Month: June 2011

  • Thinking… Metaphorically

    I’ve been trying to tease out the constituent parts of human problem solving for a little while now. This is the kind of topic you’d think you could just find a couple of books to read or fish out a couple of papers to get some insight. Unfortunately my searches aren’t bringing up as much as you might think. Although I will heavily recommend Jeff Hawkins’ On Intelligence. His book is still pretty theoretical but it has a lot of great ideas that have been slowly sinking into my thinking.

    Nevertheless I have recently contrived a crude theory of human problem solving in my own head. I’ve come about these ideas from reading a few books, some introspection, the rope experiment I mentioned last time, and watching a ton of playtesting of puzzle games. It’s probably shallow, wrong, and not worth anything, but I’m going to subject it to you anyway.

    I think there are two major components to our problem solving.

    • Permutation
    • Metaphor

    The first one is easy. Permutation just means “trying everything”. I have believed permutation to be super important for a few years now. I don’t think the idea of just trying every possible action is going to shock anyone but from playtesting puzzle games I think you’d be surprised just how important permuting is to our problem solving. A lot of puzzles are solved by permutation alone. When you’re just “playing around with something” you are permuting. When you solved one of those “get the metal loop off the rope” puzzles and it just falls apart in your hands you are solving by permutation. When you are thinking out every possible outcome of a chess move you are permuting. Permutation is our go-to workhorse for getting things done. To do it all you need is two hands and a memory. It’s super versatile (you can apply it to pretty much any problem, regardless of how diverse) and it’s generally pretty effective, if laborious.

    So permuting is an obvious part of problem solving but it’s clearly not the whole story. It doesn’t explain how people make leaps of intuition, and it doesn’t explain how people play games with very large solution spaces like Fantastic Contraption. In Fantastic Contraption permuting is very hard. You can put a stick or a wheel anywhere and connect them to anything. It would take you days to solve a problem by just trying every possible machine. Luckily, we have a wonderful gift for metaphor.

    By a gift for metaphor I mean we have the ability to see similarities between seemingly disparate things. Consider this typical metaphor: “an angry man is like a simmering pot of water” (that’s a simile, which is a kind of metaphor). I argue that we are using the familiar pot of water to help understand the more rare angry man. We take some traits shared by the two objects: that they could both quickly convert to a state of painful unpredictability. At the same time we don’t get confused about other qualities of boiling water and anger that are not related. We don’t think of the man as being wet or becoming a gas, for example. This is second nature to us, but if you were to try to write a computer program to do this you’d quickly see how magical the talent is.

    We can use this ability to see similarities between new problems and previous problems. Consider the rope problem from my previous post. A number of people wrote me saying they thought of the rope solution first. In fact I think more people come up with the swinging solution when the problem is posed as a thought experiment than when they have the ropes in front of them in person. I’m going to guess that this is because it’s harder to permute in our heads so we jump more quickly to our metaphor engine. Then we recognise the similarity of the rope problem to a tire swing, or a clock pendulum, and we’re off to the races. Metaphor also helps explain mastery of a game. As you improve in chess, for example, board states start to become analogous to eachother and so the outcome of previous games start to inform future games.

    Permuting also helps us discover new metaphors. By playing with things we are always seeing them in a new light and there is an ever increasing chance that we will see a sudden similarity between the current problem and something we’ve seen before.

    In this way Metaphor is your ability to take what you understand about one problem and apply it to a new problem. That is a super powerful talent that we all have. I believe it’s this ability, along with just trying random stuff, that makes our species so adaptable and so good at videogames.

  • How Complex is BrainSplode!?

    Playing BrainSplode! in Honduras

    When we were in Honduras last year we had a pretty crappy Internet connection. It was pretty slow and we had to pay for our bandwidth by the meg. When you pay by the meg suddenly podcasts and torrents are less fun. Fortunately there were a few indie games I sucked a lot of fun out of. They kept me entertained for days and all they asked for was a few megs of bandwidth.

    One of those games was BrainSplode! by Rich Edwards. BrainSplode! isn’t even a proper game. It’s just a prototype. But it’s so good it drives me crazy that Rich is continuing to prototype stuff rather than just double down on BrainSplode!.

    You can think of it as a game about programmable howitzer shells. I like it for a couple of reasons. One is because it is incredibly, rediculously fun. Another is that you can so easily enumerate the complexity of BrainSplode!.

    I will give a very brief description but you should really just go play it. Brain Splode! starts off as a very familiar ballistics game ala Scorched Earth or Crush the Castle. But it mixes in some Roborally/SpaceChem style programmable elements. Namely, you can chose to change the direction of the shell, fire off a booster rocket, pop a parachute, or any combination of these three actions at any time after the shell is fired. You do this by lining up three ‘actions’ to take before you fire the shell and then activate them in turn by pressing the mouse button.

    As an example, I can fire the shell high and to the left, then I can make it face backwards, then I can fire off a booster rocket which sends it flying to the right and then pop a chute to slowly glide towards my final target.

    Manually calculating ballistics trajectories

    BrainSplode! has something like 6 variables to play with. Two for the cannon, one for how I program the shell, one for how I set the direction changer (assuming I use one) and then another couple for when I choose to activate each action.

    Since a lot of these variables are along a continuum and not discreet choices it’s hard to enumerate the total number of options available but we can easily see that the solution space is huge. In fact I think it’s too big. BrainSplode! is yet another game in a long series of games that I love but utterly fail at convincing my friends to play. I pestered everyone I knew and almost no one else finished BrainSplode!. I think that has to do with the complexity of the solution space.

    There is some good empirical evidence that the two variables of the ballistics game on their own represent a comfortable level of complexity (Angry Birds). So I think BrainSplode! is stumbling outside the optimum complexity fun-zone. As an interesting experiment I’d like to try pushing it back into the fun-zone. Here’s how the experiment goes. You can play along at home.

    1. Download BrainSplode!
    2. Beat BrainSplode! normally.
    3. Go back to level 7 and set the cannon to shoot up and left with minimum power (put the power meter in the very top left of the square). Now beat level 7 without moving the cannon.
    4. Do the same for level 6.

    You have to beat the game first because steps 3 and 4 are very hard (since the game isn’t designed with them in mind) and if you didn’t the difficulty curve would be way out of whack. By holding the cannon variables constant we push BrainSplode! back towards the optimal-fun zone where the complexity is more manageable. I think as players and designers we intuitively understand when something is too complex or not complex enough, but only after the fact. We pretty much have to build something and play it before we know if it needs more or less stuff glued onto it. Worse still after we’ve been living with a game for a while we lose perspective on the optimum complexity: “I have no problem understanding the game I’ve been making for 4 months, I don’t know why other people are having trouble”.

    I’d love to understand the relationship between complexity and fun better. Mostly I find it an impenetrable fog. I’m always trying though and BrainSplode! is probably the most fun experimental complexity playground I’ve yet to find.

     

    p.s. if you’re really looking for a BrainSplode! challenge I have two more for you:

    – Try level 7 but with only one set of programed commands. You can use parachutes and rockets and everything, but you can only set them once  at the beginning and never change them. This includes the green change-angle power. You can use it but you must chose one angle at the very  beginning and never change it. You also have to leave the cannon locked in one place, but you can set it anywhere you like. No changes to  anything! (if you play RoboRally then think of it as having all your registers locked).

    – Then try level 6 the same way.

    I’m pretty sure I’m the only one in the world to get these ones. Even Rich shied away from these challenges. Happy ‘Sploding!