Random Tables without... the tables

Inspired by this thread:

Random Encounter Tables

I've always enjoyed the various creative prompts and instant imagery which can come from random tables, random encounters, game oracles, and other similar techniques.

However, I often don't want to bother with actually spending time outside of play writing up the things, only to discard so many elements for the actual game. What if you're only playing once? Is it worthwhile to write ten elements on your chart, when you will only actually use one? (Sometimes it is, yes... but it's pretty inefficient, as well.)

One of my favourite gaming/GM techniques is to use input from the various players, without consulting one another, and then treat them as "random table" results. In this way, you get a variety of input, nothing is ignored, and the creative spur, the unexpected, comes into play.

For example, I have, on several occasions, played In a Wicked Age... this way. So long as everyone has a clear sense of the genre and tone for the game, it works very well. Every player writes one story element (without consulting with the others), and then we throw them together. It's much like a "draw" from the oracle... except with no oracle.

My favourite trick, though, is to have players write up some elements like that for the developing story, and then do something random TO them, before throwing them back into the game.

Here's a simple example (not a good one, sorry - just for illustration). I have four players, and I say:

"You all recently arrived in the Frontier Keep. Who is the most interesting or important person for you, here? (Make up an NPC.)"

"What's a weird, inexplicable rumour you heard last night? A strange event which left you scratching your head, but deadly curious?"

Then I make a little chart.

(A)_____________________ secretly works for (B)____________________, who seeks revenge against (C)_____________________, who is dependent on (D)___________________ in a terribly illicit way.

I roll dice and randomly assign one NPC to each slot. Then I randomly assign (again, dice) a rumour to each NPC, and consider how they are related - generally, something jumps out at me: either the NPC is responsible for the rumour, or, perhaps, the rumour is a hint of something that happened TO them (if the former doesn't make sense).

I've done it the other way around, too:

Shown with a list of "random items" or story elements, and handed them to the players to assign in some fashion. My Lady Blackbird one-shot hack, for instance, has a list of NPCs and a list of dramatic and unlikely reputations. Before we play, I tell them to take turns assigning a reputation to each NPC. I love it! Every time I play the scenario, the NPCs take on a completely different flavour (e.g. the players decide the little girl is a cannibal!)

This isn't exactly like using random tables, but I've gotten such a kick out of it that I've started doing it pretty much every time I ran a game of some sort. I think there's a conceptual overlap here, mixing "randomness", "uncoordinated input" (each person throws in an idea), and "unexpected coincidences".


  • Cool idea!

    I tend to just use a d6 table. Whenever a character does something and I can't predict the outcome, I openly write out the numbers from 1-6 and invite suggestions. (6 is always "something else make a new table")

    Eg. Nuclear bomb. A non-scientific character pulls out random wires:

    1) nothing happens
    2) boom!
    3) countdown slows down
    4) countdown speeds up
    5) the ricochet scares a cowering scientist out of hiding (this was probably suggested by one of the players)
    6) something else, new table (mutation, space warp etc.)

    Everyone can see it's fair. When the characters knocked a satellite out of orbit, I could've ruled that it probably crashed in the sea. If I'd ruled it hit a population centre, they'd think I was clocking the game against them. When open rolls on a sequence of d6 tables resulted in it crashing on Sydney everyone knew it was fair - just very bad luck.
  • edited October 2016
    It's been mentioned many times before, but one of my favorite shortcuts for decisions is the 50/50 binary tree. Every time you get a "no" answer and step down this tree, you are moving from a "more likely" choice to a less likely one, so this follows "Zipf's Law".

    Zipf's Law is a fascinating glimpse at the outline of our own epistemology. It shows itself not only in the classic "Zipfian Distribution" and the "80/20 Law" but in the stochastic behavior of any large random dataset featuring preferential attachment (such as the evolution of species, or the distribution of differently-valued objects in a random sample). It will even determine how many - though not which - of your sessions will become the most memorable ones, and just how memorable. It is a weird aspect of how our reality works.

    For GMs, Zipf's Law is most realistically (and simply) modeled by using a "Binary Tree" approach, coupled with the "Most Obvious Thing" principle. This is laid out in the DayTrippers GameMasters Guide, so if you've read that book you can go home early. For the rest of you, it works like this:

    Deciding What a Thing Is:
    - think of the most obvious or likely thing it could be
    - roll a "Yes/No" die (high means Yes, low means No)
    - if "Yes", that's what it is
    - if "No", think of the next most likely thing
    - repeat until you know what it is

    Detailing a Thing:
    - think of the most obvious or likely trait
    - roll a "Yes/No" die (high means Yes, low means No)
    - if "Yes", the trait is true; go back to step 1 for more detail
    - if "No", think of the next most likely trait
    - repeat until you have a useful number of true traits

    "Is the encounter friendly or not?" = 5 so yes, it's friendly.
    "Is it magical?" = 3 so no, it's mundane.
    "Is it someone the PCs know?" = 6 so yes.
    "Is it Jake?" = 2 so no.
    "Is it that old woman from the swamp?" = 4 so yes, it's her.

    Each time you get a "No" result, you will think of something less likely than the previous candidate. And if you receive a "No" to that question, the following candidate will be even more unlikely. But since you only end up stepping down that slope 50% of the time, the ontological distribution of things in your gameworld will follow a Zipf Curve just like things in the real world(TM), even though they're completely random. The most likely things will be twice as frequent as the second-most likely things, which will be twice as frequent as the third-most likely things, and so on.

    By structuring your fictional ontological queries this way and letting the dice do all the work, you are mechanically guaranteeing that the distribution of characters, creatures, objects and qualia in your fictional world will fall along Zipfian Distributions, emulating this stochastic law (paradox!) of the observable universe.

  • Hunh. I've never heard of Zipf, but that's very roughly how Mythic GM Emulator works as well.
  • That's great stuff. I got the 50/50 technique from Eero - he used it a lot in our OSR games a few years back.
  • I often use this technique in my own meanderings, but I do not assume that the generational probabilities match the in-world probabilities. That is, there might be a 50% chance that some event occurs, but I assume each check is twice as eventful moving forward.

    "Is he strong?"
    Coin flip. Yes (1/2). Rather strong (1/4).
    Coin flip. Yes (1/4). Very strong (1/16).
    Coin flip. Yes (1/8). Incredibly strong (1/64).
    Coin flip. No. Stop.

    "Rather strong" here doesn't mean in the 50th %tile. It means in the 75th %tile.
    "Very strong" doesn't mean in the 75th %tile. It means in the 94th %tile.
    "Incredibly strong" doesn't mean in the 88th %tile. It means in the 98th %tile.

    Oddly, I wouldn't square the probabilities going in the other direction. If I got three "No" results, I'd say that was (1/2)^3 = 1/8 = 13th %tile, as expected.

    But if we established the weakness as the interesting trait, and asked HOW WEAK, I might square it. Oddities of my mind, I suppose.
Sign In or Register to comment.