notes: the art of producing games
research notes from the art of producing games, by david mccarthy, ste curran, and simon byron. ilex presss 2005. notes mostly verbatim. emphasis mine.
pac man. holds the record for best-selling arcade game of all time. given the decline of coin-operated gaming, this is a record it’s likely to retain indefinitely. the game took eight people 15 months to design – four working on the hardware, four on the software. p16
the legend of zelda. despite numerous attempts, no other game or series of games has managed to replicate the mix of combat, exploration, and puzzle-solving with which the series is synonymous. p17
an unforeseen consequence of the introduction of more mainstream consoles was the death of the bedroom programmer. the ability to experiment with all aspects of game design was restricted to formats that did not require approvals from their manufacturer. as the open formats declined, so, too, did the commercial opportunities for those keen to design and publish any type of game. development teams of ten and above ceased being the exception – and throughout the decade (1990s) this figure often increased tenfold. p18
deus ex is the logical conclusion of several design trends that began to appear toward the end of the 1990s; its conspiracy-laden setting is home to a richly multilayered, emergent game universe that grants unprecedented player choice and narrative flexibility. thought no game has yet matched its scope, including its own sequel, many developers have begun to emulate its systemic design. p23
design document drawn up
playable proof of concept / vertical-slice demo created
technology and tools initiated
art direction finalized
funding and materials (e.g. dev kits) obtained.
team scaled up to full-size
producer oversees schedulling and logistice
programmers create tools and finalize technology: physics, AI, renderer, etc.
artists create and animate characters, textures, backgrounds, vehicles, etc.
level designers create levels and missions
sound designers work on bacground music, audio etc.
playable game is tested by testers
console approval met
playable game is localized for foreign territories
planning for sequels and expansions
game is advertised and distributed. p26
when a games goes into full-production mode, the team is scaled up, with additional artists, programmers, and designers coming on board. by now, a clear design should be in place, with an unambiguous direction in mind for visual style and the like. it’s at this point that programmers write the code that drives the game. using special development kits – scaled up versions of the target console that feature increased memory and a hard drive – they write engines and tools; while the game engine generates polygons, shadows, and textures, code is also required to simulate lifelike physics and the AI behavior of characters in the game.
during this process, the programming team works in tandem with artists, sound artists, and level designers, who between them create the visual appearance of the game, and populate it with enemies, objects, and sound effects. toward the end of production, a quality assurance department will test playable versions of the game to make sure that they function properly, and throughout the process, a writer or writers will create any necessary dialog or important background information. p26
level designer / mission designer
character artist / animator
localization manager. p28-9
a roadmap or flowchart of the game is required, listing the main sections that the player will experience, and the possible routes through to completion. depending on the nature of the title, this can be a straightforward, linear path or it can occupy dozens of sheets of paper taped together where player freedom is offered. p34
virtools offers a pc, linux, and xbox engine. incorporates a high-level graphical user interface and an extensive library of over 450 ready-to-use behaviors, combined with all the low-level access that programmers need to create custom features. p35
a publisher’s pitch wishlist
sales sheet – genre, target platform, general gameplay, new technology, key features
design document – comprehensive overview with more detail and breakdowns of characters, setting, levels, etc.
technical design document – backup procedures, version control, build
risk analysis – things that can go wrong
project plan – development schedule and software
cost forecasts. p38
the most common sources of funding, outside of individual investors, are either appealing to venture capital funds, or recourse to completion bond financing. the latter is a novel financial instrument that sees an insurance company guaranteeing, or ‘bonding’ interim milestone payments, with a publisher stepping in to pay the bond, plus interest and fees, when the development of a game is complete. p41
the current norm is for developers to use a small-scale team during the preproduction phase; to originate the design of the game, and to take the key decisions about its ultimate vision and direction. by the end of the process, this vision will be embodied within a variety of materials, such as design documents and technical demos, which will help the developer pitch the idea to publishers, and help keep the direction consistent even after the development team has geared up to go into full-scale production. by using a small team, costs are reduced, and projects can be started concurrently with other projects that may be drawing to a close. p46
a growing number of developers are eschewing the laborious process of creating written documentation and technical demos, choosing instead to pursue the more steamlined approach of the vertical-slice demo. instead of creating a written design, which may become superfluous, the focus is on creating a working demo that encapsulates all the features of the finished game, and which will be used as a consistent reference point throughout production. p50
the game designer must work closely with the producer to establish realistic milestones, and to make sure the team manages to hit them. this can involve brutal cuts to the feature set. p52
evolving from the god-game trend of the mid-eighties, the rts (realtime strategy) places the layer in a strategic military conflict, and challenges them to defeat their opponent. balancing defense and attack is usually the key, and good realtime strategy games offer the player more than just one way of approaching each battle. arcade skills are only sparsely required. this game’s about thinking. p53
though the genre (puzzle) seems to be considered a kiss of death by many publishers these days, an excellent puzzle game can be produced on a shoestring budget and has the potential to reach many more people than, say, an excellent firstperson shooter. there are very few people who don’t know what tetris is. the key, of course, is finding a balance of reaction and intellect just as sublime.
all videogame rpgs (role playing games) owe their existence to the pen-and-paper version of dungeons and dragons, and much of that heritage is still visible today. payers take the role of an adventurer and embark on a series of quests that ultimately lead to the narrative’s resolution. combat in an rpg can be based on arcade-style reactions, but more often than not it’s a mathematician’s delight, the computer rolling a hundred dice to decide who wins. p54
the most sedate of genres, the management game gives gamers a chance to play god, or at least the head of a medium-to-large organization. good management games manage to involve the player in what could, superficially, seem like the gaming equivalent of spreadsheet software. in other words, play down the bureaucracy and play up the humanity, hide the math and give the player plenty of feedback.
massively multiplayer games require a decent internet connection and a desire to play games with strangers, two things that automatically exclude a good percentage of gamers. those that remain participate in one of gaming’s newest forms, and one whose experience is almost entirely dictated by the players themselves. many new challenges await those who would design a massively multiplayer experience (everquest, phantasy star online). p55
if a team member has a question, he or she refers to the design document first. they should not have to refer to anything (or anyone) else. the design document is a concrete guide to what happens where. a kind of navigational chart, without which the design team are lost at sea.
a truly emergent game (of which there are only ver yfew) gives the player a sense of greater control and freedom within the game world. as games are about control at heart, having a greater breadth of possible objects and systems to control can make the game more enjoyable. it is however also possible to make a smaller, tighter set of features equally enjoyable, by making each individual interaction as strong as possible.
there are two main benefits of systemic approaches. firstly, it is more efficient from a developmental point of view, because you are able to reuse assets and technology in different instances. the fewer special cases you have in a game, the less developmental time you need to implement things. the second benefit of systemic approaches is that the player experiences a greater consistency in the world. this consistency helps them construct strategies and make assumptions in a way that makes them feel more powerful. p62
i want the player to avoid failing as much as possible, while always being aware of the threat of failure.
i try to make low-level gameplay elements as enjoyable and polished as possible
i try to keep a strong disciplines image of the end product in my mind all the time.
i try to work upward in a hierarchy in the course of the project. for instance, working on low-level elements (e.g. control) and trying to lock them down, then moving to mid-level elements (e.g. combat) and trying to lock them down, finally moving to work on higher level elements (e.g. missions) and trying to lock them down. you should try not to move to another level in the hierarchy until you absolutely satisfied you have solved all of the problems in the previous level. p62
in 2004, he estimated that game developers spend 80% of their time creating technology and assembling the game, and only 20% implementing and refining game features and artwork. so it’s clear that choosing or creating the right tools, and making sure that they are stable and efficient, is of the utmost importance to the quality of the finished game. making the wrong decisions during preproduction can produce far-reaching complications that disrupt the entire development cycle. p64
one advantage of creating in-house tools for a pc title is that they can be shipped with the game to enhance their appeal on store shelves. by allowing users to create their own content the replay value of a game is enhanced, which makes it a more attractive purchase. but the discipline of creating tools with a nonprofessional user-base in mind can also be beneficial. probably the most desirable characteristic of any development tool is a user-friendly interface, and designing tools that are going to be released to the public forces tools programmers to address this issue of usability. this also reaps positive repercussions for their use internally by the likes of nontechnically skilled artists, testers, and so on.
half-life in particular managed to give rise to a burgeoning mod scene made of of talented amateurs who use the existing game as their core technology, modifying it to produce an entirely new product. some popular mods even manage to attain commercial release, such as the phenomenally successful counter-strike. quite simply, the mod scene is the closest equivalent that the videogame industry has to the independent movie scene, allowing games to be created for a fraction of the budget of full production models. p66
havok physics. the trend in videogame design toward emergent gameplay has heightened the necessity for featuring realistic physics, once the sole preserve of racing games, across all genres. havok’s cross-platform technology has been used in a number of high-profile titles, most notably valve’s half-life 2, to provide features such as ragdoll effects, detachable limbs, and vehicle physics. it is supported by a set of tools that are compatible with a number of other content creation packages. p69
alienbrain studio is an asset management tool that allows developers to version, track, and store all of their game’s assets during and after production. not the most glamorous end of the middleware spectrum, but essential nevertheless.
one of the biggest videogame releases in recent years, valve software’s half-life 2 is built upon havok’s physics middleware. havok’s technology provides developers with tools for content creation, tweaking, debugging, and profiling, and it allowed valve to fundamentally redesign the core mechanics of its paradigmatic firstperson shooter. p72
increasingly a specialized art, a programmer’s precise role in game creation depends on their skillset. the most ‘hardcore’ position of all is usually considered to be the engine programmer, the man brave enough to take the machine right to its limits and create a framework on which the game’s structure and scenery can be hung. good engine coding doesn’t just require low-level assembly language, but an obscenely acute mathematical brain. others may be more at home working on artificial intelligence routines, where an understanding of how to transfer human logic to an artificial environment is the base currently. online specialists need to combat latency in telecommunications. tools teams work to provide tools that enable designers and artists to work more easily, or tweak middleware to their team’s requirements.
the specific area in which a programmer works will largely dictate how they spend their time over the course of the project; engine programmers will work with the art department to decide if and where visual compromises need to be made in order to preserve the game’s fluency, while AI coders pay attention to the criticisms of testers and designers, and tweak their routines appropriately. p80
the ‘camera’ is the position from which the players sees the action, and getting that position right in thirdperson perspective games has become something close to an art form. and, like all art forms, there can be no correct answers, but there can certainly be things that are horribly, horribly wrong. the problem is simple. movement in 3d games must be relative to the camera, because otherwise there is no way of mapping the control system to visual feedback; a hundred-and-eighty-degree shift of the camera will leave the controls backward, for example, and is completely counterintuitive. but a camera cannot be rigidly fixed to the back of the player – it must be flexible enough to move around objects fixed in the environment that would prove impossible for a real, solid camera to pass through, and intelligent enough to ensure that the player is visible at all times. more than that, since the player’s control system shifts slightly every time the camera moves around, ti must be wholly predictable, never twitchy, and never frustrating. in essenced it must do what the player wants it to do, second-guessing faultlessly throughout an unpredictable adventure. and that’s a challenge.
more and more games are being shipped with online components, causing more and more headaches for programmers forced to rely on the unreliable nature of third-party communications hardware. the problems for programmers of persistent worlds start with the simple conundrum: should game data be stored on the user’s local machine, or locked safely away on the server side? well, the server side, obviously, because anything else is insecure and can lead to hacking, cheating, and devaluing the game experience for everyone. but what data gets stored, and where? every byte transferred has to be transferred by thousands and thousands of players, and that’s going to cost money. how many servers does the company need? and even then, will that system be secure against the latest console cheat devices?
and that’s a slow-paced genre. speedier things become simply impossible. p84-5
physics is a buzzword in game development at the time of writing, but the appearance of emergent gameplay is causing problems for the programmer. players want worlds to behave with as much realism as possible, but computes aren’t anywhere near capable of reproducing real life particle by particle. shortcuts must be taken, but which ones? p85
middleware libraries were the next logical step, enabling teams to use off-the-shelf game components like renderware graphics or havok physics to avoid the work of developing those pieces themselves. then, the large-scale move to game engines began, enabling teams to start with a 100% complete, fully integrated game framework, and focus on those areas of gameplay and technical features unique to their game.
an AI programmer’s job is to create a software framework suitable for managing the movement and interaction of a diverse set of computer-controlled characters; this is a matter of engineering. but his job is to also make those characters interesting to interact with or challenging to fight; that’s a creative task. p89
a good level designer has to be, like so many on the development team, a multiskilled creature. since level design tools vary from one game to the next, it helps if they’re able to learn fast, but those with a good grounding in the pc mod tools bundled with many modern 3D games will find the technical aspects of creating worlds little trouble. level designers are the architects of the game-design world, constructing environments to specifications delineated in the design document, and a fine sense of shape and dimension is an essential trait for the role. p90
epic and id’s policy of bundling their tools with their games means that there are more budding fps (firstperson shooter) designers across the globe than for any other genre. the nature of the game, too – run around and shoot just about everything – makes getting a functional level running pretty easy, but just because it works doesn’t mean it’s any good. level designers working on narrative firstperson shooters must cloak their game’s (almost obligatory) linearity with clever architecture. those working on levels for multiplayer combat must create a rock-paper-scissors style hierarchy of locations for the player, so that every advantageous spot makes fighters vulnerable in another respect.
key to both types, though, is pick-up placement. players should often feel on the brink of running out of ammunition, but rarely actually fall completely short. likewise, the locations of health packs must be far enough apart to cause consternation, but just within reach of the player as they brace themselves for game over. p92
the key word here is ‘adventure’: one of the main motivations in a standard tpa (thirdperson adventure) is exploration, which places pressure on the level designer to create an environment worth exploring. if the player doesn’t find the world aesthetically interesting, they won’t explore it; equally, if they don’t think there’s any point in twisting along that precarious precipice because the game’s never offered them rewards before, they’ll be uninterested in risking their lives. p93
vice city – the game world, and one huge level – was designed to exude entertainment at every street corner. streets unexpectedly seque into river-jumping ramps, stairwells become routes to stupid stunts, and the airport isn’t a real airport, it’s just an excuse to crash lots of cars. it doesn’t matter that the mission design within the world is hit and miss, because there’s so much fun to be had in ignoring the challenges set for you and enjoying the rich interactivity of the game world.
lemmings has a gentle difficulty curve, and the first handful of levels teach the player what skills their tribe are capable of, and how to use them in combination. coming up with a way of doing that, but still keeping the levels involving and entertaining is as big a challenge as setting relatively incomprehensible problems what will stump addicts for days. p95
this is probably the most important element in a successful level (tweaking). with so many creative talents on the team, it would be foolish not to invite and heed their input. similarly, play testing is a crucial part of building a level. play test as soon as you have a working model of the level, revise based on feedback, then play test again. p97
newtek’s lightwave 3d chooses simplicity to court its customers. evolving from the amiga-based video toaster that proved popular for low-budget cgi in the early-to-mid nineties, lightwave 3d’s brutal animation system is compensated for by the speed at which it lets users create scenes. in other words it’s simple but effective, with a deceptively large toolset, and it’s increasingly popular with smaller companies that are producing games within short time frames. p103
games which aren’t tied to standard earth rules let their artists’ imaginations run free. rez is a superb example of constrained abstraction. faced with the opportunity to create five different levels representing the internals of a computer, the designers take electronic, organic, and mythological themes and weave them together with a sophistication that matches the game’s synesthetic background. artists plunging into abstraction’s deep space must work within their own guidelines to create something that is still internally consisten; a complete free reign would create a visual mess. p104
the solution, as demonstrated in timesplitters 2, is an approximation of reality. the game has plenty of human characters, but they’re rendered in an exaggerated style that intuitively tells players that they’re not in a space that follows humankind’s visual rules, and allows the game designer to render a coherent world well within the capabilities of the system. the main problem is getting the style right; while few people have a problem with the way the real world looks, and artistic interpretation is much less universal. p105
we’ve come a long way with building static worlds that are visually rich, back up the action, and make it dramatic and exciting to play, worlds which are composed and lit to look stunningly beautiful. the trick is going to be to keep all of these qualities and push our worlds to be even better, while at the same time letting people interact more. once everything has the potential to move, shift and change, then keeping creative control is a new challenge.
animation and AI cross-over: at the moment, games characters are rather lifeless and wooden. in the future i expect to see game developers spend much more time on animation systems that put much more characterization into the game animation itself rather than cut-sequences…i can see the game animator’s role changing significantly with an entirely new set of tools to allow them to ‘direct’ virtual actors and define their personalities and responses made up from much more complex blends of animation data. p106
frequently, traditional rules need to be ripped up. composition classes often emphasize the importance of repetition – and videogame music may have historically been famous for tunes based around a single central theme – but tastes have changed. modern gamers expect diversity. a 30-hour game needs more than 30 minutes of music.
research has shown that videogame players can rate the audio track forming as much as 30 percent of their overall enjoyment of a given game. p113
each level begins with a simple techo beat. enemies appear on screen and can be shot individually, or grouped together in a high-scoring chain. gameplay-wise, it’s as on the rails as you get – but the real interaction comes from the tv’s speakers. every time an enemy is targeted, it emits a beat – be that a synth, a rim-shot, or any other type of electro pulse. as they’re destroyed, they also emit a sound – not an explosion as such, but yet more beats. everything is timed to perfection. rather than producing a rambling, jumbled mess, the techno soundtrack works brilliantly, even appealing to those who despise the music-game genre. most who fall for its seductive beauty find themselves treating it more like an album than a game – simply skipping back to the start and listening to it all over again. p114
you’ll be able to have contrasting melodies – say, one ‘good’ piece of music over a ‘bad’ piece – played simultaneously. it’ll be new music that we’ve never heard before. people will be able to work interactively with it because they’ll have grown up with it. it’ll house horrible to us, of course. p117
movies and tv are linear experiences. each moment is timed to perfection. you’ll tug at the heart strings of someoen and then maybe throw something else straight at them. but all the time, the music is timed perfectly to the flow of the film. but with games, the player is directing it.
take something like a thirdperson game. you can have a bit of music that represents going to the next puzzle, which can be a linear piece. but then when the player reaches this puzzle, hours could be spent trying to solve it and immediately the linearity drops away.
people often get stuck over this idea of interactive music. but it’s quite simple to produce a never-ending piece that reflects the state of the game and still remains musically ‘correct.’ we do this by creating a pool of musical phrases, which we splice together at random. if you have enough of these you can create the illusion of a never ending piece. in simple terms, all you then need are three states – neutral, good, and bad. it’s then just a case of swapping between these three pools depending on how well you’re donig in the game. p119
while the arrival of movie talent has met with varied success, there can be no doubt that standards in game narrative are improving. and this is necessary for the medium to be taken seriously. often, videogame stories are rubbished as cliched and derivative. the industry itself seems to dismiss the importance of plot, choosing instead to emphasize in the box-blurb the number of levels and weapons.
the adventure genre still leads the way in videogame narrative – though that has diversified over recent years to include more action-orientated titles.
city of heroes features 20 different ongoing story arcs as villain groups menage paragon city and react to player victories and defeats.
most games tell stories using linear sequences of cut-scenes to top and tail the sections under the control of the player. although cinematic techniques have improved over time, it’s remained this way since the industry experimented with interactive fiction. bold attempts have been made to try to break free from this rigid format – but as games such as tomb raider have proved, it can work very well.
there are three vital documents necessary in developing a strong in-game narrative. the first is the story, told from start to finish. this, typically, will be under 20 pages long, and will detail the events specific to the game. locations, actions, ‘inciting incidents’ – all need to be scribbled down in chronological order.
games should adhere to a structure – for many, the three-act structure is suitable. this structure includes elements such as the ‘inciting incident,’ which is an event or action that radically changes the balance of the protagonist. in a movie, the inciting incident needs to happen within the first quarter, but in a game many argue that it must come much sooner – ideally in the introduction.
if the inciting incident doesn’t happen immediately, the game needn’t be pedestrian up until this point. prince of persia: the sands of time – a game with a wonderful narrative, and perhaps the most perfect ending ever featured in a videogame – kicked off with a dramatic set-piece, but it wasn’t until those sands of time were unleashed and the wider picture revealed that the game truly began.
the inciting incident will set up what many call the ‘obligatory scene.’ that’s where the main character is going to resolve the tale – be that through conflict (that is, defeating those who sought to harm him) or restoration (putting things back to how they were). it’s called ‘obligatory’ for a reason – the audience expects it, and will be left naturally disappointed if the ultimate conclusions doesn’t meet that expectation.
the end of every ‘act’ should build to a peak, ensuring the pacing rises and falls. these peaks should build toward the end of every act and drop off at the beginning of the next section. it goes without saying that the end of act 3 should boast the largest climax of them all.
once you’re happy with your plot, it’s important to draw a detailed timeline of what every character featured in the game is doing and where that character is at any one time. often this information won’t be used directly – but the audience can sense whether you, as a storyteller, really know your world. if you don’t, they quickly stop believing. it will also provide you with natural references to integrate into the script.
finally, it’s useful to have a background document that details every character’s history, along with a summary of the overarching plot. this will provide a rich source of additional information that can be intoduced into the script where appropriate, and provide the writers with the background that they need to get the characters’ dialog and voices right. p121-23
the broken sword series success is due to two main factors. in the trilogy to date, the narrative has drawn on real-life mysteries and legends – the knights templar in the first, the mayan prophecies in the second, and the voynich manuscript in the third. this has not only provided the series with a solid backdrop, but also piqued the interest of many who have played the game.
more obviously, much of broken sword’s success can be attributed to the will-they-won’t-they? relationship of its central duo. goerge stobbard and nicole collard’s continual fighting is typical of many male/female pairings, but it’s rare to see such interest handled so eloquently in a videogame.
of course, the game’s style helped enormously. but rather than rely on long cut-scenes to advance the plot, cecil keeps the action in-game and dialog-driven. even in the third game’s attempt to drive drama through video sequences, players needed to remain alert, as life-or-death decisions were usually not far away. while these ‘action sequences’ were not entirely successful, they do provide alternative thinking for those wishing to avoid lengthy cut-scenes. p124
warren spector has almost single-handedly taken it upon himself to pioneer the nonlinear action adventure, with deus ex proving to be his most successful attempt to date. while the standard plot – shady government agency, viruses, terrorism, and so n – could have proved derivative initially, it was the storytelling skill and implied freedom that really captured the imagination. most impressively, he did this in a firstperson game – typically the most linear of all genres.
of course, no computer or videogame could ever be truly nonlinear. but spector and his team structured the game’s back-end to offer the illusion of freedom, advancing the same story through many different mechanics. players with a penchant for gunplay could find npcs reacting to them differently later on, aware of their hardball reputation. brilliantly, the game’s expansive back-story is never fed down the throats of its players. a wealth of background information can be found in various documents and computer terminals – as well as from in-game characters. kojima would have forced players to read every single character. spector has enough confidence in both his players’ intelligence and his own storytelling ability to permit a greater degree of freedom.
cut-scenes are rarely intrusive – a necessity, thanks to the game’s perspective. switching from firstperson to thirdperson would detach players from the in-game world, shattering the feeling of existing within a realistic environment. indeed, it’s the consistency of the story and the aplomb with which multiple solutions were designed and implemented for almost every section that enabled players to become so immersed in a rich, atmospheric tale. p125
planescape torment was based around an interesting concept. a traditional role-playing game set in the advanced dungeons & dragons world – with its familiar rules and structure – it ran the danger of disappearing among a sea of similar titles. but what made the game stand out was the quality of the plot and the strength of its dialog – two things rare in the world of videogames.
assuming the role of a character who cannot die, planescape’s world is structured but nonlinear. players opt to progress as they see fit, treating npcs with contempt, admiration, or something in between. this could have resulted in a jumbled mess of conversational anomalies, but the sophisticated storytelling engine permitted the developers a huge degree of freedom – freedom that is handed over to players as they journey through torment’s world, forging friendships, alliances, and creating grievances as they see fit.
the player’s moral alignment and affiliation with different factions are impressively flexible and have a significant impact on the course of the game. characters can switch character class allegiances on whims – unusual for a genre that typically restricts such decisions. the important premise actually works for the game, rather than against it. what’s more, the lead character’s immortality means that players rarely abuse the save or load system, maintaining the solidity of the game world.
planescape marked a bold step in videogame narrative, and one which was not entirely commercially successful. a shame, as its brave attempt to introduce intelligent role-playing should have been rewarded with much more than a swift appearance in the bargain bin. p125
there is no doubt that plot must support gameplay – gameplay is king. but if the plot and the gameplay are genuinely intertwined then they will need to be designed in parallel. ideally elements of the plot would develop systemically – so the author designs the rules for the plot which the player then creates. to an extent this happened in grand theft auto iii – with an overarching linear plot running in parallel with player-created systemic subplots. this is one of the reaons why the game was so well received. in doing this, we got a glimpse of the future of storytelling in the interactive medium.
but isn’t gta’s narrative actually fairly basic? it’s a series of missions interspersed with cut-scenes…clearly, in gta, you had the overall linear narrative, but within it you had player-created systemic substories. that’s what makes the game so appealing; the freedom of the systemic, within the structure of the linear. that’s why it worked.
i would argue that its staccato storytelling is part of the game’s strength. the player can play the game at their own pace, and the pace rises and falls with a great rhythm throughout each mission.
in an adventure there is a lot of story to tell. a golden rule is to hide the exposition. we do this by integrating the story and the gameplay – which means that our linear cut-scenes need only be relatively short. we also realized very early on that if you have two protagonists, then those characters can exchange dialog that can be witty, fun, and also convey exposition.
you can hide exposition in several ways. one effective way is to integrate the narrative with the gameplay itself – so as the player advances through puzzles, the story is gradually revealed.
the other way is, instead of having a few great chunks of exposition, you actually break it up into lots of little sequences. so you’re continuing to entertain the player, hopefully making them laugh, hopefully making them gasp, but at each point the story is being drip-fed, so we don’t need the extraordinary lengthy cut-scenes that have become prevalent in many japanese games.
what we could do in our first game lure of the temptress is not that much different from what we can do now. we designed the game so that the player could go and talk to a huge number of characters about what was relevant at that point, but as the game advanced, their dialog was replaced to reflect new situations. and that was quite interesting because it meant you could go back and talk to people, and they would say completely new things based upon the fact that the world had moved on.
the other thing that people really liked was that we had characters who’d walk around doing their own thing – a system that we called virtual theater. i’ve always regretted the fact that we didn’t really exploit virtual theater in later games. but the way that characters had freedom to do what they wanted to conflicted with the adventure idea of a multilinear narrative in a nonlinear environment. p126-8
the crash bug is truly the bane of the developer’s (and tester’s existence. pity the gamer then, who, for no apparent reason, is forced to witness their game resetting itself. unless they’re playing the magnificent eternal darkness, which uses simulated technical flaws and glitches to evoke a deeply unsettling atmosphere as your character’s sanity gradually dissipates. thus, during key sequences depicting the gradual onset of psychosis, the player’s tv appears to change video modes; items appear to go missing from the inventory; the game will apparently reset; and, most unsettling of all, the game appears to delete your saved game after ending a game session. p135
while conventional games have budget in the millions, high-quality cell and tv games can be produced for less than $150,000, typically by teams of three to five – although often by individual lone programmers. and whereas console games can take between two and three years to develop, cell or set-top box (stb) games are developed in a few months.
in every developed country apart from america, a higher proportion of the population owns a cellphone than a computer. there are well over a billion phones in existence now, and with each new generation the hardware specifications improve. the current wave of 3g phones offers a quality of gaming comparable with early playstation games – albeit on a smaller screen – although the market for these high-end phones is quite small at present.
on standard cell phones, it’s a crowded market. there’s currently no approvals process for games developers and publishers, so anyone can, in theory, produce and distribute a cell-phone title. intellectual property abuse is rife. p178
games distributed free to consumer are funded by revenue from advertising partners. this model requires a huge audience in order to attract a business-sustaining income which means the game itself should avoid anything that will restrict any potential players from signing up.
for this reason, these games are usually browser-based, scripted in java, shockwave, or flash – and so can, in theory, be enjoyed by anyone who can connect to the internet. often card, puzzle, or simple arcade games, they eschew sophistication in favor of attracting and entertaining a mainstream audience. these games can be developed by almost anyone as the programming languages are relatively unsophisticated, making them the perfect platform for aspiring developers keen to hone their skills. however, without the power of a major portal behind them, they’re unlikely to hit a widespread audience but can still be attractive to advertisers as part of viral campaigns.
the pay to play model can be divided into three main categories. monthly subscriptions vary from a few dollars to an average of $15 a month, and offer unlimited play over the period. like a broadband internet subscription, which is usually required in order to connect at suitable speeds, these plans are perfect for heavy users or those who prefer to calculate their monthly outgoings in advance. this is by far the most popular model for mmogs.
metered subscriptions are the equivalent of the cell-phone pay-as-you-go scheme and are perfect for those not willing to commit to a longer contract, although they can often work out more expensively for habitual users. some games, such as z-opolis, offer a sliding scale that makes playing for longer more economical by reducing hourly rates the lnger a user is online.
the one-time fee is a model that is usually offered in conjunction with other revenue schemes. at its crudest, it’s the distribution in a boxed, retail version, usually with playing time bundled as part of this cost. credit card details are taken at registration, with the player asked to actively cancel and re-subscribe to a full subscription (rather than upgrade) after the trial period. one-time fees can also be used as entry fees to special items, tournaments, or as payment for additional or episodic content.
it may be necessary to team up with a third party in order to meet the technical requirements of an online game. a partnership with an isp such as aol or msn or a portal such as yahoo will offer the necessary exposure to a large customer base. additionally, dedicated gaming sites such as gamespy and mgon can help in driving consumers to specific online games and often produce sdks to enable developers to integrate their games with the site’s infrastructure.
dedicated server support can obviously add to the running costs and will usually fall under the remit of the publisher. however, more and more games are embracing peer-to-peer technology. this neatly sidesteps the need for dedicated servers, relying instead on surplus player processing power to host networked sessions. p180-1
on average, online gamers stay with a game for 10 to 12 months, which means that about 10 percent churn out of a game in any given month. developers continually update their games, and add upper level depth in order to reduce churn.
new massively multiplayer online games have about 20 minutes to attract a new user into a game. in that 20 minutes, a new user should be able to do many thing. they should be able to create a character. they should be able to take that character on a mission, and not be killed. indeed, they should win the mission – and be told how great they are, and given awards and honors, etc. they should then be sent on an additional mission. most mmogs today do not allow the above, but the newer-generation products we are working on currently will be designed this way.
currently there is lots of item selling and trading, but it is not sanctioned by the companies publishing the products. this is a legal limitation, not a technical limitation. if publishers acknowledge that these items have real-world value then they also have real-world liability if any of the items are lost or stolen in the game, and the publisher does not replace them. for this to change, the legal atmosphere would need to change first. p182
the main problem so far is that 99.9 percent of online games are medieval fantasy role-playing. p183
the main difficulty with controlling player action comes with too many people gathering in one place at one time. when this happens, it stresses a couple of things: the graphics-generating capability of many home computers, and the processing power of the single server that is hosting the large group.
in order to overcome this issue, many games are designed to spread people out among the many servers by starting people at different locations, and having many different paths to complete a mission. in city of heroes, we also have the ability to create multiple copies of the same exact area when too many people want to be in a certain place at one time. p183
gamers want action and excitement, combined with depth. old-style mmogs gave only one or the other… next-generation mmogs will deliver both. p184