Recent 10
Post:
Post: Pixel Shape Generators for Minecraft
Photo: Minecraft in Coffee
Photo: Mr. Snail
Post: Quincy 2009 Election: The Levy Limit
Photo: February Snowstorm
Post: Quincy 2009 Election: Tale of Two Taxes
Photo: Sailors' Home Cemetery, Quincy
Photo: Yawkee Way Banners
Random 5
Photo: Waiting, for the Bus in the Snow
Photo: Nickerson Beach and Boston Skyline
Post: Change is Here
Photo: Pedestrian in Snow, Washington Street
Photo: Red Tree in Bleak Landscape

Myst and Science: A Prelude to a Braid Review

When first I played Myst I said to myself, "Somebody gets it. Somebody has seen the true value of the computer for making puzzles." But let's skip over Myst for a minute, I want to talk about science.

What is science about? In the United States, the school curriculum focuses very heavily on the scientific method, the cycle of Observation, Hypothesis, Experiment, and Analysis used to answer questions about our world. Across all grade levels, students are given a question, asked to hypothesize the answer, and an experimental procedure and equipment to answer the question. Repeat adinfinitum . Here is the problem: the scientific method is such a small part of science, such a tiny step in the whole process, it is wildly misleading to focus as hard as we do on it. It is like trying to teach computer programming by focusing only on keyboarding skills. You can never see the forest if you focus only on the trees.

Science is really about models: model building and model testing. Have a facet of the world you want to explore? First, you build a model of how it works that explains any observations you've made so far. Most importantly, you have to make one that makes testable predictions1. Only then are you ready to use the scientific method to try and match those predictions with observations. The experiments themselves are the baby steps of science2, the steps that lead us to the larger goal: refining our vision of the world around us.

What does this have to do with puzzles? The fun part of science is the interplay between the building of understanding and the testing of that understanding. Almost all puzzles hand you the first part and expect only the second. They tell you the rules of the game, all of the constraints, and the goal, and your job is to figure out how, within those rules, to get from point A to point B.

For mechanical puzzle designers, this approach is mandated by the fact that they are making physical puzzles in the real world. The maker of a blacksmith's puzzle cannot say "Well here are the rules of physics for this game!" and would be hard pressed to hide the goal or inner workings of the puzzle from a player3.

But what about computers? Computers can function as a total black box, hiding all of the mechanics of anything you can imagine and visualize on a screen. Myst is the undisputed early master, in my mind, of using this great gift to puzzle designers. In Myst, the player is given nothing. No goals, no idea how the fantastic-looking world they see works, just a series of objects to interact with. The player must explore, posit, and test constantly, slowly building up models large and small. At the highest level, they must figure out the ground rules of the Myst universe, and what happened on Myst island. But more dear to me are the individual puzzles, and to prevent my spoiling people who've not played Myst I'll just give an early and extremely simple puzzle as an example.

In the clock tower is a machine with three numbered discs and two levers, with another level off to the side. From another puzzle, the player knows he or she has to spin the discs to make a specific three digit number appear. And so the player begins to play with the machine, pulling the levers and watching which lever rotates which discs. At some point, the machine will stop responding until the third lever (which resets the machine) is pulled, and the player realizes that the puzzle has a time limit.

I've watched several people do this puzzle for the first time, and what follows is the fun part. At this point most assume the puzzle is now solvable4, but it isn't. The player is actually still missing information, their model is incomplete. When they realize that they can't solve the puzzle, they go back into experimentation mode: trying to find out what else the machine can do. Through lots of prodding of the machine, they'll eventually figure out that the levers can be held down, not just pulled once, to rotate different discs. Finally, the puzzle is solvable and actually almost trivial now that they player is equipped with an accurate model of the machine.

A quick list of how much the player has to discover and understand to solve even this simple puzzle:

  • What the desired end state is to solve the puzzle (discovered by solving a puzzle elsewhere)
  • Which discs the two levers rotate
  • The time limit
  • The reset lever's function
  • The ability to rotate different discs by holding down levers

Many of the game's puzzles are similar in theme: an isolated system with unknown rules, all of which need to be discovered to solve it. Most are like the example, 90% trying to understand the mechanics and 10% actually solving.

To play Myst is to do real science to explore an imagined world, and that is pretty darn neat. It's just a pity that most game designers don't use the power of the computer in this way, to create interesting worlds and puzzles and initially hide their inner workings from the player.5

As a preview for Wednesday: Braid pretty much hits it out of the park, in the same way that Myst did. It gives you a beautiful world with imaginative physics for you to explore, and has a well told story as icing on the cake.

Footnotes

1 That idea is summed up pretty brilliantly by my undergraduate electromagnetism professor in the following dialogue with our discussion group:

Professor: "I have a theory that electricity is carried via invisible faeries. Prove me wrong."
Students: "Well, we cannot see any faeries."
P: "The faeries are invisible."
S: "Well we can detect them by watching for mass changes in materials as current flows"
P: "The faeries are massless."
etc.

In the end, you end up with invisible, massless, totally undetectable faeries. The point is: if a theory makes no predictions to test, what value has the theory? The faerie theory cannot be refuted, but it also says nothing concrete about the physics world that we can use. People with bizarre theories like Mr. Time Cube bristle when their work is called pseudoscience, but the lack of predictions is why they aren't doing real science.

2 I suspect experimentalists would be very peeved if I didn't mention that their job is extremely hard. The difficulty of designing an experiment that hones in on what you really want to test and gets a conclusive result is incredible. My point here is that it is only half of the picture.

3 The few puzzles that can hide their mechanisms from the solver have to do so in very clever ways. A favorite example of mine is two concentric disks, one placed inside the other. The inner disc contains a half dollar, and the goal is to liberate it from the outer ring. What the user cannot see, however, is that the two disks are connected by pins which are held in place with springs pointed inward. To remove the coin, one must spin the puzzle with enough velocity to throw the pins outward against the power of the springs and quickly pull out the coin.

4 Unless they really do a lot of pre-thinking before playing with the machine much. Everybody I've watched, though, goes right into trial and error. Who wants to think hard when you can do something fun like pull mysterious levers to see what happens? I sure went right into pulling levers like a madman when first I played.

5 Someday I'll write about the games that do worse than simply being boring adventure and puzzle games: they have inconsistent logic. Games with inconsistent internal logic actually punish the player for trying to learn, and there are a surprisingly large number of them.

Comments? E-mail me at: "me" at this domain (adammarquis.com). Home