Sunday, December 13, 2020

A New Paradigm for Random Encounters

    I've done a couple of random encounter tables this year, where I have taken results from many different systems and created selections for specific localities in the Wilderlands.  In doing so I have used a straight d20 paradigm, giving each monster a 5% chance of appearing.  While I've accepted this, I have also felt that a linear progression isn't right and even a regular Bell Curve is problematical at times - such as when a Star Spawn of Cthulhu has a 5% chance of doing a TPK just by showing up.  Once monsters get tough enough, any Dungeon Master who strives for what's been termed 'Gygaxian Naturalism', has to ask, how do normal people survive when such threats are walking around.

    Late in November 2020, Goblin's Henchman over on MeWe shared a post where he discussed using non-homogenous random tables, based on distance from a town to determine what monsters may show up.  The idea is pretty simple create a random encounter table, set your weaker monsters in the lower range, say 2-12 where you can  use smaller dice to generate the curve.  As you move away from civilization, add tougher monsters and increase the size of the dice you roll.

Probability Breakdown

A simple table compares the probability of any combination being rolled by two of the standard gaming dies.  Looking at any line, for example rolling a 5 - shows how the probability falls off precipitously as the size of the die increases; going from 25% when rolling 2d4 down to a mere 1% when rolling two d20. 
I was going to stop at 2d20, but when I started crunching numbers, I realized that 2d20 wasn't going to drop the probabilities enough - so I added the 2d100 progression also.  THEN I realized that was too long AND it STILL didn't give me the progression I wanted (see below), so I added 3d10 and 4d10 columns.  

Roll    (2d4)      (2d6)       (2d8)      (2d10)       (2d12)       (2d20)       (2d100)       (3d10)       (4d10)  
2 6.25% 2.78% 1.56% 1% 0.69% 0.01% 0.01% - -
3 12.50% 5.56% 3.13% 2% 1.38% 0.50% 0.02% 0.001% -
4 18.75% 8.33% 4.69% 3% 2.08% 0.75% 0.03% 0.003% 0.001%
5 25% 11.11% 6.25% 4% 2.78% 1.00% 0.04% 0.006% 0.002%
6 18.75% 13.89%7.81% 5% 3.47%
1.25% 0.05% 0.010% 0.004%
7 6.25% 16.67% 9.38% 6% 4.17%
1.50% 0.06% 0.015% 0.006%

Even going to 2d100 wasn't enough, I ended up calculating the odds for combinations of 1-4 d4, d6, d8, d10, d12 and d20.  I did decide that going beyond 40 lines in a table was more than I'd like to tackle routinely.


    The first change to my approach to building the random encounter tables is that I now need to categorize the encounters by strength.  I have five rather broad categories in mind.  (1) Local inhabitants, farmer, merchants etc.  not usually combat threat but may have information of adventure hooks for the characters, (2) weak monsters - arbitrarily I'll say 1-3 hit dice; I'll need to tailor the definition for Runequest and other such non-leveling systems - I think.  If you're looking for an example, I'm thinking anything weaker than an ogre, as one ogre can put some hurt a party of 1st level characters. (3) middling monsters - say 4-8 hit dice, (4) strong monsters - of 9-15 HD and Oh My God, We're Doomed! monsters of 16+ hit dice.

Taking those categories and adding in some desired frequency by range data produces this matrix.  (I'm thinking of the Wilderlands 5 mile hexes, would also work for ACKS 6 mile hexes.  Would change the range buckets if using 10 or 15 mile hexes; if you're using D&D 30 mile campaign hexes, go straight to the 4+ Hex column.)
Category     0 Hexes      1 Hex         2 Hexes       3 Hexes      4+ Hexes  
Local Inhabitants 85% 75% 50% 15% 5%
Weak Monsters 15% 20% 30% 20% 15%
Middling Monsters - 5% 15% 40% 30%
Strong Monsters - - 5% 20% 40%
Oh My God! We're Doomed! - -- 5% 10%

Thinking about it, probably just use the 2 and 4+ Hexes ranges for waterborne encounters to simulate the increased covert mobility of water creatures compared to land creatures.

Results vs Plan

Playing around with a spreadsheet for several day led me to this distribution of encounters in a single table.  I used examples from the last Random Encounter table I constructed.

Roll                 Category                    Example  
2-7 Local Inhabitant Farmers, Merchants, Local Patrol, etc
8 Weak Hawk (ACKS)
9-11 Local Inhabitant Farmers, Merchants, Local Patrol, etc
12 Weak Orcs
13 Middling Troll
14 Weak Al'mi-raj (FF)
15 Middling Ankheg
16-17 Strong Chimera
18-20 Middling Weretiger (ACKS)
21 Weak Giant Skunk
22 OMG Star Spawn of Cthulhu (CoC)
23 Strong Rakox (Gamma World)
24-25 Middling Cthonian (CoC)
26 Strong Apparition (FF)
27 OMG Small Warrior (MA)
[1 HD, but # appearing 50-100]
28-29 Strong Yexil (GW)
30 Weak Wood Nymphs
31 Middling Cockatrice
32-36 Strong Bearoid (MA)

With this distribution, I could check the probabilities generated by various combinations of die rolls and find the ones that fit my desired distribution the best.

Category     0 Hexes
   1 Hex
   2 Hexes
   3 Hexes
   4+ Hexes
Local Inhabitants 87.5% 75% 43.1% 20.5% 6.1%
Weak Monsters 12.5% 16.7% 29.2% 18.9% 15.6%
Middling Monsters - 8.3% 20.4% 40.6% 38.3%
Strong Monsters - - 7.4% 23.0% 30.1%
Oh My God! We're Doomed! - -- 7.0% 10%

I could play with the encounter category distribution and try to refine it further, but I think I've dialed it in enough for this exercise.


     I don't know if I'll use this when I construct random encounter tables in the future.  The idea sounds good, but in practice you need tables that allow you to roll for monsters of specific strengths.  D&D has done that for dungeons since the beginning, but not for hex crawls.  When I add in my favored way of adding whimsy by combining results from every game system on my shelves - well that would be a project in and of itself to create a set of master encounter lists.
     The other thing I don't like about this approach is the players will learn the numbers to run from and the numbers to ignore.  Such meta-game knowledge I think would lessen the immersion of the players in the game, as they will react differently how their characters would.  After all if the player knows the rustling in the bushes is a wood cutter or shepherd or such, the they lose the mystery and excitement that this time it might be the Blatant Beast come to hear the epic poem it tasked them to create the last time they encountered it.

Your thoughts?


  1. Not bad. But due to your layout, you're getting things rolling over onto the right sidebar.

    1. Yeah, tables have to be hand coded in Blogger. I installed a CSS template in the page header that's supposed to handle it but lost some internal tags when copying and pasting HTML.

  2. A really interesting idea, I agree that more than 40 and the tables get unwieldy. I have been using d20 + a modifier to shift along 40 catagories in a liner way in the wilderness. Hadn't given the bell curve much thought. The original source mentioned applying it to a megadungeon which i might explore a bit as i am currently using the old "levels" for depth in a horizontal megadungeon, this may fit better.