[...] it wasn't until Civilization Revolution [2008] that my eyes
were opened to the full extent of people's irrationality regarding
random events.
We had decided it would be neat to display the odds of each battle
on the screen, partly because statistics are fun, but mostly to
address a particular issue that had turned into a running joke on
the message boards. The problem stemmed from the fact that there
were no guaranteed wins in any matchup--the odds might be incredibly
long, but the underdog always had a shot--and this led to the
occasional absurdity like a spearman from an underdeveloped nation
defeating a military tank in battle. [...]
But we thought maybe it would help if we showed the players their
odds before the fighting started, so they could understand that
there were real numbers behind these unlikely battles, and not just
a vindictive, petty AI.
We were wrong. Not only were they unimpressed by the long-odds
evidence, they fought back even harder on the short-odds information
they could now see.
"Sid, the game is messed up. I had this battle with a Barbarian,
right? The odds were three to one--and I lost!"
"Well, yes," I would agree. "Sometimes that's going to happen."
"No, no, you don't understand. Three is big. One is small. I had the
big number." [...]
It didn't matter how many different ways this conversation played
out, I couldn't convince our testers that it made sense for them to
lose a three-to-one battle roughly one-fourth of the time. Past
certain odds, people expected to win no matter what, but also to
occasionally prevail if they were the underdog in the same
situation. [...]
So we changed the actual odds behind the scenes, and made sure that
the player would win any battle with odds of three-to-one or
greater. This might have been unfair to the computer AI, but we
never heard any complaints, and once players were given the
advantage, they reported having much more fun.
"Sid. There's another problem."
"Uh oh. What happened?"
"Well, I had this two-to-one battle, and I lost. Which is okay, I
know we've had this discussion. But right after that, I had another
two-to-one battle, and I lost again!" [...]
Again, emotions trumped logic, and we had to accept that. So we
started taking into account the results of previous battles, and
making it extra unlikely for too many bad (or good) things to happen
in a row. We made it less random, so that it could feel more random.
"Now are you happy? Anything else?"
"Well, now, here's a really weird thing. I had a battle, the odds
were twenty-to-ten, see? And somehow I lost."
"That's . . . the same as two-to-one."
"No, two is only one more than one, but twenty is ten more than
ten. I mean, do the math!"
So we added another "correction."
I presume the testers were not exactly 8 y.o. kids. Then by what
chicanery & deceit did they manage to graduate from school?