Every GDC there’s always tons of interesting side conversations spawned by elements from the talks. One of the more innocuous comments that started a number of conversations was from Sid Meier’s keynote. He told a story about playtesting Civilization Revolution – when presenting players with simple odds, like 2 to 1, they would expect to win disproportionately (more than two times out of every three).
When they won a few times and then lost, they would be surprised. Although the probability defines that they should lose occasionally, they perceived their strength as much greater. They would expect to win even more disproportionately when the odds were presented as 20 to 10, despite the same probability. I’m sure plenty of folks in the talk were blown away by the revelation of just a handful of our deep seated cognitive biases, especially when dealing with probability.
Ken Perlin, who I happened to be sitting next to during the keynote, pointed out that part of the problem is that the system was actually quite simple. Would an army that was twice the size of a force it was fighting win at 2:1 odds? Probably not, other stuff being equal. The human brain is good at looking for patterns in large complex systems. We see them even when they’re not there. Even the simple presentation of two cartoon armies fighting suggests more non-linear factors.
Later on, Daniel Benmergui dismissed it as “player pandering” – which it is, in a glass half empty kind of way. Saying Sid Meier panders to the player is kind of like saying a professor is being academic – true, but is that automatically a bad thing?
Or, was Meier just trying to communicate what he wanted as clearly as possible (especially if the key elements to convey were not related – it helps to simplify ancillary elements). Civilization is still today the best game series that shows you can layer a little bit of information and historical context with an entertaining game. Is a child going to learn all of world history from it? No, but it has certainly inspired many kids to learn more about the time periods and historical figures in the game. If it that was a positive effect, was that still wrong? Or do the game design ends justify the game design means?
I didn’t take it as Meier saying “Here’s this problem, here’s how to solve it.” Instead I just assumed it was “Here’s this problem, here’s how I solved it.” You need your own answer to that question, but as a game designer it’s inexcusable to be unaware of how people perceive such a system and the related probabilities.
It’s inherently a player communication problem – the representation given to the player did not match the model suggested by that representation. You could have solved it by changing what is shown to the player (abstracting the notion of strength in a way that aligns better with their expectations of the odds/stats given), or you could make the model more complex to match the one suggested. Meier added probability based on previous events to line up the results with player expectations. You could just as easily add more rules to the system to model non-linear effects (soliders could have a morale bonus in that situation, which would boost the odds even more).
Frank Lantz had the most inspiring summation, which seems like the best ideal to hold ourselves to: “Good games use players’ cognitive biases. Great games make players aware of their cognitive biases.”
All from a little math.
The problem seemed to be that Meier’s title for his talk (“Everything You Know About Player Psychology Is Wrong”) implied that he was going to present The One True Way, when of course he was just talking about what worked for Civilization. I don’t know if he’s aware of the controversy he caused, but my impression of the talk was that he was just giving examples from his own experience and not saying that it was the only way to do things.
I find it really interesting and a little frustrating that the talk Rob Pardo of Blizzard game talked about the EXACT same issue about players’ perceptions of randomness, and how Blizzard adjusts their probabilities to work with how players THINK randomness works instead of how it actually does. Nobody seemed bothered by this talk.
That’s true about Meier’s title – I took at he was just sensationalistic for fun, but I think people did walk away with his approach as a guiding principle. I also heard that similarity w/Pardo’s talk (I didn’t go to it though).
Yeah, it was a bit frustrating for me just how many people didn’t already have some grasp of that problem, but as an AI programmer I’ve had to deal with this all the time so it’s hard for me to judge how widespread it should be. Coming from that background, too, their solutions are the lazy solution – you’re just avoiding making the model deeper because you think it will be harder to communicate that than some fuzzy probabilities, but you miss opportunities for a richer experience (and, counterintuitively, it often ends up being easier to convey to player when you’re adding rules to the model than when you’re adding a range of probabilities).