GDC 2012, Day 2
Thursday, March 8, 2012 at 9:17PM
Sirlin in Game Developer's Conference

Designing Games for Game Designers

Stone Librande is creative director at EA/Maxis, and on the weekends he teaches students game design. His talk was about how he uses board games, card games, and other non-digital elements to teach game design. He believes that no computers are needed to teach design, and that they actually get in the way. The fundamental concepts can be gotten to more quickly without the distraction and complication of software.

He was inspired by an article from Greg Costikyan (someone in comments will link that article?) but he said it was too hard to get students to read some long thing. Actually, he oftened mention things he's done that didn't engage students enough and how he changed things up to get them interested, and it often involves making a game out of things or tweaking the rules of things, so that's kind of meta-interesting, if you know what I mean. Anyway, he summarized the essay he liked with a very simple diagram. I don't have time to draw, but I can use some boring words for you.

A game is a START CONDITION with an arrow pointing to a GOAL. Inbetween those two things is one or more OBSTACLES. A dotted line around the entire thing represents the RULE SYSTEM that the player interacts with in order to change things in the system, and it lets them make moves from the start condition toward that goal.

I think he taught these concepts very well. By just telling you that, or showing you a digram, you might not fully and deeply understand it. A better way to understand it ("it" being the system described) is to...play a game about it. Playing a game is like the ideal way to understand a system. So has devised many simple games, like dozens and dozens that he uses for this stuff. I don't remember specifics here, but stuff about rolling dice and getting different colored poker chips that let you make different moves or something, and a goal about having X number of points. And the catch is, hey lets change things around now! But we'll only change the start conditions. Then you see how much effect start conditions really have. Ok now let's keep start conditions fixed, and change around the goal. Notice how it makes some of the rules extraneous, like they don't do anything interesting any more, but other rules still work fine. In each case here, we made a very differnet game.

One example of changing start conditions was Backgammon. There is historical evidence that a very long time ago (thousands of years I think), all the black player's pieces started on the black player's first space, while all the white player's pieces started on the white player's first space. So the start conditions were different. This is kind of boring though because it means there are several turns that are just filler, they don't have enough of an impact on the game to be interesting. (Note: this is exactly why I made Flash Duel's board 18 spaces rather than En Garde's 23 spaces.) Anyway, later on in history the start conditions of Backgammon changed and the pieces got to interact even on turn 1, so that's way more fun.

Another excercise he did is present students with a game where each player has a starship that has 4 slots on it where you can put dice. Each player gets two 4-sided dice, one 6 sided die, and one 10-sided die. Two slots are for weapons, one slot for shields, and one for engines. When two ships fight, first you each roll the engine dice. The player with the higher result gets to attack with both weapons, while the player with the lower result only gets to attack with one weapon. The shield roll subracts damage that weapons deal.

He had players configure their starships however they wanted, then they all played each other. This is actually an *excellent* lesson that is very relevant to what we do every single day in development of Sirlin Games, because the whole point of it is balance feedback the players give. Some players very strongly claim that some configuration is the best and "overpowered" or whatever. Why? Because they played like 2 games or something and this anecdotal evidence convinved them. So it's a lesson that if you're that sure about game balance from that little evidence, you just don't know what you're doing. When such a claim comes up, Stone then shows them how to use Excel or something to just compute the expected value of damage of one configuration compared to another. And of course this often shows that the "overpowered" ship is actually weaker than the other guy's. Stone said he didn't know if there is a single best ship, or if there is some rock, paper, scissors set of ships that counter each other, and left it as an exercise to the reader.

Just a quick note on that, it's hard to even type that last line without mentioning what I think is the key concept to both Poker and Yomi: Donkeyspace. Even if you knew how to play Poker or the Yomi card game optimally, playing optimally would not be the most successful strategy in a tournament or ladder. Playing optimally means you are playing in the least exploitable way, which is a pretty sweet. But imagine you played against someone who was playing really terribly, like RPS where they blatantly only ever play rock. You should not play "optimally," you should play to exploit their badness. So even if there was a single best ship in Stone's example, it's entirely possible that playing that ship would not win a tournament, and that someone exploiting other people's bad ships even more would win.

Oh I forgot to mention that from the very beginning, for all these games he says he encourages the students to really try to win. He taunts them and Stone himself plays and doesn't hold back, letting everyone know he will gloat if he wins. This is specifically because he's telling the players to feel what it's like to play these games, and not look at them in a detached way like designers. Interesting, as I'm certainly familiar with that problem. I can often only see a game as a designer these days.

Another of Stone's games was interesting, and I'm not actually sure what to make of it. He gave students a board (a grid, like a Chess board) and put 4 tanks on one edge, and gave simple rules for how the tanks behave (automatically, no player control of tanks). On the other end of the board, he put a robot. The robot is trying to survive long enough to get to the end of the board, to the last row where the tanks are. The student's goal is to write rules for the robot such that it's a very close call whether the robot wins or the tanks destroy it. So it's a tuning/balancing problem. Students are in differnet groups, each group coming up with their own solutions. The trick is that once each team has their robot rules tuned pretty well, he then changes it up and takes out the tanks entirely. Now the *robots* fight each other.

The kind of mind-bending thing here is that he says usually it turns out that the robots all have a very close match against each other, even though they all have different rules (it's asymmetric) and they weren't even designed to fight each other. They were all designed to fight the *same* set of tanks though, so somehow it usually ends up fair vs each other. I'm not sure if that's some genius thing that maybe I could use somehow, or if it doesn't even make any sense at all. At least he got me thinking!

That said, I thought the most interesting thing was Stone's example where he said most teachers would give some lecture about B.F. Skinner, but that's boring even to him, so instead he made a game out of it. (Oh boy...here it comes.) The "game" is that each player gets two dice and there are no turns or anything, anyone just rolls whenever. If you get double sixes, you get a poker chip. If you have a seven showing when someone else gets double sixes, you lose a poker chip. That's it. Notice that there is no strategy at all, and no decisions to be made. This is not even really a "game" at all. The secret is that Stone is trying to see how long he can get students to stay interested in this "game" which is entirely based on skinner box mechanics, with the students as the rats. He can usually do 30 minutes (wtf) and his record is 1 hour (haha).

First the students do it a bit, then realize they should roll the dice over and over really fast. Then he I think he puts a time constraint, like it lasts only 1 minute or 2 minutes or something. Or maybe not, I wasn't clear if that was part of it, though that constraint might make it even more intense. Anyway once the students show any signs of being bored (and he says they are pretty much always really exicted by this game), he opens up a package of Pepperidge Farm cookies, the expensive kind, and lets the smell waft through the room. He says now you get cookies instead of poker chips. Students find this incredibly awesome and they love this game! The thing is, you only really need like 2 cookies, getting more cookies after that isn't so great. How to keep it interesting?

Real money. He throws down a bunch of quarters and says now you get real money for double sixes. College kids need quarters for laundry and vending machines, and they are broke anyway so now this is way exciting to them. After a bit of that, he says let's up it even more, and fans out twenty dollar bills. Do I even have to tell you how exciting the game is to these kids? lol

The crowning jewel here is that Stone said he got kind of tired spending twenty dollars every time he did this session, so he figured he could cut his losses by playing too. Maybe he'd win some of the dollars back. "And I have to tell you," Stone said, "...this game is pretty damn awesome!" Haha. I hope you all realize the power and depth of that statement. Even a highly accomplished and knowledgable teacher of game design who is playing a thing that isn't even a game, a thing with literally no decisions, gets swept up into the skinner box training. That's how powerful it is.

Mind Games: Brain Training for Game Developers

Scott Crabtree said that there are lots and lots of sessions at the conference about how to make better games, but his is the only one on how to have a better brain. If you have a better brain, you can make not only your current game better, but all future games you work on better too. And of course, your life outside of the games you work on would benefit. Maybe you think that's hokey, but I think he's exactly right. The only problem with this session for me, is that I have read way too much on this subject, so the information wasn't really new to me. It might be new to you though, so let's go over some of it.

The old view (in brain science) is that after childhood, you can't really change your brain. It's all kind of set and that's that. There was a period of scientific debate, and now there is basically no debate left: the old idea is wrong. While it's true that your brain changes much more easily until you hit your early 20s, it definitely can change at any point in your life. You can rewire it, and meaning change your synaptic pathways, enabling new ones, disabling old ones, and even generating new brain cells. It just takes effort, but you can do it and you should do it.

When we age, we all get "plaques" in our brains, which are pockets of deadspace where neurons no longer fire. If you have a whole lot of connections in your brain, the obstacles that these deadspaces present aren't so bad because there are many other routes your neurons can take to get somewhere. With fewer connections, this can be a big problem though, like a roadblock on your only road to somewhere. Excercising your brain means your mental decline will be much, much less than it otherwise would be.

Your brain functions are "use it or lose it." Doing a certain task often forms new pathways, meaning it actually rewires your brain in a measurable way. If you don't do something for a long time, those pathways can fade away, again physically changing the structure of your brain. Scott came up with the REACH acronym to help us remember how to improve our brains.

R is for repetition. Doing something over and over is necessary to make a brain-change. If you focus on a new skill, you should try it often for at least a month.

E is for effort. You have to do something that requries significant effort, meaning something that induces "flow." It must be challenging, but not so hard that you can't do it (and no so easy that you aren't being pushed.)

A is for attention. It must be something that uses all your attention. That takes up all the bandwidth of your pre-frontal cortex. Don't multitask, you need all your attention on this one thing.

C is for complexity. The more complex and challenging the activity, the more rewiring you'll get.

H is for health. Your brain is part of your body, and it is greatly affected by you getting enough sleep, nutrition, excercise, the right level of stress, and so on.

A couple more notes on that. Scott very strongly recommends against multi-tasking. He says study after study shows how terrible it is. Your attention actuallly can't be on more than one thing, so what you do is rapidly switch where your attention is pointing. This is known (from science and all) to make you stupider, meaning you have much higher error rates and you have a lower effective IQ. It also has long-term bad effects making you less able to filter out irrelevant information, and more prone to distraction. Multi-tasking creates an *illusion* of offering challenge, because it feels hard to do a bunch of things at once, but actually it's just not possible, he says. Instead, we just switch attention and get all those bad effects, we don't ever have attention on several things. (I wonder what Scott would say about Starcraft.)

He says the opposite of all that bad stuff is meditation. He doesn't mean meditation with any kind of religious, spiritual, martial art, or other baggage. He simply means a state of relaxation where you practice focusing your attention without letting it stray. He recommends focusing the feeling of your breathing and he notes that being good at it requires a slightly different skill than you would expect. As an analogy, if you have to balance on one foot, what being better at that skill really means is actually that when you start to fall just slightly, you notice sooner than someone who is worse at it, so you can correct it sooner. Likewise, a master of mediation is more self aware about what his (or her) attention is on, and is so aware of it straying from being focused on breathing, that he or she can correct it and get back on track very easily.

If you're in a state of flow, that means you are probably rewiring your brain. The state of flow is fragile though, and distractions knock you out of it. It can take about 20 minutes to get into a full state of flow (hmm, I would think much less than that, actually, but I think he said that). He definitely said that you lose your flow after 1 or 2 minutes of distraction. So it's actually *crushing* when someone bothers you during that time if the conversation lasts more than 2 minutes, or if you get distracted by some new e-mail or IM or something. Try to ignore all that if you are in a precious flow state.

Being stressed out is seriously bad for your brain. Having some stress is good, we're talking about about more extreme stress that's bad. It activates the fight or flight response, which Scott explains is actually the "freeze, fight, or run" response, but that isn't as catchy (it's correct though). Even though we got this from situations like dangerous tigers being near, you get it now in a stressful business meeting, where it's just sucky and counter-productive. You freeze up, and your ability to think clearly goes out the window. Fight / run doesn't help you either.

Scott explained that even though your brain is only a couple pounds, it uses up 20 or 25% of your body's energy or nutrients or something. This shows why your overall health is so important to your brain functioning well. Also, it relates to how badly high stress can fuck up your thinking. When you get that fight or flight response, your limbic system sucks up glucose and a few other chemicals he mentioned that I forget, and powers you up for running or fighting. Remember, you brain takes a ton of power to work, and you just lost a big chunk of it to the limbic system, so you're not in a good shape to think critically or make logical decisions in that state.

If you are in stressed state over something where your evolution is working against you (so...not a tiger chasing you), then you should try to relax in order to restore your brain function. Deep breathing rather than shallow breathing is one legitimate technique to do this. Another one is visualizing a safe and relaxing place to you. Scott explains that these techniques can only do so much. If someone's screaming at you, the breathing thing actually might help, but the visualization thing is probably not great to do right that moment. It's just another tool to try to return to rest state. Oh, and another one that actually works (like, by science experiments and stuff) is touching your lips. Scott says he likes to give us just science data without his own opinion, but on this point, he will add his opinion on *why* touching your lips works (and it does work, he says). He thinks that usually when your lips are touched, it means you're drinking something, eating something, or making out with someone. Either way, it's good stuff, so that's how you're conditioned. He also notes that basketball player Kobe Bryant famously licks his lips before foul shots, possibly as (unaware) self-medication for the stress. Also, Scott asked how many basketball fans were in the room, and I think only two people out of a huge crowd raised their hands, which seemed to make him sad, lol.

Scott also explained the research on improving strengths vs improving weaknesses. You might think that you should work on your weaknesses, but the data shows that you get much more milage out of improving your strengths. If you really, really suck at something, it takes enormous effort to suck slightly less. That same effort can dramatically improve your abilities and cause much more brain rewiring if it's something you are at least kind of good at. For example, in one study, people were taught how to do speed reading and were given a test before and after. The people who could only read very slowly increased their speed 50%. The people who could read really fast increased their speed 600%.

Lunch

I had lunch with long-time friend and game director of Uncharted 3, Justin Richmond. Justin told me about a session he went to about Valve switching Team Fortress 2 from premium to freemium. He said at first, you paid for the game, and Valve kept secret how many players there actually were. People thought it was a lot but it was only 20,000 concurrent, which was less than outsiders realized. Oh by the way, this is all like second-hand info and I'm probably butchering it.

Anyway, Valve switched to free and got some astronomical amount of players. They also learned that best way to get more players (well, besides making it free) was to do patches. Every single patch, a bunch more players show up. Even if the patch has no content at all, and is just bug fixes, this still works. Ok, so Valve started selling in-game hats, just cosmetic things that didn't affect gameplay. They didn't want to sell gameplay affecting stuff because that would ruin things. People bought the hats. I pointed out how this ruined the careful aesthetic of the game, but Justin said these hats were designed by Valve and they at least had some artistic integrity in mind. Except...then Valve thought it would let anyone make hats! Hmm, now they will mostly be crap though, what to do about that?

Next, Valve decided to make a contest about it. (By the way, making a contest out of something seems like it always work in any situation. Note to self, remember that.) The top X community-created hats each month as voted by the community would get into the game. Result: lots more hats sold, more money for valve. Then people said what about weapons? We want weapons! So valve said ok fine, we'll make new weapons and sell those, but they will all have drawbacks so they aren't more powerful. I said but wait, surely some are actually more powerful anyway. Justin said yeah but everyone buys the weapons and loves them. I said ok whatever, at least you can directly buy them.

How about selling even more hats though? How could we do that? Valve then put out more detailed specs on how to create hats, and made demo models of hats to help people even more with the tech side. And they decided to give 20% of revenue for user-created hat to the user who created it. This attracted a much, much bigger set of hat makers, including real artists, and the hat market exploded. Justin says there are dedicated hat-making companies that make hundreds of thousands of dollars of just that, and nothing else. I said really? Justin said yes really.

Then he started talking about community created weapons, and an unexpected thing happened. A guy sitting 3 feet away corrected him, and said he worked for Valve. Yes, really. (And yes he was there the whole time.) He said for community created weapons, it's actually just the model and the "idea" of how it works that the users create, and that Valve has to do a lot of work to really implement the gameplay correctly for them. Then I said to the Valve guy that it would cool to that for a supposedly competitive game, if it had the FEWEST weapons they could possibly make, as in the most elegant, most tightly tuned and balanced game. Instead they have some sprawling game of a 120+ weapons, and it's kind of anti-competition mindset there.

Valve guy responded that the hardcores love it though. I said that that seems to be an answer of the form "more money is the right thing to do" and mentioned that players may also want more cocaine or whatever other crazy thing, and there are other considerations, like caring about what competition actually means for example. So how about that, how about caring about an ideal here and committing to that with the best possible game? He said that's what Counter-strike does, and it still sells a lot, so for an e-sports type thing like I'm saying, they have Counter-strike. For a "just for fun" game, it's a different audience, so it's ok to do things that are bad for an e-sport. This was, at least, a much better response than I expected. So I said yeah ok, good explanation. That said, it's a bit weird to think about the concept of a player in a competitive multiplayer game who is "hardcore" (by Valve guy's own label) yet you wouldn't want the e-sport ideals. Apparently this is the entire set of TF2 players, heh.

The Art of Diablo 3

Christian Lichtner talked about Blizzard's art process and ideals in Diablo 3. He stressed that the art has to support the game design, that it has to support the theme of the game, and that it has to pay homage to earlier Diablo games.

Actually a lot of his talk was really showing how important art serving gameplay is. He showed many screenshots of earlier versions of the game compared to current versions, and it was pretty striking. Often, the earlier versions looked fantastic by some measures. Like one showed a bridge that had great detail and just looked awesome. The quality of the art in the earlier screenshots looked on par with what other companies might ship and think was fine. But in every case, the more current version illustrated just how deep the problems were with the earlier ones. Each time, the newer screenshot made me realize just how many problems the earlier ones had, and how hard an art director's job is to *know* that. I mean you'd have to look at those old versions (which look good in many ways), realize the problems, visualize the solutions, and get everyone else working towards that.

The issues I'm talking about where almost all about readability. About the player knowing clearly what is what, and what is happening. Christian explained that they use three layers for art, the background, mid ground, and foreground. The newer screenshots showed how the backgrounds now all have the contrast turned down, and a painterly look to them. The player's character and the monsters are in the mid ground, and these models are intentionally more detailed than the background, have higher contrast, and have more lighting applied to them. This makes them really pop, and when you look at old screenshots in comparison, the characters were often camouflage way too much. High contrast backgrounds might look good when you are are admiring a background, but they really get in the way of telling wtf is happening.

The fore ground layer is used for the special effects from all your abilities. These are intentionally the most saturated and brightest, and the whole point is to make them really, really pop even way more than the characters. When you do moves, this is what makes them feel super awesome.

Another thing he mentioned is that Blizzard likes to do things that are archetypes or expected (such as having a "barbarian" class, and having the globes for health / mana on the HUD like before) but switching things up 20%, to throw you off a little and make it unique (the barbarian is actually an old guy). Also, Christian says "barb" when he means to say "barbarian" and I suspect this was weird and grating to almost the entire audience, considering he said it about 100 times.

He also showed extensive concept art progressions of how they get the heroes and monsters to look the way they want. While a lot of this made sense and showed them working towards a certain thing, the weirdest thing was about the monster Diablo himself. Christian showed a slide with Diablo's name (the monster, not the game logo) and an image of Diablo. This image was I think the most fantastically awesome of any illustration shown the entire time. Then he told us that was the Diablo 2 interpretation, and that they wanted some other thing for Diablo 3. Why? I have no idea. Then he showed a bunch of different takes on Diablo that all seemed worse, ending with some other version of him that was at least better than the previous concepts, but kind of no where near the first image shown. And that's the final Diablo 3 interpretation of him. Hmm ok.

GDC Microtalks 2012: One Hour, Ten Voices, Countless Ideas

I was a speaker in this session. This was 10 presentations by 10 speakers, each one under 6 minutes. Each speaker uses exactly 20 slides that auto-advance after 16 seconds per slide. This is a very difficult format to prepare for because it's so intense, so fast, and every single damn word and image counts. Also, it's logistically difficult to get your words to match up with the slides unless you really polish your talk, which it seemed we all mostly did.

I went first (well, after the session's organizer gave his micro-talk that introduced the concept of micro-talks). Mine was on the time pressure in games, and how I think there is a mistaken notion that games where you think for a long time are "strategy" games, while fast games aren't. I think fast games can be strategy games too.

I talked about the difference between conscious thought and unconscious thought, and how conscious thought is much slower, taking about 0.3 seconds to coalesce, while unconscious thought can operate much faster. For very simple things like arithmetic, conscious thought produces higher quality decisions, but for very complex things, unconscious thought produces higher quality decisions. I gave some examples from real science and stuff.

Games that have us rely on unconscious thought are interesting in that they are accessing some part of us that we don't usually have access to. So you get to learn about yourself by playing games like that, and learn about other people too. "I can learn more about someone by watching them play 10 seconds of Street Fighter than 10 hours of an RPG," said Seth Killian of Capcom. In an RPG, everyone goes through the same motions pretty much, while a game tapping into rapid fire unconscious thought decisions like Street Fighter is pretty revealing.

I also quoted Yomi expert Waterd103, the master of timers: "It annoys me that Yomi timer is so long that I have time to write down notes, analyze patterns, do some math, review history. Is insane, because since I CAN, then I feel I HAVE to. And it's not that much fun."

Well put. If you have a bunch of time, you do a bunch of boring busywork using conscious thought which ironically could very well have you make WORSE decisions than if you didn't do it at all. Making the correct decision involves an enormous number of variables and actually your unconscious could very well do better at it than if you instead filled your consciousness up with some data that felt like the whole story but totally wasn't. So long timers basically suck here on decision quality and fun factor.

Another interesting thing about unconscious thought is picking up patters in what other people are doing (this happens to be the whole point of Yomi, but applies to many other games too). I quoted science studies about how people don't have conscious awareness when they are picking up on and exploiting patterns. So you benefit from picking up on these patterns and change your decisions for the better to take advantage of them, but on a conscious level, you don't even know there *is* a pattern. In a hilarious one, after subjects already picked up on some pattern in a very clear and measurable way, they were asked if a pattern existed and they all said no. When they were told a pattern existed, and that they should use conscious thought to predict stuff, they all did worse than if they didn't know any of that and just used unconscious thought. When given unlimited time and cash rewards, no one could (consciously) find the pattern that they ALL took advantage of unconsciously. Most hilariously, when professors from the experimentor's own department who were aware of the research took part, even they could not find the pattern, thought they too used it to their advantage unconsciously.

The point is this stuff is really interesting and takes place outside of your conscious awareness. Time pressure is a good way to force decisions into this interesting realm.

CliffyB of Epic Games gave a rousing speech about how he'd get into games right now, if he were starting out. Mysteriously, Brandon Sheffield covered very similar ground in his speech (which actually wasn't a bad thing at all). I found his advice stuck with me more than anything else in the microtalks. He talked about how the way you make it big is to do something small that is so intensely personal to you. It has to be something legit, something that your heart is in, no matter what crazy thing that is. The quirkiness and individuality will work in your favor if you truly love this thing. He gave examples of various iOS games and other indie games that were big hits, and mentioned that Notch didn't do market research on whether people would like a sandbox game, he just build minecraft to be the most awesome thing for him personally.

Brandon said several times that this thing should be SMALL. CliffyB also cited project scope as the #1 most important thing (also saying SMALL). CliffyB said the rookie mistake is 10 features to 15% instead of 3 features to 85%, because the second one is at least shippable. He also said your first game or few games will suck for sure, so get them done sooner by making them small. Brandon emphasized keeping it small more to keep it tight and focused and clear of vision, in addition to the logistical reasons like actually being able to finish it. I kind of wonder if I should create a design for an iOS game that is along the lines Brandon recommends, a very small project that expresses whatever thing I am most able to express. Too bad I don't really have an iOS programmer to explore such things with.

That's all for today.

Article originally appeared on Sirlin.Net (http://oldsite.sirlin.net/).
See website for complete article licensing information.