Tales of the Rampant Coyote

Adventures in Indie Gaming!

Randomly Speaking

Posted by Rampant Coyote on November 9, 2012

Craig Stern of Sinister Design had an insightful essay on randomness (or, rather, non-determinism) in games this week.

I thought I’d talk about it a little bit.

I’m a big fan of Go (I even dropped a serious chunk of change on Many Faces of Go – which was well worth it, IMHO). Chess a little less so, but I do enjoy the game (although I suck at it). When I was a kid, one of my favorite games was Mastermind.  These are all deterministic games with no artificial randomness.  But for me, RPGs have always been joined at the hip with the idea of slinging dice. So I’m very reluctant to go full-on deterministic with them. And Craig, too, is finding some places where gameplay can be expanded with some careful addition of some random (or at least, less predictable) elements.

In somewhat random order, here are some thoughts on randomness in RPGs – particularly computer RPGs:

Asymmetric vs. Perfect Information

In my Artificial Intelligence classes in college, we tended to talk about “perfect knowledge” or “Perfect Information” games – particularly Chess and Go – with respect to AI. Perfect information means all players know the game state and all possible moves. Nobody’s got any cards they are hiding. I *believe* this also means than the results of any move are also known in advance, so all moves must be deterministic, but I may be wrong here. But an otherwise fully deterministic simulation can be made a lot more interesting through asymmetric information. The players don’t know everything. Even in a straightforward RPG, encountering an enemy without fully knowing their powers — and having to play cautiously while figuring out what they’re capable of – is an example of asymmetric information. Ideally, the AI wouldn’t automatically know what your party is capable of, either. I was always suspicious that even the dumb animals in Knights of the Chalice automatically knew who your vulnerable spellcasters were. But even a simpler possibility here – with perfect knowledge of the enemy position and composition – would be to have each player hold three special move cards that the opponent can’t see… whether these were obtained randomly or ‘purchased’ before a combat.

This is another avenue for making a deterministic game less predictable.

Reality is Random

I was stupid and not paying attention when driving a few days ago when making a left turn. I nearly got into an accident. It wasn’t a near miss or anything, but I felt really stupid about it. I have no idea how many left-hand turns I’ve made in my life, but it’s a lot. These are not difficult actions. I don’t go into an intersection expecting a 1% chance of failure when I turn on my left blinker. But even trivial acts go wrong sometimes. And sometimes we get lucky and win the lottery.  There are so many complicated factors in play with every action and plan, so many uncontrolled variables, that our human minds effectively have to boil things down to chance. And we have to make plans around chance. We buy insurance. We take risks. We play the odds. We create “Plan B.”

I’ve done a bit of simulated combat with fencing weapons, padded swords, martial arts, paintball, and the like.  Enough to know that what seems simple and automatic in practice can become devilishly hard when the heat is on. There are plenty of reports on actual gunfights (and a few eerie videos on YouTube) that illustrate how ‘shootouts’ are amazingly… well, random. Guys who can place high in shooting competitions can’t seem to hit the broadside of a barn at point blank range against a deadly opponent shooting back at them.  Real fights are fast, frantic, messy affairs, and gamers would be infuriated with probabilities that mirror reality. Actually, older wargames (and D&D) tried to simulate this by claiming that attacks were an abstraction. In older editions of D&D, your hit chance represented the sum total of a number of attacks made during the course of a round, perhaps the one “good” shot that had a real chance of doing damage.

Simulationists (and I have at least one foot in that camp) tend to enjoy the randomness of real life thrown into our games.  Though we also like having enough control over the variables to be able to pick and choose our chances.

It’s Fun to Gamble

I forget the studies involved (it’s been a long time), but it’s been demonstrated that anticipation and excitement over an unpredictable result is much higher than for a predictable one.  Putting coins in a slot machine is a lot more fun than buying something from a gumball machine – even though your average return on investment from the gumball machine is higher.  Our most memorable moments in our “dice & paper” RPGs often come from moments where we defied the odds – either with a positive or negative result. One of our favorite stories involve a “boss” vampire going down at the beginning of round 1 with a combination of a good initiative roll on the part of the rogue, an Arrow of Undead Slaying, and a bad fortitude save on the part of the vampire. Yes, it ruined my big, high-level boss encounter. Yes, it turned a big climactic fight into almost a joke. But the players LOVED it. So what’s the problem?

Another – negative – result came from a monk in a bad melee who encouraged the spellcaster to launch a fireball on his position. “Don’t worry, I can take it,” he promised. But his enemies, he was sure, could not. After all, he was 80% sure of avoiding all damage from the fireball entirely (improved evasion), and even on failure, would take only half damage, and had a few more hit points than the spell would do *on the average*. Worst case, he could be healed and prevented from dying at -10 hit points once the fireball killed all of his opponents. You can probably guess what happened.  The damage was extremely high, and the saving throw roll was extremely low. The monk was reduced to -11 hit points, killing him immediately. The player, fortunately, took it in great humor, rolling up a new character and laughing it off. We all do, now, too.  It’s still a running joke.  It was one of those great moments in gaming.

And randomized treasure? Seriously, sometimes I think that’s the only reason Diablo ever caught on. Randomness makes a game interesting.

Randomness Facilitates Friendly Competitive Gaming

It’s sort of a given in game design that randomness plays an important role in friendly competitive games. It allows players of unequal skill to play together,  and gives all players a chance to write off losses as “bad luck.” On the extremes, you have games like Candyland that are pure luck, allowing children to play with their parents and have an equal chance of winning. By contrast, beating a more skilled player at chess may lead to the suspicion that they deliberately threw the game.

Too Much Randomness (or Lack of Control Over Variables) Irritates the Hell Out of Players

One of the discoveries I made in the making of Frayed Knights: The Skull of S’makh-Daon was that the earlier incarnation of the game system – which relied on bell curves to deliberately be “less random” than your average pen-and-paper RPG – was extremely frustrating due to its randomness. The later version was considerably less random, yet arguably still too dependent upon random numbers in the end. I have a few theories why the level of randomness acceptable in D&D feels less so in an RPG like Frayed Knights, and more plentiful combat is certainly part of many of them.

The risk of failure is fun. Actual failure, considerably less so.  Playing in “ironman” mode with permadeath turned on in a game is – for me, at least – at least twice as thrilling as a normal game.  But it’s a hundred times more upsetting when you lose.  Which way those balance may depend upon the game and you, as a gamer. For most of us, the higher the risk, the more control we want over the variables. When a split-second of poorly timed lag (a certainly undesirable form of randomness) can yield perma-death, we’re not going to be happy.

I thought the drama star system would help offset some of the frustration with “bad luck” – and I believe it did – but as I’m tweaking the system for part 2, I’m working do diminish some of the ‘spread’ a little more.

A Happy Medium

Anyway – I enjoyed Craig’s essay, and it also amused me. He’s slowly getting sucked back into putting more non-deterministic elements in his games. I’m feeling compelled to reduce some of the randomness in some areas of mine (while possibly expanding  it other, more interesting areas…) I think it all comes to there being a happy medium in there somewhere. Of course, it varies by game and play-style so it’s hardly universal. Or predictable. You might even say it’s kinda random.



Filed Under: Design - Comments: 10 Comments to Read

  • WhineAboutGames said,

    I’ve had one or two people wish that LLTQ had a random mode, where you could never know what was going to happen and just had to face whatever challenges were randomly generated with the skillset you’d chosen to build.

    At the current level of difficulty, though, that would be just plain unplayable. No one could ever win… okay, not no one, but perhaps a more nethack-like percentage?

    (I think some of that is, deep-down, just a request for a longer game with more stuff to encounter.)

    I’m personally pretty happy with the way a game designed to never be random ever worked out. It is still something of a gamble for most players because the consequences are sufficiently complicated that they can’t foresee everything that’s going to happen.

  • Armaan said,

    Great counterpoint to Craig’s article. As with most things, having a balance between complete determinism and randomness is probably the best way to go.

    I think the reason highly random gameplay of traditional D&D works is because with D&D you’re not really playing to win. You’re playing to have good times with friends. So when your character dies from a bad roll, you can all laugh at how “that didn’t work out like I expected”, reroll a new character, and then keep playing. Even if you fail to win the module you’re playing, it’s no big deal because winning is not really why you’re playing.

    But with a CRPG, you are playing to win (unless you’re playing a roguelike). If your character dies, you’ll often end up losing a lot of progress toward that goal. Not such a big deal in a roguelike where death is expected, but definitely annoying in a story-based game, especially when you’re heavily invested/interested in the story.

  • Felix said,

    That’s the gist of the problem: randomness is all right, but lack of control is not. In real life, we can mitigate risk pretty damn well with preparation. In RPGs, all too often we can have the best gear, excellent attributes, tactical advantage… and get flattened by a mouse with a lucky dice roll. Life’s unfair, you say? Well, I have different expectations from a game.

    In my abandoned attempt at a roguelike, what I did was keep the number of dice rolls at a minimum (and yes, I went with bell curves too). Every combat turn, you roll once for attack and once for defense. Everything else is modifiers — but there’s a lot of them, and they can vary in hard to predict ways. That kept things interesting without them being arbitrary. For example, a snake can’t get lucky and kill you, but it *can* get lucky and poison you. That *also* shouldn’t be enough to kill you outright, but it will force you to avoid the next critter instead of fighting it… and then you’ll still have to eat for natural healing to do its job. You can screw it up, but it will be your own fault. Or you can get really really lucky, but then all you gain is getting past obstacles faster.

    So that’s my solution: enough randomness to spice things up, but not nearly enough to make tactics useless. Enough factors to make things varied, but not to obscure the workings of the game. Complexity, not confusion.

  • Picador said,

    I have a few theories why the level of randomness acceptable in D&D feels less so in an RPG like Frayed Knights, and more plentiful combat is certainly part of many of them.

    I assume that this is part of your aforementioned theories, but it seems obvious to me that randomness in games creates anticipation (and therefore enjoyment) only when there’s a ritual around it. Rolling dice (at craps, or in a tabletop RPG), waiting for the roulette ball to settle in place as the wheel spins, seeing each of the three slot machine pictures come up one at a time… these rituals create suspense.

    Nobody would play a roulette machine that took your bet, then popped a message up on a display screen after a second: “You win” or “You lose”? Dice, roulette wheels, even the mechanical components of slot machines: these give a focus for our attention and anticipation in a way that behind-the-scenes calculations do not. The strength of computer RPGs in calculating and modeling the virtual world behind the scenes becomes a liability in creating anticipation based on randomness.

  • Rampant Coyote said,

    Actually, Picador, that’s not something I thought of before. New food for thought! 🙂

  • Andy_Panthro said,

    Of course with computer gaming, it’s quite possible for the game to tell you one thing and for the result to be something quite different.

    For example, in XCOM, you could easily make the game display an 80% chance to hit, while behind the scenes making it 95%. This would fit better in most people’s minds, because missing at 80% feels unfair (even though it’s only a 4/5 chance).

    And now I’m wondering what change in behaviour you might see if you referred to it as a “20% chance to MISS” rather than an “80% chance to HIT”. Would people become more cautious?

  • Darklord said,

    Interesting as always. 🙂


  • Noumenon said,

    Gumball machines are totally better than slots.

  • CdrJameson said,

    I like Picador’s ritual point – The computer version of board game Space Crusade had a dice-rolling ritual for deciding hits and it certainly ramped up the tension.

    I’d add that in pen & paper games the mechanics are all exposed – you know what will give you +1/-1, you know the probabilities on 2d6, you (generally) know what your enemy can do.

    In CRPGs the desire to make it easy to play, to handle all the ‘unpleasant’ bookkeeping often means you haven’t got the foggiest what the underlying rules are, and thus how to make a well informed decision.

    Briefly mentioned in Craig Stern’s article: Randomness in AI response really, really pays you back. It gives a great illusion of intelligence.

  • Good randomness, bad randomness « No Time To Play said,

    […] writes about board games and what we can learn from them, Jay “Rampant Coyote” Barnson plays devil’s advocate a little — an important counterpoint, as it turns […]