Episode #64 is live!
Jon and Dirk are joined by Soren Johnson once again, this time to discuss conflict in games. Topics covered include: What is a conflict? Why is its representation in games nearly always violent? What other types of conflict are there, and how can designers best utilize them?
Most games prominently feature conflict of some sort so this topic was probably long overdue. As such, we covered a lot of ground from different types of conflicts (military, economic, diplomatic, political) to how conflict plays out during a game (symmetric, asymmetric, lopsided).
I’ve long wanted to see a good game about political conflict, but the sheer immensity of this task is hard to overstate. Humans are naturally very attuned to the traits and mannerisms of other humans. Most of us are familiar with the disconcerting ‘uncanny valley’ resulting from a close-but-not-perfect visual representation of a human, but less obvious is the existence of an uncanny valley for behavior as well.
Over the past few years I’ve become more and more sure that it’s impossible for a game AI to come anywhere close to passing the Turing test. The task is simply too demanding, as even the tiniest flaw will be noticed and judged mercilessly. On top of that knowing ahead of time that your opponent isn’t a human might make the task actually impossible. We’re wired to regard human behavioral quirks as ‘personality’, but computers are not afforded that luxury – for them it’s simply random or flawed logic. Nearly all of us understand computational mathematics at a basic level, but human thoughts are inscrutable and forever unknowable. At least, that’s what most of us believe, and that’s all that really matters to pragmatic game designers.
Is it possible to make a game that roughly matches what we think of as politics or diplomacy? Sure, but it’s not the same. Representations of war in a game can truly feel like fighting a war. Armies take casualties, terrain is captured and lost, a daring flanking maneuver can turn the tide of battle. The same is true of economic simulations, where amassing a massive pile of eMoney feels very much like having a big wad of real-world cash in your real-world pocket.
But when it comes to representing human interaction we have no choice but to veer into the truly abstract. I’d say that King of Dragon Pass and Crusader Kings II are the titles which have come closest, but they work because they construct a fairly rigid framework of decisions around the human player rather than trying to make their AI characters seem truly alive.
I have no doubt we’ll see impressive breakthroughs in academic AI in the coming years, but I do doubt those achievements will ever trickle very far down into gaming. It’s not a matter of feasibility, but economics. AI programming is quite difficult and the vast majority of players are just as happy with a mediocre AI as they would be with a nearly-sentient one. I’ll still be holding out hope though, as success here will mean nothing less than a revolution in game design.