Towards a Moral Code for Game Designers

I've finally played Dan Cook's Bunni Game: How We First Met - a horrendously addictive casual Flash RTS which he has blogged about as a part of his Flash Love Letters series. You can play the game here, but I recommend you don't click through until after finishing this article, because you won't be back for an hour or two. I especially enjoyed the deer quest which happens about 20 minutes in, having had to deal with deer in the garden at home recently (The local council suggested I shoot them or, failing that, throw rocks at them. I'm not sure whether IEDs would have been an appropriate substitute).

Bunni has quite a subtle resource mechanic: flowers provide support for you units, but it is also possible to over farm resources if you have too many workers which balances the urge to rush them and requires you devote attention to getting the mix of worker and resource quantities right. Later in the game you may want to do this, as it's a far easier way to upgrade resource types without having to bomb them.

In fact, the only thing missing from the game was a click through advert for Starcraft: GotY edition. ('Thank you for playing Bunni. If you enjoyed this, you may also want to try: Supreme Commander.')

But, praise for the game aside, Bunni has an excellent example of the moral quandries of game design. One mechanic in the game is very similar that in another casual game, Plants vs. Zombies, and is an example of what I'll refer to as clicklets. A clicklet is something you have to do within the game - clicking a location on the screen - for which there is no situation where it makes sense not to perform this action. In Plants vs. Zombies, the sunshine collection mechanic is a clicklet - some plants generate sunshine, which you must manually click on as it drifts across the screen in order to collect it. In Bunni, the clicklet is tree shaking, which allows you to collect fruit and other rewards. I could not see any situation which warrants not shaking a tree when fruit is available.

There's several key differences between a clicklet and other game mechanics. The only resource a clicklet uses up is player attention: clicking to attack in Diablo is not a clicklet because it distinguishes which monster you spend time attacking. A clicklet is an atomic action: it always makes sense to shoot Space Invaders, but you do this through moving left or right, firing, not by clicking on the Space Invaders individually. And a clicklet is not optional: you can choose to crouch roll or not crouch roll in Zelda: Ocarina of Time while you travel across Hyrule field and it will not impact the game in any way.

I'd argue, in fact, that a clicklet is not a game mechanic at all, it is a process of operant conditioning. But luckily I don't have to argue this, because Raph Koster has already done the hard work for me:

Now, there’s no reward in this game, there’s no winner or loser, and there’s no endgame. Yet even during testing, I had to tear myself away, and when put into Metaplace Central, average session length for the day went up 50%. But… in some sense, it’s a crummy game. Why this effect? Because the Easter Egg hunt is a confluence of a lot of highly manipulative tricks.

In the example Raph gives, he took the moral high ground and removed the Easter Egg game he designed from Metaplace. I would like Dan to consider doing the same: either redesign the tree shaking mechanic so that there are advantages to leaving fruit on the trees, or remove the need to shake trees by clicking them. I'm asking Dan, and not Pop Cap (publishers of Plants vs. Zombies), because I know Dan reads this blog and I have a good chance of engaging him in a robust discussion either here, or on his blog, about this feature.

But if Dan Cook, who I know as an experienced and thoughtful game designer, can make what I consider to be an incorrect design decision from a moral stand point, where does that leave less thoughtful, less experienced or less morally scrupulous game designers when it comes to game design?

I suggest there should be a moral code for game designers: one which provides clear examples of the boundaries of which a game design should be careful straying beyond. This code should say nothing about the content of games: I'm not interested in whether a game offends religious or ethnic groups, could be considered pornography or explores issues which would be illegal in real life such as rape, pedophilia, murder or drug use. It would however, have everything to say about the way the game mechanics work so that we have a common platform to criticise games for breaking the code unwisely.

To construct such a moral code requires a dialogue between designers, about what is appropriate and inappropriate in game design. We can all point at grinding as the whipping boy of bad design, but as I pointed out in Permadeath, there is a delicate balance to be struct between grinding and necessary exploitation. I'd like to throw open the floor to comments, sugggestions and criticism, but I'll include some quick suggestions with which to start the discussion.

'The Golden rule:' The most valuable resource in your game is the player's time and attention.

'The Monopoly rule': Any game mechanic which is designed to be instructive rather than fun should be for a game intended to be played only once, and not by children.

'The Rule of Repetition': Any repetition in the game should be to allow learning of the game rules or automated away: once learning has been successfully demonstrated, further repetition should only be as a part of learning more complex rules.

'The Supermario Rule': Your level design should exist only to support 1) the learning of game rules, 2) suspension of disbelief and 3) if it is cool enough (for content that is only to be used once).

'The N-Armed Bandit Rule': Any random rewards should be delivered via an N-armed bandit where M-arms are pulled, where N > 1, N >= M >= 1 and the underlying relationship between the quality of rewards and arms pulled is non-random. Multiple pulls should be governed by the rule of repetition.

'The Rule of Seven': A player should be at most presented with seven options at any one time.

'The Rapture of Choice': Anything which is presented as a choice, should actually be a choice.

'The Playing to Win Rule': Your summits should be scalable.

'The Axiom of Cool': Coolness and utility should be two separate axis.

'The Anti-Garfield': Everything available should be useful.

What others could you suggest?

Another way of thinking about progress

Via the Greedy Goblin, Tobold has a few things to say about progress in RPGs (specifically World of Warcraft) that you may find interesting:

Tetris gets faster because that way with every level the challenge rises, so sooner or later every player gets to the point where the challenge is equal to his skill. If you can consistently get to level 10 in Tetris, you are a better Tetris player than somebody who can only get to level 5. Your skill goes up while playing, because you learn, but the challenge goes up quicker, so every single game of Tetris ends with you losing, and the level at which you lose is a measure of your skill. Then you start over, and because your skill goes up, maybe in the next game you get one level further. If you would draw a graph plotting the maximum challenge you can beat over time, you'd see a constantly rising curve, but which is getting flatter, until it levels out at your maximum potential.

Now imagine MMORPGs would work like that. You start with a level 1 character fighting level 1 mobs, and you would *NOT* get stronger by gaining levels and equipment. If your character would always remain at exactly the same strength, and only your skill and random factors like critical hits would determine your success, then the maximum level of monster you were able to kill would say something about your playing skill. Some players would be able to kill level 5 monsters with their level 1 characters, others would maybe be able to kill level 10 monsters. But obviously you'd never see 25 level 1 characters raiding Ulduar.

Reaction to Exploitation and Evasion

In a rec.games.roguelike.development thread started in reaction to the permadeath article I wrote, Ray suggests:

I think a grindless roguelike will occasionally murder your
characters for no good reason, due to the analysis I outlined
above. And I think I *am* prepared to fully explore the
implications of that conclusion. What do you think a grindless
roguelike looks like?

You'll probably want to read the rest of the thread - there's a lot of insightful points made by a number of rgrd regulars.

It's like free heroin in your pocket

The only thing saving me from complete addiction to Civilisation Revolution on the iPhone is the fact it isn't completely stable. I've had two crashes in the 3 days I've been playing it (And when I say 3 days, I mean every waking moment).

That, and the AI isn't the best...

Canon

I somehow want to relate this to my recent thinking about narrative but will get around to it later...

Late to the party

Mr Whirly is the new cake...

The Function of Narrative in Games: A Theory

Looking at the balance of narrative vs ludology in game design, I keep thinking about what it is that narrative does. I’m not especially interested in the ‘effects’ of narrative on the player so much as the ‘function’ of narrative. That is, what role does narrative have in the process of playing of a game?

The simplest and most compelling theory is narrative exists to help the player continue playing the game. Narrative does this by encouraging the player to have an emotional investment in the story, by tying together a series of potentially unconnected events and places (the ice level, the fire level), and giving perceived value to repetitive actions. This overarching function of narrative at any point in the game has two immediate goals: telling the player what actions they need to do next, and reminding the player what they have already done.

Telling the player what they need to do next can take the form of a quest diary or journal that a player can refer to at any point. It becomes increasingly important as the complexity of games has increased. Introducing the 3rd dimension immeasurably increases the difficulty of navigating the playing space, as does other ‘realistic’ effects: dust, clouds, smoke, and moving the colour space towards brown, wet and reflective increase the requirements of filling your levels with shouty men standing where you need to be, pointing at things you need to shoot (This is epitomised in the single player campaigns in the Call of Duty series of games). In this sense, to rephrase Clint Hocking, if the best we can achieve with narrative in games is to prevent the screen scrolling to the left, then we have failed to use it to its full capability.

Reminding the player what they have already done is in some sense less critical: it helps with the elimination process for figuring out where to go and what to do – but the necessity of repetition in games may confuse even this issue. But it ensures the player knows what they can do – what capabilities they have at any point in time. More importantly, this aspect of narrative is what carries the emotional investment of the game – because it reminds the player if nothing else of the time they have already spent playing.

But what the narrativist debate in games is actually about is less the function of narrative, and more that this narrative experience requires an externally agency for it to be effective - an author, who provides the narrative to the player through cut scenes, traditional story telling mechanics like dialog and literally dropping story text, in the form of diaries and recorded messages throughout the game space.

My experience with roguelikes suggests external agency is not a requirement. Narrative exists in roguelikes in two forms, in a similar ‘what I’ve already done’ and ‘what I have to do’ form (it would be disingenuous of me to omit the fact I’ve chosen a narrative theory that provides this correspondence). The ‘what I’ve already done’ consists of after adventure reports (AARs) and day in the lifes (DiTL) written by players of roguelikes and posted to the forums and shared with other players. The ‘what I have to do’ form consists of a check list of items that have been learned by repeated play as necessities for successful progression in the game.

Narrative in this sense is a user generated experience – but other narrative forms are equally user centric. Literature provides this by mimicking the internal voice through the process of reading; film, television and first person shooters by superpositioning the viewer over the camera location. Games do this by creating a space where a set of psychological needs can be replicated and fulfilled. The difference between traditional narrative and games is that with games the player is not only required to read the narrative, but must make decisions to propel it forward.

The simplest narrative is takes the ‘and then’ form – perhaps told by a child. ‘And then we went to the park. And then it was raining. And then there was a red ball. And then I fell down.’ We are remarkably adept at constructing a compelling internally consistent story from this sequence of statements. But if I was to ask you come up with the next statement in the sequence of events, you’d immediately pause. You’d need to consider the likely sequence of events (was this child’s care giver there, were they seriously injured, is there a hospital near by, does someone have a phone) and provide a statement that follows consistently. That in itself is not difficult. But where the current gap with games occurs, is that you must not only consider a whole set of information, the game designer must have anticipated all of this criteria as well, and provided for the set of decisions you are likely to make. The correct answer, of course, is ‘And then Zoe helped me up.’ – it’s a retelling of a game of Left4Dead.

It is the gap between the decision making capabilities of the player, and the game rules built by the game designer that makes narrative problematic. We overcome this game through repeated play, as I suggested, through learning the rules of the game and improving our skill in navigating the game space. But linear narrative fails to provide a supporting framework for repeated play: ‘And then we went to the park. And then we went to the park. And then we went to the park. And then we went to the park.’ is a far less interesting a story. We focus a lot more on the exceptional events than the everyday. But we need that everyday framework to ground this in – reread the first example with the insertion of ‘And then zombies attacked.’ as every second sentence to understand how this framing helps understand context.

I firmly believe that roguelikes have solved the narrative 'problem' through permadeath: which forces the player to experience this repeated play, while focusing the mind of the designer on how to provide sufficient variety with each play through to make the experience unique. The roguelike narrative of a single play through is united, I’ve argued elsewhere, by the meta-narrative of repeated interesting and hopefully unique failure which guides the player’s learning of the rules of the game. The player overcomes essentially the same challenge over and over, as the player improves in their ability to use the resources they have, over an ever changing, ever escalating topology of surmountable obstacles. It is no coincidence that this also describes Shadow of the Colossus, a game which has been widely lauded for its synergy of game play and narrative.

This meta-narrative theory, replaying the essentially the same game over and over but with harder to solve problems, can help explain the strength of genre in games – when you move from Doom to Half-Life to Far Cry 2 you are not so much playing new games as replaying the same game again with more complex variations. And it can help explain why some narrative devices from other media fail to survive the transition to gaming.

Imagine you are a designer of a first person shooter which has included a successful widely read writer from the beginning, lauded for the previous work he has done in this game genre. Not only has he developed the plot, character designs, history and many script elements to support the kind of exciting, perhaps revolutionary game experience you are looking to produce, but he’s included a plot twist about a third of the way into the game which introduces and provides a coherent explanation for an innovative game mechanic element which features for the remainder of the game. How do you market the game?

In the example I’m thinking of, you go out, and demonstrate at every possible opportunity how the new game mechanic works, giving the media footage and press material showing off the technology, and help them inform and educate the gaming public. Forget the fact that this completely destroys the plot twist for anyone who has the slightest interest in the game: in fact, you’re setting this up so that someone playing the game will spend the first third of it wondering exactly when this mechanic kicks in.

This is comparable to the Wachowski brothers before releasing The Matrix holding press conferences explaining how the machines jack people into their power plants and the BTU output you can expect from an average adult male. Instead, the brothers chose to run one of the most innovative, audio only campaigns creating heightened anticipation of a movie property which had not yet proved itself, to ensure that the first viewing of their film leaves an indelible mark on the audience. The gaming equivalent I’ve referred to: Clive Barker’s Jericho.

From a traditional narrative perspective, the marketing decision made by the makers’ of Clive Barker’s Jericho makes no sense. The twist, a movie staple, works because subverting the expectation of the audience doesn’t suddenly cause the movie projector to fail, or the surround sound system to break down. But subverting the expectation of someone playing a game can result in this complete mechanical failure. If the player doesn't know where to go next, or what they can do, the game can fail.

In this light, it makes perfect sense to spoil the twist in favour of explaining how the mechanic derived from this twist is used in game. While this is a matter of conjecture, I'm am sure early on in user testing, the need to introduce the character switching mechanic to the audience as early as possible was identified and the marketing approach derived rationally from this. Clive Barker is intelligent enough an author to have deliberately designed the narrative twist which introduces this mechanic. He merely made the mistake of breaking the unwritten rule that any narrative twist should never affect the game play (See Metroid, Metal Gear Solid 2, Bioshock, Call of Duty 4: Modern Warfare etc. for examples upholding this rule).

Similarly, the absence of a falling climax or denouement in most game narratives, much lamented by game critics, is easily explained when viewing the game meta-narrative as an escalating series of repeated actions.

The unreliable narrator is another narrative staple from more traditional media. GLADoS from Portal is the best example of developing an unreliable narrator in a gaming medium: in fact, you are explicitly set up early in the game play to be shown her unreliability by being told by her you cannot complete a puzzle, in direct contrast to the expectation that completing puzzles allow you to progress. But when presented with the twist and after having been trained throughout the game to use the available tools to avoid destruction, it is still incredible to see the number of people who fail to read the situation, and proceed calmly to their death when ordered to do so (I too am guilty of this mistake). Only the game over screen rescues you from this misinterpretation.

So if these traditional narrative techniques fail to translate readily to the gaming medium, what tools can we use successfully to empower game narrative? That's part of what is so exciting about the medium - we are still in the process of building tools and exploring their utility. I'm going to suggest two that are successful: ownership through naming and resonance. Undoubtedly there are many more.

Holy holy invasion!

I don't know where to begin, but there are so many things I want to say about this game that I didn't realise I needed to play until I read this review. I'm torn between writing a clone or buying a PSP.

Let's just leave this at Spore done right.

Randomized levels in Diablo III

One of the contributors to the PCG wiki just highlighted a link to Blizzard discussing randomized level generation in Diablo III. Unfortunately, a little light on the detail. I was hoping for more.

Give me money

I believe I've discussed this before, but Gabe Newell is also now recommending that you give me money in order to fund me developing more games...

(By me, I of course mean all those amateur and indie developers whose games you enjoy. But also me.)

Bob Smith has died. Joe Smith is sad.

Half way through reading a #gamedesign tweet started by Soren Johnson, I came across this gem:

Harvey1966 @bbrathwaite: Another fave example: Sadness when a male X-com soldier died because another had the same last name. (Married?) #gamedesign

Thu Jul 16 22:35:00 +0000 2009

ClickNothing @Harvey1966 re X-Com soldiers... still reigning example of 'deeply moving dynmcly driven 'narr elements' - from >10yrs ago.. !!!

Does naming the procedurally generated content the player discover create more 'ownership' of that content?

(PS: Should I tweet? Could someone explain to me why this would be [insert positive adjective]?)

Running hot

The answer to my Cold Start question I posed earlier:

I have seen this problem previously caused by the CPU starting on a low stepping and failing to step up due to temperature of the CPU. This was fixed by the vendor (not Dell) releasing a BIOS which started the CPU initially on a high stepping and then stepping it down following POST.

Care to suggest any alternatives?