Sunday, April 17, 2011

5 Kinds of Random Generators & What Makes Them Not Suck

Ok, so I've been (shockingly, I know) thinking about random generators. Specifically: which ones are fun, which ones aren't, which work, which don't, and all that. I feel like I've boiled generators down to a few basic types. I kinda feel like: just how Gygax included a little info on dice curves at the beginning of the DMG, he should've explained a little about different generators up there, too...

PLAYER RANDOMNESS ANXIETY PRINCIPLE or PRAP: Before I go into these kinds of generators, though, I should point out there's a separate principle at work which overrides everything below: Nearly any random generator can be fun and good if the players know it is being rolled on, they have some idea of what the possible outcomes could be, and this produces anxiety.

Like for example here's a crappy generator:

What's That Over There? (roll d6)
1-3 Bear
4-6 2 Bears

If the players know what's on this table, or have rolled on it before, or if you just say "ok, we're going to see how many bears there are" then even this table is worth more than zero so long as the PCs are afraid of bears (and if they're not, you need to work on your growly noises and read "New Mutants: The Demon Bear Saga.) Some people don't like talking a lot about the random tables to the players since they feel it breaks immersion and if that's your life then hey, live how you wanna, but anyway my point here is simply that the categories below only address how good different kinds of tables are in situations other than ones where the players know kinda what's on them and know you're rolling on them--or in addition to them.


Simple, Complete Option Generators

These are tables and methods that parse through basic, familiar choices, basically just a list with numbers in front, like:

What kind of trap is that?
1-Pit trap
2-Poison gas
3-Poison needles
4-gates crash down, locking PCs in

or like the random building table here.

The items in the generator are not clever or creative, they are just the array that you'd expect in the genre/situation you're playing in.

These are good if:

  • There are a lot of options to parse through (more than you can keep in your head easily--say more than 12), or...
  • You've got a computer program, deck of cards, chart, or other mechanism that generates lots and lots of choices from these simple options in one go. Like for example an 8-option wind direction table (north, northwest, south, southeast, etc.) is pointless but a program that independently generated wind directions for each hex of an 800-hex ocean might be useful and save you some time.

These are not good if:

  • It takes as long to consult the generator as it would to just make something up and there are so few options that you could easily have come up with anything in the generator. Most "types of traps" generators are like this: they are fun in a PRAP situation, but if you just need 5 traps because you're making a dungeon, sitting and rolling 5 times on a d8 table can easily be more trouble than it's worth. Yeah yeah, I know you like rolling on tables--but in this situation, even if you don't want to make up your own dungeon, wouldn't you at least rather be rolling on a d100 table?

Simulational Generators

These are just like Simple, Complete Option Generators only the numbers are weighted more-or-less carefully to match the probability of things happening in real life or in the genre of story you're simulating. Most tables in the AD&D DMG work this way. Here's a simple one:

What's Wrong With This Guy On Fox News? (d100)
1-98 Gay and doesn't want to admit it to himself
99-00 Something else

Another example of this kind of generator would be basing your campaign map on the map of some random real place you found. Character generation using 3d6 or 4d6-pick-the-highest or other common variants is simulational since handfuls of dice always tend toward the median result.

These are good if:
  • You use them over and over and over again and do not fuck with the results very often. They will reinforce the internal logic and style of whatever structure the generator is creating. They will, in effect, become "rules" of your world ("goblins usually use slime traps") just like the fact that PCs rarely have a 3 or 18 strength is a "rule".
  • Creating a single structure/creature/place/adventure element requires rolling on them so much that you will probably get an unlikely result that you can build into an adventure hook and there is some computerized/card-based/graphic-based mechanism that can do all this rolling for you faster than you could think up your own hook. Like you create a weighted castle generator and most of the results are typical but all that rolling eventually gives you the result that the third courtier from the left is a werewolf. Well then you've got your adventure right there, plus a believable place to have it.

These tables can be a drag if:
  • You rely on them to give you adventure ideas and--however you're using them--they don't regularly produce them. When you get an unusual result, they'll help, but the whole point of the table is that unusual results will be rare so by the time you get one you may have already thought of some. Or you've gazed longingly at the unusual results for so long you already have a favorite you want to use.
  • You do use them over and over and over again but the internal consistency you create by having these weighted results isn't having any effect on anything anybody notices while playing. Like if javelin traps are 5% more likely than dart traps who cares? Is this some important genre convention the rest of us don't know about?
  • You don't get to use them enough that the weight of the probabilities is felt. Like I've used the DMG random treasure table like twice, ever. One of those times I got the Book Of Vile Darkness. The (extensive) simulational aspects of that table counted for nothing in terms of impact on the game. I had a Random Goblin Witch Curse table--I used it once in a PRAP situation--it wouldn't have made sense to make this a weighted, simulational table.

Complete Creative Result Generators

These are kind of the opposite of simulational generators. The effort hasn't gone into weighing the probabilities--most or all of the options are somewhat improbable. This is basically a bunch of allegedly-interesting maybe semi-unique ideas someone has thought up that are put into a generator.

Here's the classic and it rules.

These kinds of results can be added to Simple, Complete Option Generators or Simulational Generators if there are a lot of results in the generator, but if there aren't then the Creative options will appear over and over and become a cliche. If a generator is all Creative Results then it should have a lot of options--for the same reason.

A good random mutation table is usually a combo of Simulational and Creative results so if you roll up a lot of mutants you get nine with tentacles and bulging eyes and like one with snails instead of teeth. (Interesting philosophical question: the Warhammer mutation tables are weighted so you can produce armies with common mutations and outliers, however you're supposed to use the same tables when rolling up a personal mutation for your Chaos Champion PC if playing an RPG. So your chaos champion is more likely to end up with a common mutation, not a crazy creative one. Is that good or bad? Oh god I can hear the GNS theorists in the comments already, forget I said anything. )

Another thing about these is they often work best on the "after you've used this result, cross it out and write your own" model.

Often these are best in extremely specific situations, so it can be kinda tough to decide which one to make: the more creative you can get with your generator, the less broadly applicable it's likely to be. I have gotten to use the 999 table like three times, ever. Such is life.

These are good if:

  • You want something interesting to happen and are sure you can handle the curveball it'll throw into your adventure.
  • You have a list of good ideas you want to use but don't have room to cram them all into the next adventure you write so you're kind of putting them "on ice" for later.

These are a drag if:
  • You use these all the time and so the PCs just get the feeling no choice they make matters because no matter what they do something completely weird the DM rolled up will happen. i.e. the world becomes so non-simulational it has no internal logic at all. Without some internal logic the DM has all the power and the PCs have none. (Jeff's Party Like It's 999 table avoids this by being a table the players decide to roll on, and by only coming into effect if the carousing goes wrong.)
  • You really don't want these results in a generator, you just want to use all the ones you like whenever you like, so it'd be better to just have a list, actually.
  • (related) The results are so specific and there are so few results that if you get a result more than once (assuming that's likely) it starts to be lame.
  • The person who wrote the stuff in the generator was creative in a direction that isn't really your bag.

Incomplete/Inspirational Generators

These are generators which, unlike the previous two, don't tell you exactly what's going on--they give you imagination fuel and require you to fill in the gaps. This is one of them, as is this, as is Jeff's idea of picking up a random issue of Dragon and making a campaign out of it. (Yeah, I'm using a lot of Jeff examples, what can I say, the man loves generators.)

This generator is a hybrid--it has some Simple, Complete Option tables and some Incomplete tables. It's a good case study--I don't like it and only used it once because:

-it was technically more cumbersome in many ways than it needed to be--all the possibilities for the first 2 tables--for instance--could have fit easily in one table.

-it took as long to generate an area and think up a unifying idea as it would've taken just to think up an interesting thing for the place without the generator.

This one
works a little better, but I've still only used it once, since it has the same problem: it takes just as long to make results that fit as it does to just think up a dungeon. (Though I get mail that other people like it and use it, so whatever.)

Yesterday's city generator post was a better hybrid, I think: 67 of the 78 results told you exactly what was where instantly (they were Simple, Complete Generators) and the last eleven (Incomplete) results gave you hints you could use or just discard and stick with the insta-results.

...which points up the thing about Incomplete Generators:

These are good if:

  • You think the self-imposed challenge of thinking up ideas within constraints is fun and you have time to do it. (Which must be at least a little true for you sometimes or else you wouldn't be GMing.)
These can be pointless if:
  • It takes so long to use them you could've just made up your own thing or you're in the middle of a session and you would rather just know what's in the dead guy's pockets now please, thank you.
  • The hints are incomplete but not terribly creative. Like if you roll on the random evil-wizard-dungeon table and the extent of what you get is: 67-80 "A monster" and that's all you get, well, hey, that didn't get me too much farther than I was before I picked up the generator.
  • The generator spits out tons and tons of incomplete results so that filling in the blanks is less inspirational and more just like the generator made you do more work than you would've if you'd just designed your own thing from scratch. In other words, these kind of generators are mainly useful because they are fun--so you actually sit down and do the prep--if they're not fun, they fail in their main task.
Breeding Generators

These are made up of elements which are, in themselves, incomplete, but which are combined with other incomplete elements to create a complete result. One example is Stephen Poag's Exquisite Corpses.

Here's a simple one:

What's your transmutational superpower?

You can absorb (roll once on this table) and turn it into (roll again this table)

1- sound
4-gamma radiation
5-kinetic energy/vibrations

The nice thing here is we get all kinds of emergent things we hadn't thought of when we threw the table together--"absorbs light and turns it into kinetic energy"--that's interesting: roaming around in miasma of darkness causing tremors all day.

The hard part about making one is there's so many possibilities it's hard to make sure all the results are going to make sense, especially when the list of variables gets long--like: "turns electricity into light" is that really all that great? You're a table lamp. Well we could change "light" to "lasers"--but then if you rolled "lasers" first your power is to turn "lasers" into something and that seems not terribly useful most of the time unless you live on the Death Star and plus if you can turn lasers into electricity, well, I mean it's cool to shoot people with electricity but why not just leave them as lasers?...

Obviously this can be ironed out, but the point is making a good Breeder isn't easy.

Also note that a Breeder is different than just stacked Complete Generators like one where you roll race randomly and then class randomly--that just gives you a combination of existing options--which is basically two Simple, Complete Option generators stuck together. Those produce combinations of known quantities--which is what most things in games are. The Breeder is also different than something like this which has a similar mechanism but is really an incomplete generator--you have to finish it yourself. Breeders create new "building blocks"--and each block is finished when it comes out.

Functioning breeders are really the bee's knees when it comes to random tables--a whole greater than the sum of its parts. There isn't much of a downside to using them, it's just hard to make one that works and that's on a subject that's generally useful. Like when LOTFP announced they were doing a book of Exquisite Corpses, the first comments were Awesome, now do one for adventures, one for dungeons, one for wilderness maps...

And that ain't easy because you can say head of a...body of a...legs of a... and you've got a complete monster that is pretty much ready to play with. Doing that with a dungeon or adventure or outdoor map would probably just give you an Incomplete Generator unless you're super clever about it.

Ok, so, there you go with that, kids.


  1. As you know I'm sorta, well, not anti-random generator per se but I don't like'em, use'em or need'em much. You've actually listed pretty much the all the reasons I don't love them so I won't go into that but I will tell you which ones I have liked.

    "Incomplete/Inspirational Generators", akin to the Super Powers charts from the old Villains & Vigilantes game were fun. I really liked the challenge of rolling Animal/Plant Powers twice, coming up with Fish and Bird abilities and trying make a seriously cool superhero out of that (Done. The Seahawk. Native American Hero best described as a Crossbetween Aquaman, Hawkman and Shaman from Marvel's Alpha Flight).

    More times then not however, I would drop a power (to avoid a weakness in V&V) and tweak a thing or be left with a ridiculous hero. Other players just made silly supers and it always felt weird.

    So the Incomplete/Inspirational Generators and Breeders are ok in my book, just not as cool as making it up myself.

  2. That's a very solid analysis.

    On in-play tables: Player drama can also include celebration, not just anxiety, for instance when THEY roll a crit.

    On inspirational tables: I use them to break my imagination out of cliche. But a really cool thematic idea trumps the random table. Fact is, I just threw out a randomly stocked level and replaced it with a wicked theme.

    Simulation tables provide plausible deniability. The troll-slaying sword, where you don't want the guy with it to NEVER run into a troll but if you intentionally make a troll it looks forced. Works the other way for trolls with elf-slaying swords.

    There's also nothing wrong with jogging your imagination reading a list that just happens to have numbers next to it. I guess that's your complete table, there.

  3. My holy grail for a long time has been a single treasure item generator to use in solo play. I want to be surprised by my own table. The difficulties are related to your categories.

    It can't be a Simple, Complete Option Generator, because there needs to be a ton of options.

    It can't be a Complete Creative Result Generator, because, again, no matter how cool the stuff I make up, it won't surprise me.

    So, it has to be a combination of a the Incomplete and the Breeder. The difficulty is, as you say, abstracting out the commonalities without allowing for too many weird anomalies. I haven't given up yet.

    And yeah, I saw those comments on LotFP and thought something similar, I spent hours of my youth trying to come up with solo adventure breeders. Its hard. My best attempt was to cook treasure/monster generation into hexagonal geomorphs. That would work for a very vanilla dungeoncrawl.

  4. @roger

    the "anxiety" refers to pre-rolling emotion.

    And, naturally, the emotion of anxiety implies both hope and fear.

  5. Very thorough analysis. Nommy food for thought the next time generators get written. Possibly you should consider making a PDF of your RPG theory posts like this one... :)

  6. I call the Mad Libs approach of your sample Breeder Generator "formulas". It's a big deal in the 20-Sided Quickie tables I was working on; I have a bunch of formulas like "MODIFIER + LAND FEATURE" for naming geographical features or "MATERIAL + PERSON + OBJECT" to describe statues of people. The all caps phrases are names of columns in the table.

    For Incomplete/Inspirational Generators, I'm relying much more on the Random Random Table (based on Lewis Carroll's Memoria Technica) with the dice rolled on a sheet of paper divided radially into areas labeled with the six Ability scores arranged around a hub.

  7. So, I take it you don't like my bear table:

    But I dig your advice here. Lots of things to think over in terms of making material more useful and creative.

  8. What's Wrong With This Guy On Fox News? (d100)
    1-98 Gay and doesn't want to admit it to himself
    99-00 Something else


    I also thought I'd (slyly) pimp my Wandering Harlot Table, as seen in Fight On!

    Which can easily turn into a whole subplot of its own. Which would make it a breeder, but I don't think I can cope with any more single entendres tonight.

  9. @Adam

    This is like the 9th time you've commented on this blog to tell people about that table. I can tell you're really proud of it.

  10. Zak, do you think there's a difference between tables that are designed to be used in adventure creation, and tables that are designed to be used during play?

    Personally I like making tables as a spur to creativity. If you have to think of twenty villainous motivations, you're probably going to run out of obvious ones and be forced to come up with non-obvious ones. It's kind of like a brainstorming technique.

    When DMing I mostly use the reaction table, wandering monster tables, and one I made myself for the class and race of NPC parties - outside of tables I write for particular encounters ('if the PCs throw a torch at this, it will explode on 1-2...').

    This started as a post about how I don't use random tables much, but in writing it I realised that I actually use them all the time.

    Also I often make up 'tables' for specific cases and roll when I'm not sure what should happen.

  11. Hmmm. This post gave me an idea, and I worked up a small table to explore that idea...

    Mid-Level Henchman of Evil Wizard Table

    Race, d4:
    1 - Human
    2 - Dark Dwarf
    3 - Human/random animal hybrid (anthro, centaur, whatever you like)
    4 - Cthulhuoid starspawned abomination

    Weapon, d4:
    1 - Axe
    2 - Crossbow that magically reloads itself (or a pair of them?)
    3 - Frost mace; extra cold damage, slows opponents
    4 - Electrically charged sword that screams to disorient opponents AND throwing and returning warhammer

    THEN add together those previous rolls and subtract from 9!

    Extra feature, 9 - (Race + Weapon):
    1: No extra feature!
    2: Surrounded by glowing holy radiance (it's just light, not an actual holy aura...or maybe it is!)
    3: Casts magic-user spells (so what if it uses an axe? it's a natural magic ability)
    4: Spits venom (play it like throwing a dagger, but it's a free action it can do every round)
    5: Pet frog on shoulder that throws poisoned darts
    6: Can fly as if affected by permanent Fly spell
    7: When killed, a demon bursts from his skin with full HP; if the demon can survive or escape with his skin or equipment, he can re-humanize himself

  12. @anarchist:

    Is there a difference between during adventure creation and play? Mostly, yeah--you have different sets of design goals however, they are nto mutually exclusive. Some tables meet both, but most don't.

    During play, DM-only tables the players -don't- know about should be fast and ones players use or know about the DM using should be slow but high-stakes. Like you don't want to say to your players "now we're going to roll for the kind of furniture in the room"--that doesn't increase tension.

    During creation, tables need to follow the specs in this post more or less.

    As for creating tables as a brainstorming technique, hey it's true, we all do that.

  13. Primarily I use tables out of play for prep. The tables I use in game are tables where player stakes and interest are high (critical tables, treasure) or setup tables for encounters (taking 30 or so seconds to set up a scene).

  14. One thing I've noticed is that I'm coming up with a lot of quick things as needed, and using table for other purposes than they were originally meant for.For example, I'll go "Okay, I need to know if there's any wind, what the temperature is, and if there are any weird smells", so d2 (no wind) d3 (neutral), d2 (yes). Then "What can I use to get some weird smells? The potion appearance table has smells and tastes. Those'll work." d8 (turpentine), d7 (sour). From that I get "The air is still, and neither warm nor cold. There is a strange odor, as though of soured milk and turpentine.".

  15. A couple people have mentioned inspirational/incomplete tables as a way of forcing you out of your creative habits; and that's definitely true.

    I'll also add that they're inordinately useful for keeping the creative juices flowing after you've exhausted your own supply. Recently, for example, I had about a week to stock 256 hexes. After the first hundred hexes or so I was running pretty dry. In an ideal world I would have taken a break and let me brain turn over a couple of times with fresh idea; but I didn't have the time, and so random tables kept me fresh in ideas.

  16. @Zak, i have a question for you.

    One the problens that i have is managing tons of paper, props, handouts, etc.

    In my new campaing im trying not to use those junk, but i like the idea of using random tables.

    How do you use then?

    Edvando, from Brazil.

  17. @edvando

    I try to get all the tables I am going to need in one place.

    Like last week I cut and pasted a lot of tables all onto 2 pieces of paper then printed them out.

    Another way is to have each table printed out on a piece of paper with 3 holes punched, then put them all in a 3-ring binder inside plastic sheets.

    Another way is to buy a blank sketchbook and put all the tables in there.