Nick's Café Canadien

Civilization veteran Soren Johnson, one of the foremost designers of strategy games and AI today and certainly one of the best writers on the subject, often remarks that the theme of a game is not to be confused with its meaning (slides here). Diplomacy may cast its players as the great powers of pre-1914 Europe, but it’s about simultaneity. StarCraft may put you in charge of Heinlein-esque space marines and alien civilizations, but it’s about asymmetry. If the theme and mechanics harmoniously cohere, then the mechanics can shed light on the theme in the way that art sheds light on the world. Pre-war Europe is an intriguing setting for Diplomacy because in all their backroom double-dealing, the empires didn’t take turns. Aliens are a good fit for StarCraft because you can map anything onto aliens, be it the collectivist swarm-by-numbers ethos of the Zerg or the judicious high-tech investment of the Protoss.

I am partial to this view, predominantly for reasons of aesthetics. If we are to conceive of game design as an art form, it does not suffice to decompose games into the artistry of constituent parts—the music, the models and sprites, the cinematic sequences, on rare occasion the writing. The aesthetics have to come from the specific properties that make something a game, whether it is played with a board and dice, a deck of cards, or a mouse and keyboard—and those properties come from the mechanics.

But that’s neither here nor there; I won’t elaborate today. Instead I want to turn to my favourite of Johnson’s examples: the evolution game. For your fill of Darwinian game mechanics, look not to Spore (which Johnson worked on), a game that is nominally about evolution from microbe to intergalactic juggernaut, but is actually about special creation. Back when I first played it, I wrote, perhaps a tad generously:

Let’s not bury our heads in the sand: by placing creature design into the player’s hands instead of leaving it up to random mutation, Spore inherently owes a lot to intelligent design. There’s still room for a real game about evolution in the Darwinian sense, where you set certain environmental constraints and preconditions, let a species run loose, and see if it survives in an ecosystem full of other models—kind of like how some engineers pit robots in mortal battle, but with adaptation.

Spore is a lot more creationistic than I gave it credit for; consider that the functional components of your custom-made species—the mouths, the horns, the flagella—are interchangeable parts from a specified, modular set, which is precisely what we would expect from a designing agent but not at all what we would expect from natural selection. But never mind all that. The evolution game exists, says Johnson, and it’s called World of Warcraft.

I would contend, however, that the Darwinian features Johnson ascribes to WoW are equally prevalent in most games with competitive and highly interactive player populations, provided there is sufficient strategic depth worth talking about. WoW is an evolution game because its core mechanic is community. Where there is a community of players and a developed metagame of optimal practices, strategic decisions are memes that compete for survival. Let’s call them strategemes.

Strategemes include everything from chess openings to Scrabble vocabulary: they are transmissible units of knowledge that players learn, study, and adopt—and crucially, copy. Copying them is not seen as unfair, but as an advantageous and often essential behaviour. They leave room for mutation, and we can perceive a frequency distribution of variations over a population of players and games.

But where does natural selection come into play? Let’s look at the exemplar we get from Johnson: the WoW talent tree.

Insert coin to continue… »

This week’s selection: The Immortal Game: A History of Chess (2006) by David Shenk.

In brief: The book’s alternate subtitle—”How 32 Carved Pieces on a Board Illuminated Our Understanding of War, Art, Science, and the Human Brain”—offers a hint of Shenk’s scope of thought. Full of colourful stories and painstaking research, this thoroughly accessible work probes into the mystery of how chess has endured for 1400 years and why it delights us still. Shenk guides us on a tour through everything from the intrigue of warring nations to the play-by-play thrill of a historic game, and muses as much about how chess has shaped humanity as how humanity has shaped chess. A must-read for hobbyists and serious players alike.

(The Wednesday Book Club is an ongoing initiative of mine to write a book review every week. I invite you to peruse the index. For more on The Immortal Game, keep reading below.)

Insert coin to continue… »

From Andrew Curry and Wired comes this comprehensive article on Settlers of Catan, a superb piece of board game journalism if I’ve ever seen one, and a must-read for players of all levels. It’s got a bit of everything: a look at why Settlers fit the market like a glove, a little about designer Klaus Teuber, an overview of the “German style” of board game design of which Settlers is the most prominent ambassador, and a peek into the complexity underlying the game’s infamously balanced mechanics.

This caught my attention:

In 2006, Brian Reynolds, a founder of Maryland software company Big Huge Games and the programmer who developed the AI behind the addictive computer classic Sid Meier’s Civilization II, set out to make an Xbox 360 version of Settlers. To help programmers develop the game’s AI, Teuber spent months exploring the mathematics of his most famous creation, charting the probability of every event in the game. The odds of a six or eight being rolled are almost 1 in 3 for example, while the chance of a four being rolled is 1 in 12. There is a 2-in-25 chance of drawing a Year of Plenty development card. Teuber created elaborate logic chains and probability matrices in a complex Excel spreadsheet so the videogame developers could see how every possible move and roll of the dice—from the impact of the Robber to the odds of getting wheat in a given scenario—compared. The end result was a sort of blueprint for the game that gave Big Huge Games a head start and showed just how complex the underlying math was. “It was the biggest, gnarliest spreadsheet I had ever seen,” Reynolds says.

I want to see this.

One of the best things that happened to the Civilization series was how in Civilization IV, lead designer Soren Johnson laid the mathematics and AI bare for everyone to see, expanding on a series tradition in the Sid Meier games to make all the data easily accessible (and therefore modifiable).

Settlers is elegant enough that I’m sure people have already figured out the math through a spot of reverse engineering; it’s really not that hard. But I’d love to see Teuber’s spreadsheet for its immense historical value as a design document alone. Surely there was a calculated rationale to everything from the fifteen-road limit to the assignment of three ore/brick hexes instead of four—and I often wonder if the perpetual endgame glut of sheep is here as an intentional crimp.

With the anniversary of the St. Valentine’s Day Massacre on the horizon, it seems highly appropriate to invoke Blaine’s Theorem and in doing so, say a few words about love, chess, and the greatest motion picture of all time and all time yet to come.

Casablanca is one of those films that nobody really falls in love with the first time through, even if they think otherwise. Most of its enduring power emanates from multiple viewings, when the film truly demonstrates its uncanny ability to resonate with almost every conceivable romantic trauma, especially those of a triangular geometry (which is to say, practically all of them). You go through life-as-such and every time, there’s always a handful of scenes that you’ll never look at in the same way again.

I haven’t watched the film in months—I only pull it out once a year as a routine, emergencies notwithstanding—but I already expect to encounter these transformative moments with respect to two scenes in particular: a) when Victor Laszlo leads Rick’s Café in a stirring rendition of “La Marseillaise” (for reasons I’m not even going to bother explaining), and b) the first time we see Rick, brooding over a chessboard by himself.

So here’s the mystery du jour: why is Rick playing chess?

Insert coin to continue… »

If Settlers of Catan is any indication, Germany sure has a grand tradition of designing tabletop amusements. Last week, German board games made headlines when a British auction house put several Nazi-era propaganda games on the block. On Friday, the BBC followed it up with a story about some of the British wartime equivalents, which included stimulating titles like “Decorate Goering” and “Hang Your Washing on the Siegfried Line.”

Regrettably, the mainstream press saw fit to stop at the level of goals, objectives and the roleplay element, and did not appear to find the game mechanics themselves to be newsworthy. This is understandable, given that the representational level of “Bomb London for 100 points” is probably of more interest to historians than the question of how exactly one goes about conducting such a blitz.

Insert coin to continue… »

Here’s something I would have posted last Thursday if I hadn’t cut myself off from the Internet in what was, in hindsight, an excellently timed and perfectly necessary pre-Potter lockdown. It’s been all over the news at a national and international level, as it damn well should be, but I feel it is my duty as an enthusiast of games of strategy and an alumnus of the University of Alberta’s esteemed Computing Science department to once again highlight the tremendous accomplishment that Dr. Jonathan Schaeffer and the GAMES group made last week. I heard rumblings of a major breakthrough about two months ago, but the details were kept under embargo. With the publication of the accomplishment in Science, it’s official: checkers has been solved.

For those of you who aren’t familiar with computing science, game theory or their related fields, what it means in layman’s terms is this: consider how with a simple game like Tic-Tac-Toe, pretty much everyone over the age of five has stumbled upon a strategy that will always play to a win or a draw. Well, it’s been a long time coming, but they’ve just done that with checkers.

There’s a considerable wealth of information on the Chinook website, where you can step your way through a demonstration of the proof or find your way to the article in Science.

More than anything else, I hope this kind of high-profile accomplishment encourages others to pursue studies in what is, I think, a grossly misunderstood and often ill-introduced branch of the sciences. I know that I, for one, had little idea just what I was missing until I transferred into their programme in my third year, a decision about which I have almost no regrets. Computers aren’t just tools that are meant to sit around generating heat in office cubicles, waiting to be thrown out a nearby window; their study is not limited to job training for information-age janitors, network witch-doctors and software monkeys. There’s a genuinely interesting field of scientific enquiry there to which few receive anything remotely resembling a proper introduction. I sincerely hope this is a step towards the elusive remedy.

Next stop, poker!

As some of you know, a week and a half ago I played in the Settlers of Catan Canadian National Championship (Western): 60 players, 4 round-robin games, and a break to a 16-player semi-final based on wins, victory points and victory point percentage. In short, I bombed; you can see the results for yourself, and observe that I did not manage to finish in the top third.

On a game-by-game basis, it was far from a disaster by any standard, as I won one game decisively (6 VPs on the table, 4 for Longest Road and Largest Army, and another one just to make sure) and tied for a close second in two others in the face of unfavourable dice. But I needed two wins to advance to the semis. As with most games that involve a tenuous balance of skill and chance, consistency in the face of misfortune is the mark of a skilled player. It looks like I have some ways to go.

We will speak little of Round 3, in which I did not show up, nor did the number 5, and I finished dead last with only 4 VPs. That debacle was almost entirely due to tactical and strategic miscalculation on my part, as I drew first position on the snake (placing my initial settlements first and last), grabbed a 5/9/10 and went for an expansionist wood/brick strategy, which I almost never do. Anyone who has heard me mope about Scrabble knows that I have a remarkable tendency to get overly experimental in untimely competitive situations. All three of the other players went for an ore-heavy game, and rolled the numbers to match, erecting early cities and grabbing piles of development cards while I sat on my hands with nothing of significant trade value. With no cities, no soldiers to keep the robber off my property and no 5s, my exit from contention was terrible, swift and entirely my fault.

But it happens. And while the knowledge that you played poorly is hardly a happy thought, it gives you the opportunity to shoulder the responsibility and think about how you might act differently in the future. In spite of the dice, you retain the impression that you have control over your own destiny.

However, three key elements of Settlers of Catan elevate its strategic complexity to a whole other dimension that I am only beginning to grasp. It’s non-zero-sum, it’s for four players, and it’s built on trade.

Settlers is non-zero-sum in that you can check a player, or you can block him from victory to some extent (ask Jake Troughton about the game he had elsewhere in the tournament where the other players held him at 9 VPs for well over an hour), but aside from snatching away the Longest Road or Largest Army, you can’t roll him back. Once you have points on the table, they stay on the table. And when someone coasting to victory settles on the same numbers as you, the rolls that permit you to develop confer no relative advantage. The dice may well push them over the edge, and there isn’t a damnable thing you can do about it.

And more to the point, in a four-player game, someone can hand someone else the game with a series of bad decisions, and there isn’t a thing you can do about it except sit there and watch. If the initial settlement placement pans out so one opponent is left with a disproportionate tract of uncontested land open for expansion, because the other two decided to crowd you instead, it sucks to be you and that’s that. Similarly, if another player decides to go through with an uneven trade, or break a three-way embargo, there’s little to be done apart from taking a bullet and proposing an even more ridiculous counter-offer. I’ve been on both sides of this situation. On the winning side, you try to get yourself into a position where someone else has to cooperate with you just to stay in the game.

I’m not an Xbox 360 owner, so I haven’t had a chance to poke and prod the AI behind the Settlers of Catan video game, but I’d be very interested in examining it for its shortcomings. Like the human element behind games like poker or Diplomacy, it seems to me that the trading element that fundamentally drives the course of proceedings in Settlers would be extremely difficult to formalize and refine. One of the biggest challenges when it comes to artificial intelligence is getting over the assumption that you’re playing against someone rational. Game theory was designed to fight rational fire with rational fire, which is why you end up with suboptimal equilibria like the Prisoner’s Dilemma. In Settlers, you have to account for deeply suboptimal opponents – the ones who, in the Prisoner’s Dilemma, could actually be hoodwinked into cooperating instead of defecting; the Kevin Lowes who will actually give away their Chris Prongers for less than nothing (to your divisional rivals, no less). What you do at a table of championship-level thinkers is and must be different from how you behave when someone is just there for the burgers and door prizes.

And I wouldn’t have complained about a little help from the dice.

Belated congratulations to my compadres Chris Samuel, who finished 13th with two wins and contended in the semis, and Steve Smith, who finished 19th with one win (Round 4 against me, knocking me out), and sat (but did not play) as an alternate.

Hockey anthem addendum: I have a lot of respect for Holly Cole for being one of Canada’s leading jazz ambassadors, but here’s a three-letter memo about her performance of “O Canada” at the Duck Pond in Game 2: WTF?