewhat ‘balanced’ PC, which always gets:

  • 1 x Low stat (yellow line)
  • 1 x Moderate stat (blue line)
  • 1 x High stat (green line)
  • 1 x 3D6 bell-curve type stat
  • 2 x 3D6 ‘counter-weighted’ bell-curve type stats that balance each other out.

This idea of balanced stats has precedence in systems like Black Hack.  In some ways it’s almost a hybrid between random and a ‘point buy’ system, because it has some self-balancing mechanisms baked in. 

For example, below are (I believe) the best and worst PC stats you can get (weighted to give 18 or 3s respectively):

  • Best possible stats: 18, 18, 15, 13, 12, 6
  • Worst possible stats: 15, 10, 8, 6, 3, 3

*** Reddit suggest that the above numbers shoud in fact be (I need to check my notes, but on the face of it, I think they are correct):

  • Best possible stats: 18, 18, 13, 13, 11, 8
  • Worst possible stats: 11, 10, 10, 8, 3, 3

Just to spell it out, there is not even a theoretical way to get a character with 18, 18, 18, 18, 18, 18 (i.e. 3D6s do not have enough 6s on them).

In my original post, to get rolls closer to 6x4D6 drop the lowest, I baiscally had two ‘Heads’ instead of a ‘Head’ and a ‘Tail’. Another way to go would be to allow the player to decide which end is the Head and Tail after rolling.

Final thoughts

Personally, I think this idea has ‘legs’ for all sorts of random/solo play RPGing. Time permitting, I’d like to make a random village by caterpillar-ing groups of 2 and/or 3 D6s. Or a random dungeon, rooms generated in the same way. So much information from one roll. The counter weighted faces are especially interesting as you may be able to link one random property to the other in an inverse relationship (or if you invert the table order, in a proportional relationship). And, with my system, I’m not even using all the information (e.g. the Face and Rump ends could be used together on a ‘D66’ table etc.).

– – –

thumbMe on DriveThru.
Mainly pimping my PWYW one-page procedural dungeon generator using Hex Flower Game Engines:
In the Heart of the Delve and Dangerous.

5 thoughts on “Caterpillar Game Engine … someday

  1. Pingback: ‘Caterpillar Method’ for Character Stat Generation | Goblin's Henchman

  2. Pingback: ‘Caterpillar Method’ for Character Stat Generation | Video demo follow up | Goblin's Henchman

  3. redlaWw

    I was just linked to here from elsewhere, and I have to say, this method has big issues that you’ve failed to discuss.
    You’ve only discussed the marginal distributions of each stat, but their dependence is a critical feature.
    The dependence on the “front” and “rear” is fine – it guarantees “front”+”rear”=21, which means two average stats or one good stat and one bad stat.
    The dependence of the other 4 stats (and, to a lesser extent, dependence of the “front” and “rear” on these), on the other hand, is problematic:
    A 6 on the top of the “head” die guarantees a “head” stat of at least 15 (7+6+another number not 1), but also indicates a likely good “top” stat. On the other hand, a 1 on the top of the “head” die limits your maximum “head” stat to 13 (7+1+another number not 6) and likely indicates a poor or average “head” stat. The “tail” stat is in a similar situation. In addition, the “middle” stat is just 7+top of “middle” die. All this together means that your performance in the “head”, “middle” and “tail” stats are highly dependent on what you rolled for your “top” stat. And this pushes the stat distributions into “tiers”, based on the result rolled on “top”. If you roll well on “top”, you’re guaranteed to have at least 3 good stats, probably 4, and on the other hand, if you roll poorly on “top”, you are likely to end up with few to no good stats – maybe a single 15 at best. Indeed, if you get 3 on “top”, your maximum possible stat is 15 from the “front” or “rear” (which can’t be 16+ because all the 6s are on the bottom).


    1. Goblin's Henchman Post author

      Thanks for the comment. In the end, this is still a dice rolling system … so if you’re unlucky you’re unlucky.

      But, the system does have some self-balancing properties, so ‘unlucky’ never becomes **REALLY** unlucky. That is, it’s not 100% self-balancing.

      Personally, I think this system is interesting as it offers some variability, but with the assurance that the resultant PC will be good/competent in at least a few areas.


  4. Pingback: 2022 | end | Goblin's Henchman

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s