Archive for September, 2007

Avoiding Chernobyl in Strategy Games Design

Friday, September 21st, 2007

Chernobyl stands as one of the best demonstrations of the destructive power of positive feedback. Huh? I don’t mean positive feedback as in the mouse made it through the maze give him a food pellet. In systems design positive feedback happens when some part of an output is fed back and added to the black box that operates on an input. Thus as the input increases the output tends to grow even more. The growth can be exponential…..explosive. Negative feedback means that the sample taken off of the output is subtracted from the process in the black box. Thus it tends to stabilize a system. If input increases a huge amount and the output starts rising as a result then the negative feed back restrains the growth in the output. Pressurized water reactors that are the workhorses of the nuclear power industry operate on a negative feed back principle. A key design principle is to have a negative temperature coefficient of reactivity. Basically as the temperature of the water increases it becomes a poorer moderater and thermal fission slows down. At Chernobyl due to bad design, bad operating practices, bad decisions and bad luck the graphite moderated reactor temporarily achieved a positive temperature coefficient of reactivity. Increased heat meant increased power which in turn generated more heat. In a fraction of a second the power output increased exponentially and a steam explosion occured. As the Ghostbusters would say that was a bad thing.

Chernobyl From Helo

So what does this have to do with strategy game design? A game is really just a bunch of black boxes with connections running between them. The players operate the inputs to the system by pressing buttons. Buttons can move pieces, draw resources, play cards and do all other manner of strange and fantasmagorical things. The output of the entire game system could be thought of as a group of some type of victory numbers. Each player has a victory number that represents how strong their position is to win the game. So the proposition is that if you wanted to you could take any strategy game and map it out as a system. Draw out the player inputs, the operations performed in the boxes, and let the outputs be each players claim on “winning” the game.

So when thinking about a game in this manner, perhaps while you are doing design, positive and negative feedback should be something you keep in mind. The Chernobyl event is an extreme version of positive feedback but that doesn’t mean you should exclude positive feedback from your design. Rather you should just take pains to manage it and ensure that it can’t go Chernobyl on you.

I’m going to discuss positive feedback on two different scales in my next entry. One is a meta scale and concerns the runaway leader syndrome and the other is a finer “systems” type scale involving the set collection mechanic.

Just a note as well. I am heading off to Finland next week to accompany my famous Art Historian wife as she delivers a lecture on performance art for the Finnish Arts Council. Besides being fantastically beautiful, Finland is one of the most wired countries in the world so it should be business as usual for me. One of the advantages of being an indie developer is the show goes wherever I do.

Adventures in Counterfactual History

Monday, September 17th, 2007

I think that well designed computer strategy games should offer ample opportunities for counterfactual ruminations. A counterfactual is a type of thought of experiment where you ask what would have happened if X was the outcome of an event rather than the historic Y. As humans locked inside this apparently multi-dimensional space time membrane we experience reality as a slide show. Even accouting for relativistic effects our best theories posit an arrow of time that runs in one direction and a “past” that is frozen in place. Despite this there are theories of infinite universes that branch of at every quantum decision. Other exotic theories postulate an infinitie number of pocket universes that have been created by “infinite expansion.” They all exist in the same reality but they can no longer communicate with each other and since they are infinite almost every conceivable thing has happened. Somewhere there is a you that didn’t get married, have kids and get fat. If you are a comsomologist please excuse my ham handed approach to the topic. The point is that according to some of our best minds the universe itself may be a huge counterfactual laboratory. Counterfactuals are also a great deal of fun. What if Hitler had been assassinated in 1944? What if Enigma had not been broken? Both these questions open up a line of counterfactual reasoning but there is a key difference. These two counterfactuals run the full spectrum. One questions a specific and highly compact event while the other touches on a much more general and nebulous consideration. Both can lead to huge variations in the time line.

It’s no surprise that counterfactuals are the most fun in military history situations. The systems which govern military conflict are such that wild turning points are possible with just a jitter of the inputs to the system….the concepts from Chaos theory that I talked about last time. i.e. the Butterfly effect, complex behavior from simple structures, etc.

Two of my favorite counterfactuals are the the Battle of Midway and the Battle of Gettysburg. 20th Century carrier warfare and 19th Century musket warfare whose outcomes seem like they could have turned on a dime.

The big question for Midway is what would have happened if the US torpedo planes had not drawn off the Japanese fighter cover and the US dive bombers had not arrived nearly simultaneously over their targets? A great website for some background on the battle is Battle of Midway The short of it is that within a few MINUTES three Japanese carriers were ablaze.

Hiryu Under Attack By USAF B-17s

The Battle of Gettysburg has a multitude of counterfactuals. Like the Midway example where the question centers on a focused and specific event, some of the best Gettysburg counterfactuals also operate in the tiny crucible of fate. What would have happened if Joshua Chamberlain had not ordered his famous bayonet charge from off of Little Round Top on the second day of the battle? This is one of the most asked but there are many others. What if the Iron Brigade had not arrived in the nick of time to halt Archer’s attack in the opening hours of the battle? What would have happened if the 1st Minnesota had broken in the center of the Union line during the apogee of the Confederate attack on day 2? Like 20th Centrury carrier warfare the landscape of the battlefield could and did change drastically because of tiny events. Enfilade fire down the line of a regiment could make even the bravest soldiers break and run. A routing regiment was often the first event in the a very real dominoe theory. Disordered retreat is infectious and snowballs quickly. An entire army could be and often was driven from the field because an enemy manuevered onto a flank or rear. If you have ever wanted to see this in action you should check out the animated battle CD-ROMs that I made when I worked at TravelBrains. Not to ruin the moment here but I have a special appreciation for the visual spectacle that is a civil war army in retreat.

What does any of this have to do with strategy game design? I have two points. My first piece of advice is that when setting out to design a game try and build in mechanics that provide for a rich counterfactual experience. The players should come away with questions not only about what would have happened if I had won that battle but also what would have happened if I had tried that strategy instead.

Secondly and this depends on the designer, the system should be stable but also allow for drastic turning points. This is likely to be a contentious assertion. Many gamers with Euro centric tastes are likely to be annoyed by a game where a tiny seemingly random event ripples through out the game’s trajectory. I like it a great deal. In Armageddon Empires I was always trying to come up with ways to allow the players to change the strategic calculus radically. Nukes do this. So does the Assassination special ability. One of my favorite counterfactuals is what would have happened if my logistics genius had not been assassinated by the enemy assassin? Those 3 extra supply points would have allowed me to …..

Well Tyrion did not appear in this entry. However, I think he would approve of small events that change one’s fortune immensely.

For Want of a Shoe a Horse Was Lost

Wednesday, September 12th, 2007

I’m still working on the next installment of the AAR “Distant Thunder.” It’s a bit time consuming so my progress has been measured. Having to jot down notes and take screenshots at the same time is a process that borders on the tedious no matter how fun the game is.

Playing the game though got me to thinking about the role of probability in games. A game like chess is a champion of Newton’s deterministic clockwork universe. When a pawn goes to capture another pawn there is no chance that it will be eliminated instead. Many players prefer this type of approach. Euro games are famous for a X – Y = Z resolution mechanic. Once X and Y are determined finding Z is just a matter of applying the algorithm. There is never a jitter in X or Y. If the Field Marshall spills hot coffee in his lap or the sergeant misreads the map it matters not. If a level 7 Knight goes up against a level 4 Spearman then the guy with the pointy stick must be defeated and the damage is always 3. The feeling is that introducing an element of chance “ruins” the feat of skill that manuevered the piece to its target.

Most war games however because they are trying to simulate “war” incorporate some type of random element into the conflict resolution. Clausewitz called it “friction.” No plan survives contact with the enemy is an axiom of military strategy. The side with the better “Go to Hell” plan has the advantage.

On the other end of the spectrum, a game like Risk of course is aptly all about the dice. You can try and manage your “risk” by using strategy but in the end the gods of probability will have their way. You can roll snake eyes 100 times in a row. It’s just not very likely. Although quantum mechanics tell us that even the smallest probability is enough to make for some wacky outcomes. Quantum tunneling is a good example. So is winning the Mega millions lottery or getting eaten by a Great White shark.

There are good arguments for designing games with either approach or some combination of the two. Armageddon Empires is a strategy game that is heavy on the probability approach. Which brings me really to the point of this whole entry. If you are going to go with the probability approach i.e. dice, random cards, roulette wheels etc. then you need to ensure two key things for good gameplay:

1. Make sure that the players have concrete ways to manage the the risk. One way is to let them prod it here or there. Tactics cards and fate points let them do this in Armageddon Empires.

2. Try and build a stable yet dynamic system. Probability can really help with the dynamic part. What do I mean by this? Well there is a whole field of study centered around “systems.” Negative feedback, positive feedback, bounded, unbounded, discrete, continuous….. I’m not going to pass myself off as a systems theory or systems engineering expert but I know enough to be dangerous…or at least to avoid the dangerous systems. A game designer should build the game system so that catastophic failure isn’t the difference between rolling a 1 or a 6 on a single throw of the dice. Or is this really such a bad thing?

Can small changes in input that yield huge changes in output still provide a fun and challenging experience? The game shouldn’t end because your killer stack built around your monster cyclops got it’s eye poked out. Or should it? I would say there is a huge difference between crafting a sensible risk/reward system and enforcing outcomes versus crafting a system where every game results in the same well trodden path to a forgone conclusion. So the real catastophic failure I’m talking about isn’t the butterfly effect as much as it is the system always pinging to 11 via the same path no matter what risks the player does or does not take. In systems engineering terms if the victory condition is the steady state then you want to make sure that the path to steady state isn’t too short, too long or too predictable.

I’m going to split this post up ala George R. R. Martin. I can’t promise that Tyrion will appear in the next post but I can say that it will be titled: Adventures in Counterfactual History: or How I Learned to Stop Worrying and Love Speculating

The Strategy Game Designer’s Bookshelf

Tuesday, September 4th, 2007

So you want to be a strategy game designer? Do you have the Right Stuff? “Do you have the right books?” is the question I’ll address here. Here are some of my favorites in no particular order. Some of these aren’t pleasure reading so be warned. Some of them I still haven’t finished and only use for getting to sleep when the stress and tension of a long work day have me all wound up. Don’t expect game design ideas to spring fully formed from Zeus’ head after you have read them. For me at least they serve as background material. Grist for my run down mill.

The Strategy of Conflict by Thomas C. Schelling

Strategy by B. H. Lidell Hart

The Art of Judgement by Sir Geoffrey Vickers * Seems to be out of print. I got it for a class in grad school

Diplomacy by Henry Kissinger

Thinking in Time by Neustadt and May

A War Like No Other by V. D. Hanson

Just a few from my bookshelf. I didn’t even bother listing Sun Tzu or Carl von Clausewitz. Most every strategy enthusiast plays a bad game of chess and has read those two. At least that’s the case with me.

Strategy Bookshelf

In Defense of a Steep Learning Curve

Sunday, September 2nd, 2007

The description of the mastery of some rules system as having a steep learning curve is something of a misnomer. Here is an entry on the Learning Curve in the Wikipedia that does a pretty good job at explaining the semantics. Steep learning curve is supposed to convey the idea that during your initial play time with the game you will struggle to master the game system. Each unit of time invested brings little return in dividends (knowledge/mastery of the system). It’s not really the slope at any given point that’s important but really the shape of the curves. Tsunami style learning curve would be the best description for a game that was complex in the initial stages so that learning was slow but at some point the brain comprehends and system knowledge is accumulated almost exponentially. Then the brain discovers hidden depth that has to be teased out of the system. The surprises are slow in coming but over time new discoveries are made. That’s the kind of game I think Armageddon Empires is.

Learning Curve

If you are reading this blog, trying out the demo or have purchased the full game then there is a good chance that you are someone who likes a challenge. You are someone who knows that easy riches without toil do not create satisfaction but only emptiness and discontent. You are someone who would take the Blue Pill. You are someone who would prefer to dine in hell with the Spartans, than sit at Xerxes’ lavish feasting table. Long live the Fighters!

Strategy Board Games

Saturday, September 1st, 2007

I play a lot of board games. Well, I set up a lot of board games and pretend to play them actually. I spend as much time rummaging around boardgamegeek as about anywhere else. If you haven’t noticed Armageddon Empires is heavily influenced by my love of board games.

Michael Barnes noticed this and wrote up a nice analysis of AE as a boardgame on his feature page at GameShark called Cracked LCD. Michael’s comments are more articulate than I could ever manage so make sure you give it a read.