Modelling ecosystems with EcoBuilder

Chris Berrow talks to Jonathan Zheng, who developed EcoBuilder. Can it be used to help model ecosystems?

Jonathan - In Yellowstone Park in the U S in 1926, the wolves were hunted by humans who didn't realize, you know, the ramifications of their, of their actions. Because they went extinct, it meant that the elk in the ecosystem didn't have a predator anymore. They started eating all the plants and then there weren't enough plants for the small animals. So the entire health of the ecosystem started deteriorating. And so my PhD supervisor, he watched a documentary on this and he thought, tthis is a cool story, but it would also make a cool game, because you could have your own simulation and then you could add your, you could decide to add the wolves to yourself.

Chris - More people that play and come up with different strategies. I guess you can assess which ones are successful and they could contribute to some genuine discoveries here?

Jonathan - Yeah, I actually, it already has. I just finished up my PhD thesis a few days ago and, um, the top three players and each of the last few levels of the game, uh, they, they got, I mentioned in my PhD thesis,

Chris - It's like the ultimate leaderboard. You know, when you played the arcade machine and you'd write your three letters, this is a different way of cementing your status!

Jonathan - And you actually get to help science as well.

Chris - So you took some of the real life models that people are using to create predictions. How accurate are they though? Because if they are slightly out, then the game will be slightly. 

Jonathan - That is a good question, actually. So they've been successfully used to model aquatic ecosystems actually. But you are right that they are the models of really complex. So they're difficult to kind of get, get the parameters right. And that's actually one of the tasks that we're asking the player to do. We're asking them, how many of these heavy animals can you add before the ecosystem collapses.

Chris - Is it better to introduce the wolves or is it better to not touch anything? And just to let it, you know, nature correct itself in a way?

Jonathana - One thing that is clear from these mathematical equations is that nature does fix itself, but it can't fix itself. If changes happen too quickly, if humans hunted all the wolves to extinction within a number of years, then the ecosystem can't really react that quickly because evolution happens on timescales that are too large

Chris - Games, usually... usually have to be fun to play. So how did you also make sure that this was actually a fun game to play, that people would want to come back and try and beat the final levels in the end?

Jonathan - Ah, yeah. So that was actually a problem that took a lot of thinking because the game studio, when they design a game, they design the simulation to be fun, whereas we were stuck with our simulation. Um, so it's a weird like reversal of the order in which you'd usually do things. Um, on the way we did, it was. People want to collect stars and it took a while to design all the levels that are in the game to be kind of puzzles that aren't too difficult, but are still kind of that they still have this light bulb moment where once you solve that, you realize, oh, I understand something now. That was especially difficult with our simulation because you know, you have to work around the mathematics. And obviously me and my team, me and the team had to understand the mathematics as well before we could, um, make the levels kind of the right balance between difficulty and challenging.