In Praise of Randomization

Once is happenstance, twice is coincidence, the third time it’s enemy action. – Ian Fleming

Here’s two things about human beings:

1. They are incredibly bad at randomizing
2. They are incredibly good at recognizing patterns

The implication of the combination of these facts is that if you’re GMing a game and you think you’re introducing something that has no correlation with what’s gone before you’re probably wrong, and your players will probably notice.  In fact, even if you’re right, your players may well think there’s a correlation.

There are times when you can take advantage of this, and bask in the appreciation of your players who think you planned something diabolically intricate sessions in advance when actually they just constructed the pattern on the fly, but a lot of the time it can be a problem.  For one thing, it can make the players suspicious and paranoid.  Since a significant part of the game world always is out to get them, there’s usually plenty of fodder for this.  Paranoid players can turtle, or turn away (or turn on) NPCs that would otherwise be able to offer them resources and support that they’re going to need against the real threats, and they can slow the game to a crawl (such as when they examine and re-examine every ten feet of a corridor).  Unintended or spurious patterns can also lead the players on wild goose chases, pursuing lines of enquiry that you know are a dead end but are hard to block off without going meta–especially since it can be impossible for the players to distinguish between the world not containing the clues they’re looking for and an in-game adversary covering his tracks.  Sometimes the dog doesn’t bark in the night because there is no dog.

That’s where using genuine sources of randomness such dice come in.  E.g., if you have and regularly use random encounter tables to make your world seem alive and bustling, then you have much less risk that the players will conclude they’re being spied on by beggars just because beggars are a bit of local color that always seems to pop into your mind when you’re improvising.  They might still be inclined that way because of a series of unfortunate dice rolls…but if the players know you’re rolling randomly on tables, they’re much more likely to take it the way a real inhabitant of the world would: coincidence, not enemy action.  The tables are important so that you’re just using the randomizer to pick from a distribution that makes sense for the setting; you’re not trying to thwart all pattern in the setting, you’re trying to emphasize the true patterns and mitigate the purely coincidental ones.  The key to remember is that the players have hardly any interactions with the world (including simple descriptions of what they perceive) compared to the characters, so they tend to grossly overestimate the representativeness of the interactions they do have.

Finally, no discussion of randomization in RPGs would be complete without touching on random character generation.  While it’s probably obvious that random character generation, just like randomizing on encounter tables, increases the representativeness of the characters as part of the population, it’s not as clear that’s desirable.  After all, a character picked at random from the population of the setting may not have a single adventure during his or her entire life (and probably likes it that way).  For a long time I was strongly against random character generation.  Not only should players be able to play what they want, but the very fact that they’re expected to go on an adventure (barring certain kinds of everyday-people-thrust-into-peril scenarios where I tended to hand out pregens anyway) ought to be a strong enough filter to justify deterministic character generation.

I’m much less dogmatic about it now, in part because I’ve been a player for the past seven years in a campaign where the GM insists on rolling the six D&D stats in order 3d6 each, no swapping or adjusting (not even racial adjustments or 2-for-1 prime requisite adjustments, so actually more strict than Basic D&D), and I’ve come to appreciate two features of random chargen even without the fun and complication of a lifepath system.  First of all, it really does make you play characters that you wouldn’t consider otherwise and that can make things fresher and present an added challenge.  You might not want to get too attached to that 5 Dex fighter, but while he lasts it really can be fun trying to make the most of him.  Second, it makes being particularly good at something rare, worth treasuring, and a genuine stand-out in the setting.  Mechanically, an 18 STR is the same in a 3d6 in order as in a 4d6 drop low and arrange, but in one you really are the strongest person you are likely to meet in the campaign, in the other you’re maybe one-in-ten Fighters (unless they’ve gone the Dex route), one-in-three who’s 17+.  Random chargen is still not my default preference, but it definitely has its plusses.