The Awful Green Things From Outer Space and Experimental Design
When I was young – and readers might find this difficult to believe – I was pretty geeky. When I wasn’t watching Star Trek or reading Isaac Asimov novels, I was not infrequently playing some board game or other with my similar-minded friends.
This is, in fact, going somewhere, though, full disclosure, this post has little to do with evolutionary psychology.
The Znutar
Anyway, one board game that I liked was called “The Awful Green Things From Outer Space.” The game took place on a spaceship, the Znutar (pictured here in a fantastic 3D production) which was invaded by, as the name of the game suggests, Awful Green Things. The AGTs had a three-phase life cycle, growing from eggs to juveniles to adults; adults were tough, bipedal one-eyed heavies.
The crew of the Znutar (which consisted of Snudalians, Smbalites, Frathms and Redundans) could attack the AGTs unarmed but, really, the AGTs were too strong for the crew, generally, to take on mano-a-mano. So, the crew could also pick up various artifacts to use as weapons. This could include cans of Zgwortz (the crew’s source of food), welding torches, a fire extinguisher, and so on. The problem was that because the biology of the AGTs was unknown, it was impossible to foresee the effect that a given weapon might have on the invaders. So, whenever a crewmember used a weapon for the first time, the player had to draw from a random set of effects that the weapon could have. The weapon might harm the green thing, have no effect, or even benefit the green thing, though thankfully for the crew this was unlikely. The fact that weapons had randomly assigned effects meant that the game was different each time one played.
Members of the crew in an Awful predicament
It also meant that to find effective weapons, you had to run experiments. So, suppose you’re the crew player, and you have crew member Sparks use the fire extinguisher on a room of Awful Green Things. You draw one of the chits from the pile, and discover that the fire extinguisher Shrinks the AGTs one life history phase. Great! A useful weapon. You have learned something valuable that will be true for the duration of the game. Every time you use the fire extinguisher in the future (again, in this particular instance of the game) the fire extinguisher will Shrink any Awful Green Thing on whom it is used. (Adults become juveniles, juveniles become eggs, eggs become poached. No, seriously, the eggs just die.)
However, now suppose that two of your crew members gang up on an Awful Green Thing with previously untried weapons. One uses the fire extinguisher on it, and the other throws a can of Zgwortz at it. Now you draw a chit from the pile, and you learn the effect of both attacks, jointly, on the Awful Green Thing.
Here is the problem. As an empirical matter, all you now know is what these two weapons do jointly. You don’t know what the fire extinguisher, by itself, does and you don’t know what the can of Zgwortz, by itself, does. If you use either one again in the future, by itself, you have to draw a chit to determine its effect. By using both at the same time, you have learned nothing about what each one does, individually. Your experiment might have been useful for this one attack on the Green Thing, but you haven’t learned anything useful about each artifact by itself.
And that is how, though I didn’t know it at the time, Tom Wham and Steve Jackson introduced me to the idea of the problem of introducing experimental confounds into a design, to say nothing of the problem of interpreting main effects when you have a significant interaction term.
Long live the crew of the Znutar!
Comments (6)