After being in transit for six hours, I finally arrive at Liverpool, home of the nascent GDTW conference, now in its fifth year. The programme of talks this year isn't the most enticing thing I've ever read - it has an academic bent to it that is a little alarming, and I'm sure I saw at least one of the talks here on the programme at GDC 07 (Ed: yea, it was Introversion's talk) - but we'll see how it goes.
First-day registration goes smoothly. I'm a little dismayed to discover that my pass doesn't include the "gala dinner" on the first day - so, we'll all just have to close our eyes and imagine what "Sapporo Teppanyaki" is like - but never mind. After getting my badge and handout - agenda, blue plastic folder with notepad and advert for Liverpool JMU, and brochure from Microsoft about the Imagine Cup - I take position on a sofa outside the lecture hall and begin writing this report, while various other attendees walk past me to register. They seem to be mostly students, though maybe I'm just being misled by game developers who look like students. Oh, though just after writing that, some people who look slightly older have arrived. I'll see if I can get a glimpse at the registration sheet to see where they're from. When a conference is deficient in content, it's compensated by networking opportunities, so we'll see how it balances out.
There's a lot of stuff on the agenda for the day. The morning comprises three "keynote" speeches from Luigi Tramontana (Craft Animations), Mark Morris (Introversion - I think he was the one who swore on stage at the Game Developers Choice Awards 2006 and got a standing ovation?) and Darren Jobling (Eutechnyx). Then we'll break for lunch. Then it seems I've got a choice - Dave Burrows (SCEE) and Simona Tassinari (Rare) are giving talks on one track, while the other track seems to be packed with 6 twenty-minute sessions back to back. I think I'll go to the short session track.
Picking up again after a coffee break, I almost double-book myself thanks to a typo on the schedule, but I spot it in time. I can choose between Dr David Stern (Microsoft Research) and Toby W. Allen (Sumo Digital) talking about machine learning and placements respectively, or I can go for more paper presentations. Again, I think I'll go for the latter - if only because the titles are so cryptic that I want to find out what on earth they're actually about.
Here we go. I've set up base camp at the back of the conference hall (the only seats near enough to a power socket...) and the welcome speeches are about to start.
Dr Abdennour El Rhalibi, the man in charge, and a man in desperate need of a microphone, quietly introduces Prof. Diane Meehan, a Dean from LJMU, who speaks about the university; then Prof. Madjid Merabti speaks about the conference itself, mentioning that it was originally started just for LJMU students and has since expanded to accept delegates from outside the university. That explains some things.
Prof. Diana Meehan
Prof. Madjid Merabti
- "Long papers" session
- Citygen: An Interactive System for Procedural City Generation
- Rapid Development of Game-like Interactive Simulations for Learning Clinical Procedures
- Pedagogy Elements, Components, and structures of serious games authoring environments
- Design and Development of Game Content and Mapping for the XNA Environment
- User-generated content sharing across different virtual environments
- Afternoon Sessions
- Psychology and logic: design considerations for a customizable educational resource management game
- The resonating spaces of first-person shooter games
- Gameplay Design Patterns for Social Networks and Conflicts
- Creating an Online Multiplayer Game with Microsoft's XNA
- Topographic Products of Experts applied to a Motocross Simulation and Simulation Stabilisation
Luigi Tramonata, CTO of Craft Animations, gives the first keynote speech about his company's technology "SpeedAnimation." Founded in Gothenburg - Luigi helpfully takes a few minutes to explain exactly where Sweden is, and why it's nice. The nightmare situation that SpeedAnimation sets out to solve is this: A week before release, the publisher says you need more in-game animation - a lot more. An animator can produce, on average, two to three seconds of animation per day, so that's obviously a problem.
The first observation Craft made was that you have to abandon the laborious manual keyframing approach and go for simulation instead - procedurally generated animation. Secondly, it needs to be possible for the author to interact with the animation while it's being generated. The result is a system where the animator "drives" the animation using something as simple as a regular game controller, and the system "records" the animation, generating keyframes at the appropriate times. It's a kind of puppeteering approach, incorporating game technology itself into the animation pipeline. Thirdly, it needs to be modular - combining separate elements together into a single animation instead of having to record everything separately. Craft's primary toolsuite covers wheel-based vehicles, aircraft, and camera control tools. Further rigs are available as freeware plugins.
Next he spoke about the animation pipeline and workflow for using this system. As much as it may be easy to create animation using the game controller, creating the right animation - creating something that is precise and that is what you want - is harder. So, he gave us the four "secrets" of the system: automatic control (that is, AI assistance); "autonomous mode" (following pre-laid paths, though possibly allowing the user to override the path with the controller); "low quality to high quality" (creating the initial recording using low-accuracy physics and low polycount to achieve good framerate, and then rerunning it with high-quality physics and high polycount for the final result); "hierarchical camera techniques" (stacking camera effects up by parenting cameras to one another); and "pre-rigged models" (setting up your models with effects and so on before animating - I didn't really understand why he was pointing this out because it seems like if you didn't do this the system wouldn't work at all). Rigged models for the system can be created in-house; Craft also provide a rigging service for you if you'd prefer. We finally got to see a live demo, too, which was quite nice.
Finishing up, Luigi spoke about what they're working on for the future. Discrete Differential Geometry based physics were mentioned, giving "soft body" style physics; Character/facial tools, something that had been conspicuously absent from the rest of the talk; network based systems, such that multiple people can participate in "acting out" a scene simultaneously; sound, eventually; and lastly, an SDK for developing new plugins. Then he dove into undisguised sales pitch - positive press responses, customer listings, testimonials, couple of demo videos, etc. I sort of switched off at the end there.
This was something of a "history of Introversion" talk.
Uplink, their first game, came out in 2001. Originally they wanted to make it 3D, apparently, but after simply failing to make it work well they dropped it and stuck with 2D, focusing on making the game immersive rather than graphically impressive. With a business plan, they got a grant from their university, set up a website, and posted on a couple of hacker community websites. Within a month they were unable to continue burning and packaging CDs themselves and had to work with an external manufacturer. With web sales established, they got a distributor and got into the high street stores as well.
Next came America, and that's when things started to go slightly wrong. They signed with an American publisher who then went bankrupt; they wasted money on advertising that didn't yield worthwhile results; and they wasted money at E3. Still, things overall went right - Uplink had cult status, good press reactions, strong sales (even without an American publisher paying them) and a thriving community.
Next came Darwinia. They decided early not to try for "photorealistic" games, acknowledging that they didn't have the capacity to generate content on the same level as the "big boys." Darwinia started out as "future war," a mere concept of battles between gigantic armies on CG terrains. It took 12 months before the actual game mechanic started to become clear, and Introversion were almost out of money at that point - they estimated they'd survive another month. At 31 months it was finally ready for release...
Another quick post-mortem: It took far too long to create, the initial demo was poorly constructed (hard to get into, and not much fun to play, turning off many potential buyers), and they overpriced it. However, they won the IGF awards (worth $25K), got reviews in the 90%s, Steam was a massive boost to their distribution numbers, and the icon of the Darwinian itself has become a strong brand identity.
Next, their most recent game: Defcon. This time, they took the right approach to time management: Chris Delay, Introversion's lead designer and coder, decided to try and create a game based on the film Wargames in 24 hours. By the end of it, he'd produced something with a familiar-looking world map, but not much more; so he decided to see how much he could get done in 7 days. Fast forward to day 357...
Postmortem time: They got the release time wrong by an hour; all their multiplayer servers died on launch day because of the high load; they had to spend 8 months getting the Mac and PC clients to successfully play games together; and they couldn't handle all the support requests. On the right side of things: 85% reviews, only a year in production, and 30,000 sales in the first month.
So what's next? Multiwinia, a new game set in the same world as Darwinia, due out next year.
So how has Introversion survived? Mark explained their thinking:
- Create radically different, deeply creative games
- Require creative freedom, so we work to minimize restraining influences like publishers and IP owners. Introversion haven't yet partnered with a publisher until the game is finished.
Keeping away from publishers is obviously not a common approach within games. Mark posed the question: what is it that publishers do for us, exactly? They "help" with development, by providing cash, market research, and testing/localization support; they handle the sales channels, dealing with distributors and platform holders; and they do your marketing.
But most of that, Mark says, we can do for ourselves. There are companies around to do testing/localization, and you can perform your own market research (it's not like game developers don't know at least something about what gamers want). Digital distribution takes most of the hassle out of the sales channels, and some distributors are explicitly working directly with developers anyway. And you can do your own marketing - it's not like publishers ever did that good a job of the marketing anyway, being as they were distant from the people with the best understanding of the game and more interested in putting their marketing dollars behind the guaranteed hits. The only thing that's really left for them to do is fronting the cash, becoming straightforward investors in the project. Even then, Mark warns, you need to be very careful about what freedoms you're giving up to get that cash.
Talk turned to IP, a subject fresh in Mark's mind as he'd given a presentation on it fairly recently. His position was as an advocate of developing your own IP: you've got fewer creative restrictions, reduced overheads (specifically in time spent bidding for licenses, and in time spent pandering to license-holders' demands), and the well-discussed commercial advantages of owning your own IP (opportunities for licensing it out yourself, for example). He also pointed out that by putting your company behind its own uniquely identifiable IP instead of taking on work-for-hire from other companies, it puts your reputation in that IP - in your games - rather than in your ability to deliver things on time and on budget for other developers. It's a higher-risk situation, of course, but it's potentially a much higher gain.
- A fundamental belief in pushing the boundaries of game design
- No publishers and no license holders in the way means you can stay as close as possible to the end customers - the gamers.
This was a very business-focused talk, as you'd expect; quite a lot of facts and figures were being thrown around.
Darren is the directory of business development at Eutechnyx - he "brings in the deals," as he put it. And he's a fairly natural person to have given this talk - he actually sold Eutechnyx to an American oil magnate at one point, and it took him six years to buy it back again. The opening part of the talk came across as a bit of a sales pitch for Eutechnyx - they've won awards, shipped million-unit titles, yadda yadda... I suppose it made some degree of sense to explain why Eutechnyx is in the position that people want to buy it.
So, why are publishers looking to buy independent developers? Darren's theory was effectively "because publishers aren't very good at making games anymore" - specifically, they're not good at making them efficiently. They had a load of cash back in the days of the PS2 boom, so they were able to roll onwards without trying too hard to keep costs down - but that cash is starting to run out now that the next generation is here and production costs have gone up. Meanwhile, the independents didn't get fat off the PS2, and have been running an efficient operation while the publishers have been hanging loose. While it used to be that publishers would buy independents to get at their IP, now they're buying them for their production capacity - a ready-made, highly efficient game-making machine. And of course, if they don't buy a particular independent, there's always the risk that their competitors will.
The independents, meanwhile, tend to find the offer of several million pounds to be quite tempting...
There are three independents that produce driving games (Eutechnyx's genre) that have been bought up in recent years: Evolution Studios, Bizarre Creations, and Sumo Digital. All UK companies are required to annually file their accounts with Companies House where they can be publically inspected; Bizarre Creations and Sumo Digital both filed abbreviated accounts containing a minimum of information, but Evolution Studios filed full accounts - which Darren would retrieve from Companies House and analyze to try and work out what they were up to. (At the time, I found this pretty funny, but it's good business sense).
Evolution are the people behind the World Rally Championship series - pushing out a new game in the franchise every year from 2001 to 2005. In 2005 they themselves bought Big Big Studios; and then in 2007 Sony selected them to do Motorstorm for the PS3. This was the beginning of a relationship that would see Evolution go PS3-exclusive.
Darren showed us some graphs, pointing out interesting artifacts. Wages and staff numbers were on the rise consistently from 2001 to 2005, but between 2005 and 2006 they stayed level - because Evolution were having trouble finding good new people to employ, Darren speculated. The "cash in bank" graph also showed fairly steady rising until a sharp drop in 2006, but it turns out that that's because the company voluntarily paid out dividends to its shareholders. So, things were looking good. Why did the sale happen, then? Darren listed some possible factors:
- The perceived success of Motorstorm.
- Pressure from the shareholders - though given that they'd just been paid a nice fat dividend and the company's performance was apparently on the up, this was unlikely.
- A standard part of exclusivity deals is the option for the platform holder to buy the developer after X titles or X amount of time, so maybe Sony took that option.
- Perhaps the developer was sick of making the same game over and over, and selling it and starting over would have been an opportunity to get away from that. However, there's really nothing to indicate that that's the case - there was nothing really stopping Evolution from diversifying a bit, and the yearly iterations of World Rally Championship were more than just re-skins.
- Cashflow. Again, probably not the case - things looked fine for Evolution.
- "The battle." By this, Darren simply meant the day-to-day grind of owning and running your own studio - there are an awful lot of things you have to do that detract from the time you can spend actually making your games, and becoming part of a big company where someone else can worry about accounting and legal etc. can be quite an attractive concept. Darren suspected this was probably the greatest factor in the sale.
- The cash in the purchase, as noted earlier, is always tempting.
Was it a good deal, and for whom? It certainly looked like Evolution won from it. The sale price was twice their turnover (most sales are closer to once) and fifteen times their profit (while most are more like five times). There have been worse deals. Though Darren thinks it might start becoming more standard, and that if Evolution had held on longer they might even have gotten more - content creators, and independent developers, are due for an explosion in value.
What are the upsides of a sale, then - why would you want to do it? Number one on the list: having lots of money. Well, indeed. Then there's the idea that your working capital is somewhat secure, being part of a larger whole that can help smooth out bumps; the "big company benefits" like share options and company cars; and, of course, the end of "the battle" - you can focus on development without all the stress of peripheral things. Someone else is worrying about them now.
Of course, there are downsides too. Number one on the list: having lots of money. Erm... Darren described the nightmare scenario - "you've got all this money, so you've got to buy a new house, and move, and you can't bring your friends with you, and you've got to meet all the new neighbours, etc..." I'm not entirely sure about his original assertion that you are simply compelled to buy a new house - perhaps that much money induces impulse property buying, I don't know - but anyway. It's likely to be the case that the structure of the indie and the buyer will be incompatible, so the indie will be forced to change, quite possibly for the worse. You've now got big company politics to deal with, less control over development and more time spent reporting on what's happening to your higher-ups. It's also harder to get passionate about a big company - people develop a "do your job" mentality, it becomes just another job. People who previously were entrepreneurs are now mere employees, which is not good for their motivation.
And, of course, there's the risk that the people you're selling to turn out to be total idiots. When Darren sold Eutechnyx originally, it was to a man named Jack Irons, a Texan oil millionaire who fancied owning a games company. To run the company, Jack brought in Bill Graham, an ex-cop from Detroit. Neither of them knew anything about running a games company, and it showed. Darren was losing $100,000 a month in shares thanks to the pair of them.
Having been through that experience, it's no surprise that Darren's view is that independence is priceless. Entrepreneurs are simply incompatible with big business. There are some amazing opportunities for content creators on the way, that will increase their value significantly, so you should hang on if you're able to.
What amazing opportunities are these? There are a lot of formats to play with, a shortage of skilled developers, quite a lot of money available, and a very high demand for original IP. Online and casual gaming is still showing a lot of opportunities. And Darren specifically called out Korea as a place developers should be keeping an eye on - 75% of the population have broadband, and the avatar economy (paying real cash for virtual items) is very popular there.
Of course, there are some hazards as well, at least for UK developers. The big one is China. Salaries are low there (and we're not talking slave labor, the cost of living is simply lower). India is likely to be a similar threat. The weak dollar means that for US financiers, UK development is now 70% more expensive than it was a few years ago; and in Canada the government is paying 37.5% of all developer's salaries. There's still a shortage of experienced staff. Not to mention that there are signs of a recession in the UK's future.
Darren ended his talk with a demo of Ferrari Challenge, Eutechnyx's newest title. It's quite pretty. They managed to persuade Ferrari to allow the modeling of damage to the cars, something they've never allowed before. He also showed off some of the graphical special effects, calling out their rain system as an example - rain will form puddles of varying depth on the track, and raindrops create ripples on the surface of those puddles. Apparently, "that's what next-gen is all about" - a comment I was somewhat taken aback by - but for a company that specializes in creating realistic simulations I suppose it makes sense for them.
"Long papers" session
After taking a break for lunch (and a pretty decent one at that, by conference standards) and to chat to a LJMU student called Dave, the next sessions started. Instead of going to the main hall where the morning's talks had taken place I went to the smaller "Shanghai Syndicate Room" for the long papers session. This was supposed to be six 20-minute presentations, back-to-back, where the speakers present their papers (the full text of which was in the proceedings, as you'd expect). There's only so much you can do with that format; some of the papers really didn't lend themselves well to an oral presentation, some of the speakers were definitely not comfortable with speaking to an audience, and one of the speakers - the person giving the talk on shadow generation - wasn't even present. Still, there were some good questions and a couple of video (or live) demos that kept me awake.
This was a good session. George Kelly presented the Citygen system, a tech for very fast procedural generation of cities right down to the individual-building level.
Citygen city construction begins with a "primary" road network - these are the main roads in the city, and are constructed by an artist who places the endpoints of the roads. The system then generates a road that follows the terrain. The primary road network really determines the overall structure of the city, dividing it up into "neighborhoods."
Once the primary network is established, each neighborhood is then filled with secondary roads according to a number of parameters that affect things like the average length of roads and the scale between a neighborhood filled with independent trees of roads versus one filled with a highly connected set. These parameters can be used to create distinct characters for each neighborhood, ranging from "down-town" (a classic American grid structure) through to "suburbia" (irregular, winding trees of road with decent amounts of green space left over).
Finally, the spaces between the secondary roads are split up into "lots" of varying shapes and sizes, and each lot then becomes the basis for a single building.
There was a lot more detail in the paper than I've described here, but I suspect it'll be online before long, so you can check out the details when that happens. In the meantime, you can download the sources directly from Subversion and have a play with them yourself at http://www.citygen.net/. The software includes support for exporting generated cities to COLLADA, so it might come in very handy for anyone in need of a quick city...
This talk was good to the extent that it actually ended up winning the "best paper" prize, though like many of the talks it was a "games and education" focused topic which wasn't of so much interest to me. The researchers at the Universidad Complutense de Madrid worked with Harvard Medical School and Massachusetts General Hospital to develop a games platform that integrated with the hospital's e-learning platform. To prove the viability of the platform, they also developed a game on top of it, one intended to teach a complex medical protocol (one involving 98 separate steps) to medical students. To make things more difficult, the cost of creating games for the platform had to be close to the cost of creating existing learning resources - which are mostly just HTML text files. The resulting game was a Myst-like point-and-clicker - the most effective use of their shoestring development budget, as most of the content could be produced simply by taking photos of hospital rooms and equipment - which fully integrated with the e-learning system, recording scores, grades, etc. Apparently the system hasn't yet been fully deployed as they have to resolve some political and legal issues with the hospital (e.g. trademarks visible on equipment within the game), but pilot schemes have proven effective and they're confident that they'll be able to move forward soon.
This was another talk aimed at the games-in-education crowd. To aid in design of authoring tools for "serious games" used for teaching, Stephen Tang presented the results of his investigation into the common elements and components such games require. Past research projects in the area have either been feasibility studies or design guidelines, while this was more of an attempt to extract some preliminary design patterns.
So, there are three elements of "pedagogy" within a game, three kinds of things that people learn: The properties and behaviors of objects within the game (e.g. this is a barrel, it is heavy, it is explosive), the relationships between objects within the game (e.g. when my bullets impact the barrel, it explodes), and the solving of problems in the scenario defined (e.g. I could take out that gun turret by shooting the explosive barrel next to it). So far, so reasonable.
Tang then went on to present the four components that a serious game breaks down into:
- Screens: informing learners of what they are supposed to learn
- Cut-scenes: showing learners what they are trying to achieve. More effective than screens as they catch the learner's attention better and help stimulate recall.
- Tutorials: An interactive but guided learning sequence
- Levels: An interactive, unguided learning sequence, with full freedom to experiment.
I got almost nothing from this talk, mainly because I could barely hear what the speaker was saying (and I was sat in the second row). The only clear concept I was able to discern was that content management systems like Documentum are quite a good idea. Beyond that, it seemed that all they'd done was just taken the "Racing Game" starter kit for XNA and modify it a bit.
This talk suffered from something of a failure to commit to a direction. I went in expecting something like what we got for the first section: the notion of a user creating content in one virtual world (like Second Life) and then using it in another (like World of Warcraft, I guess?). They observe that content created for one engine is usually not compatible with content created for another engine (ok, sure) so they propose that engines communicate with a "Service Utilization Framework" that translates objects (i.e. their behaviours, appearances, etc) into a format that the specific engine can understand. So far, so importer. I asked whether they'd explored things like the ability to restrict the behaviors of an object - no creating super-weapons in Second Life and sharing them with WoW - and they said they hadn't explored that yet.
After that start, though, they seemed to veer off on a bit of a tangent. Talk turned to the mapping of virtual objects to real-world ones, with a demo of a virtual phone connected to a real one; when the real one rang, the virtual one did, and when he answered the virtual one, the real one was answered as well. Kind of cool, but I'm not sure how it was relevant to the title of the paper.
We took a short break for coffee. Due to an error in the programme, I originally thought I could choose between a couple of big lectures or a second set of long papers, but on closer examination I realized the lectures went on an hour later than their section heading suggested. Oh well - the talk about machine learning sounded like it might be interesting, but I decided to head back to the Shanghai Suite for the paper presentations instead.
This was another reasonably good presentation, one of the education talks that produced something reasonably concrete. Hanno Hildmann presented an approach to teaching children "positive moral values" (i.e. good citizenship), using an example game called Utility Tycoon. Of particular note was his observation that encoding the desired learning targets as implicit parts of the game itself would be much more successful that simply preaching to children; rewarding them for making the "right" choices by granting them bonus points, extra game content, etc, would cause them to explore and learn those choices for the purposes of getting the most out of the game; then, ideally, they take that learning out of the game context and apply it to the wider world. I don't think this is a particularly new idea, but having examined some of what "edutainment" had to offer a few years ago, it's nice to hear that at least somebody finally understands how to do it properly.
Hanno did however emphasize that his approach was so far unimplemented and untested. He's looking into implementations on mobile platforms though.
Mark Nicholas Grimshaw (or "Sir Fragalot", as he seems to be known online) is a wonderfully clear speaker, but unfortunately my ability to absorb this talk was impaired by a lack of familiarity with the background material - Mark got right in there with terms like "acoustic ecology" and "paraspace" that left me a little behind. I seem to get that a lot with sound talks...
Mark's first point was that while graphics is constrained to a flat 2D plane, using tricks to create an illusion of 3D, sound is truly 3D. It's the only part of a game that isn't being "simulated" in terms of player interaction - there's no tactile feedback, no smell, no taste, sight is being fooled with a two-dimensional illusion, but what we hear is always actual sound. That's worth considering in terms of how it can contribute to immersion.
He defined the term "resonating space;" it's hard to accurately summarize the concept here without just quoting verbatim from his paper, but roughly speaking, it's a locale in which sound happens, and all the properties and data affecting the way that it happens - anything that sound interacts with (i.e. walls, floors, etc - anything sound can bounce off, be absorbed by, or generally collide with). Mark also emphasized that these spaces are not constant, but can change over time, because geometry in FPSes often changes (moving platforms, etc). A resonating space may not have exact dimensions, but there is still some notion of a "big" space versus a "small" space - cathedral versus air vent. Working to help the player perceive their role (i.e. their position) within the space immerses them and invites them to start interacting with it (i.e. making noise). The two most effective factors in helping the player to do this are localization (conveying where sounds are coming from) and reverberation (conveying how sound is bouncing off surfaces); these anchor the player in 3D and set up the boundaries of the environs respectively.
Both effects face a significant obstacle in that the configuration of the player's speakers are beyond our control. Mark didn't propose a solution to this, but he described the ways in which this can cause problems quite fully. He also noted that, due to the near-infinite possible ways in which sound can be located or reverberate in games, modern games use "acoustic shaders" (i.e. real-time signal processing) to achieve them instead of trying to pre-record a lot of differently-adjusted sounds. Well, indeed.
The rest of my notes are somewhat disjointed... I learnt the word "acousmatized," which means "sound with no visible source" (i.e. offscreen), and I noted that sounds which are not associated with real-world objects get interpreted by the player through their past knowledge of the game, or, failing that, cultural knowledge - if you have an alien sound that sounds a bit like a dog, the player will think of that sound as "the sound that sounds a bit like a dog." Mark also said that nondiagetic sounds (sounds that are not part of the game world) tend to be rarely used in FPS games, which I guess is true.
I briefly asked Mark where music fits into all this: he plays with music turned off because he feels that it breaks immersion. Hmm...
This talk was about "social networks" in the sense of societies of characters within games, rather than things like MySpace or Facebook. They'd performed a study of a bunch of games - Oblivion, Ico, UFO Afterlight, Crusader Kings, Civilization VI, Canis Canem Edit, Splinter Cell: Double Agent, and Fa?ade - and observed 31 distinct patterns in societal interaction between NPCs or between NPCs and the player, such as Loyalty ("The continued goal of being part of a social network") or False Accusations ("Untrue statements that can be made to affect the social network"). Aside from a slightly curious use of the word 'Fraction' instead of 'Faction' throughout their paper, it's worth a look for anyone who wants ideas about how their NPCs could choose to interact.
I think by this point I'd started to get fed up with sitting through sessions, so I may have been overly harsh to Colin Dodd who presented the simple game he'd made with XNA when I started heckling him. "One of the downsides of XNA is that it only supports C#..." "Er, sorry, excuse me, that's not really accurate..."
Anyway, his game looked quite nice and straightforward; also, because XNA 1.0 has not included any networking or matchmaking functionality, he implemented networking using System.Net and direct exchange of IP addresses instead of a matchmaking server. To make the IP address exchange less error-prone, he came up with quite a nice solution that represented each IP address as a sequence of four words, where each word maps to a single octet of the address. Made exchanging IP addresses much simpler for players, though I didn't see what the UI was like for entering the IP address back into the game.
A fairly solid AI talk. "Topographic Products of Experts" is a kind of mapping, similar to Kohonen's Self Organising Map, that preserves topology - points which are mapped close together share some common feature while points which are far apart do not. Benoit Chaperot explained how they'd implemented the technique in a motocross game, along with a multi-layer perceptron and a couple of Kohonen maps, to compare the results.
Well, it wins some and it loses some. The time taken to train the ToPoE was by far the worst of the bunch, with the larger ToPoE clocking in at almost an hour of training compared to the others 10 seconds or less. Decision times (i.e. the time taken to process the AI each frame) were also not great, taking at least twice as long as the MLP. On lap time, it did manage to beat the Kohonen map by a couple of seconds - apparently this is because it allows for the interpolation of data, rather than the Kohonen approach of finding the best fit, giving more accurate results - but it still couldn't beat the MLP.
The rest of the talk was about the issues they'd encountered while porting the game from Visual Studio 6 to Visual Studio 2005, and the story of a memory leak they'd tracked down.
Stephen Tang presented this paper, which seeks to "describe" games for learning and define some terminology. Unfortunately, things got off to a very bad start with an attempted description of "Games for Learning" versus "Digital Games" - something which the audience had to get him to clarify was supposed to mean "Games for Learning" versus "Games for Entertainment" - not a great start for a paper that seeks to clarify terminology. Anyway, they compared the former with the latter in three categories - rules, play, and culture - and illustrated the different approaches the two areas take to those things, e.g. rules in games for entertainment are focused on playability, while rules in games for learning are focused on supporting the activity of learning.
He gave definitions for some of the terms used to refer to the field:
- Edutainment - educating children through entertainment media like television.
- Educational games - a subset of edutainment - often referred to as "games for kids." There was the somewhat cynical observation that kids have lower expectations of games, so production values can be lower, while older players have experienced better games. I'm not sure how long that'll hold up, but oh well.
- Digital game-based learning - e-learning, but promoting experiential learning. Uh, ok. (They don't give any definition for e-learning, sorry).
- Simulators - highly precise models of reality sometimes used for training and teaching.
- Serious Games - a term coined by the Serious Games Initiative that extends educational games beyond training and "classroom" education to include things like health and public policy - for example, the UN's Food Force game.
I wasn't invited to the gala dinner, so I went out with a couple of friends from Juice Games for some drinks and a curry. Oh well. As I said, we'll all just have to imagine what that dinner was like.
Ernest opened the day with his keynote, "Fundamental Principles of Game Design." Yeah, fundamental. He warned us at the beginning of the talk that it would be a "remedial" level - pretty basic stuff for anyone who's already got experience - and that it was really there to serve as the opening part of the game design workshop he'd be running later in the day.
I'm not going to go into detail here, as Ernest shot through pretty much every aspect of game design in overview mode. This is the same talk he gives at the beginning of his Fundamental Principles Workshop, so he doesn't give the slides away.
Speaking of the Workshop, it was apparently due to start after this first talk, and being a hands-on kind of person I inquired as to how I could attend it - only to be told that I wasn't registered for it and so couldn't attend. Looking at the crowd of people who left the main conference room and headed for the workshop room, it looks like the only people who were registered for it were the Liverpool JMU students. I don't see anything about registering for it on their website, unless that's what they meant by "Tutorial."
Somewhat disappointed, I returned to the main conference room - now much emptier without the presence of the JMU students - and sat down for the next talk.
This was quite a cool talk, full of information, but somewhat disjointed - Isaac, an eleven-year veteran currently at Sierra Online, talked about things ranging from brain chemistry to the requirements for play, and as far as I can see there was no overarching structure holding it all together. The title of the talk was "Emotional Artifacts: A Game Designer's Model;" he eventually defined an "emotional artifact" as "a product of human craft used to elicit emotional response" - games, painting, poetry, music, sculpture, etc.
Other topics he covered included the idea of "affordances" - "characteristics or functions offered by objects," as in, an affordance is one possible thing you could do with an object. What an object actually affords and what we perceive it to afford are of course different things, but it's when affordances are unclear (for example, a pull-handle labeled "push") that you are seeing design failures. He also covered the idea of a model as "a thing that represents something else," and discussed how people create mental models of the world and things within it - the same goes for your game world, and the model your player creates in their head might not match the one your designer's got.
There was a lot of stuff here, but unfortunately the notes I ended up taking were so disjoint that it's near impossible for me to construct prose from them. Words like "Wittgenstein" and "surplus resource theory" lie around like debris from a head-on collision between two trucks carrying whitepapers.
Private Virtual Worlds: Getting Real About Being Virtual
This talk didn't say very much but took quite a long time to say it. On the one hand, I dislike this, as it meant I was very bored. On the other hand, it's great because it means my writeup is very simple.
Some organizations have their own private 'virtual worlds' (a la Second Life) for collaboration and... stuff. The emergency services out in the US somewhere used one for modeling a major road accident that had happened on a busy road, and planned out their response. One company is using a virtual world for conducting meetings. There was really nothing here beyond "some people have extended videoconferencing/email/IM/etc into full 3D simulation environments." Just like any other corporate communications medium, things like security are a concern. Break for lunch! Chew thoughtfully while watching cartoons on a screen behind me and wonder whether to try introducing myself to Ernest Adams. Eventually decide not to, partially due to shyness and partially because cartoon has become epically gripping. Upon returning to the conference room, we were told that of the three presenters who'd appeared on the schedule - Alice Taylor from Channel 4 Television, Thierry Platon from BIP Media, and David O'Reilly from Travellers Tales - only Alice Taylor had actually turned up; the other two had cancelled. Thierry was going to be talking about the trend towards casual video games (his talk was to be titled "Are all video games going to become 'casual'?") and David would have been talking about his path into the Industry.
Break for lunch! Chew thoughtfully while watching cartoons on a screen behind me and wonder whether to try introducing myself to Ernest Adams. Eventually decide not to, partially due to shyness and partially because cartoon has become epically gripping.
Upon returning to the conference room, we were told that of the three presenters who'd appeared on the schedule - Alice Taylor from Channel 4 Television, Thierry Platon from BIP Media, and David O'Reilly from Travellers Tales - only Alice Taylor had actually turned up; the other two had cancelled. Thierry was going to be talking about the trend towards casual video games (his talk was to be titled "Are all video games going to become 'casual'?") and David would have been talking about his path into the Industry.
Games for learning, games for innovation, and games as public service technology
Despite the long, lofty title, this was a fairly straightforward talk. Alice Taylor is the commissioning editor for education at Channel 4, one of the major British television stations; her job, it seems, is to be in charge of paying people to make educational content, both for the television station and also for things like C4's website. She's the kind of person you might pitch to, and, despite her apparent "outside-the-industry" situation, is very definitely in touch with things.
"Education," in this context, Alice defined as "teaching you something you'd find useful outside of school." That's fairly broad. As far as games are concerned, she quoted Raph Koster: "fun is learning" - not everyone agrees with that, I know - but accordingly, she's more interested in games that teach almost accidentally instead of the stale edutainment approach of "generic game in educational setting." As an example, she pointed at Age of Empires - kids who had played it were more likely to know what a trebuchet was (a medieval catapult) than kids who hadn't. The game doesn't go out of its way to teach you what a trebuchet is, it simply includes it as something you can build and use in battle, and through playing with it you learn what it is and how it can be used. This approach has certain parallels to the Montessori method of teaching, which favors hands-on experiential learning over abstracted concepts that are harder to grasp.
All games can teach things in this way. But if that's true, Alice suggests, then one can see how it is that they're held responsible for it - though more than they should be, in the case of people like Rockstar, who suffer because parents and retailers aren't taking as much responsibility as they should be.
So, what exactly is C4 looking for? They're commissioning content for 14 to 19 year olds, and from January they're looking to really focus on platforms beyond just TV. As far as games are concerned, Alice had five words - "Playful, social, useful, connected, and fun" - a couple of things to show us, i'minlikewithyou (which, after toying with for ten minutes, I still don't really understand) and This Spartan Life - and a couple of people to look into, Amy Jo Kim and Danah Boyd.
Middleware in Game Development: Homura IDE
Because the other speakers hadn't shown up, and Alice could only talk for so long, the conference organisers decided to throw this session impromptu. Unfortunately, the only notes I appear to have taken from it is "GET ME OUT OF HERE" so clearly I wasn't enjoying it. From what I remember the session was about 50% showing off a plug-in someone had written for Eclipse for making games in Java (I think it's here, if anyone actually cares), and 50% telling us what middleware is.
We took a brief break for coffee, and then the people who'd been at the design workshop rejoined us for the final section of the conference.
Skills issues for the Games Industry
Tony Hughes, from Liverpool JMU's International Center for Digital Content, presented their findings on the current problems around the games industry's workforce. Firstly, the industry's not diverse enough - we need more women and ethnic minorities (though there was no mention as to why). Secondly, the skillset most in demand is "design" - I'm not sure exactly what that consists of - and it's particularly requested by the independent developers.
Lastly, he presented some of the issues that studios have reported with fresh graduates from the games courses - not good feedback, summarised with "not appropriate to what the industry actually needs." Graduates are lacking in teamwork and communication skills, they can't take constructive criticism, and they've got no concept of budgets or schedules. I'm not sure how much of this is inevitable with graduates - a lack of ability to take constructive criticism, for example, I suspect is simply down to lack of experience rather than a failure of education; while a lack of understanding of budgets and schedules is perhaps a more clear indicator of something missing from course content.
The final session of the conference, this was a simple question-and-answer panel moderated by Jon Weatherall and featuring Ernest Adams, Dave Burrows (SCEE), Isaac Barry (Sierra Online), Neil Jones (Rare), and Tony Hughes (ICDC). Unfortunately, there was no microphone being passed around the audience so I didn't catch all of the questions, but here's some of what was asked.
Q. If there was one skill that you could create infinitely for free, which would it be?
EA: Leadership and project management. The vast majority of games that fail, fail because the team can't work efficiently, or is confused, or just gets itself into a mess for one reason or another.
DB: Low-level coding - not so much understanding assembly language, but understanding the workings of the computer (especially the cache).
IB: I second Ernest's suggestion. Plus communications skills (particularly empathy).
NJ: I agree with Dave about the low-level understanding; also, many people don't have good math skills.
TH: I'm an analyst rather than an industry guy so I can't really comment, but I will say that to improve communication and math skills requires a lot of work, right down to school age.
EA: Students nowadays start with C++/Java - they don't get the low-level skills.
Q. Art has gotten a lot more complex; the skills involved have 'accelerated.' Have you had to change the way you work much?
EA: I've not changed the way I do anything in 10 years.
DB: Artists used to be technical; then, not so much; and now, they are having to be taught shaders. So it's come full circle, but there's still room for pure artists.
IB: Designers coming out of the academic system have more of an abstract understanding of what they do. I'm constantly learning and trying to acquire new skills that are relevant to making games (or cancelling projects).
NJ: I've only been in the industry three and a half years so nothing has changed for me. The only thing that's really changed is the amount of art required.
Q. What do you think of game degrees?
DB: You need to be careful that you have something more in there to differentiate you - five people all applying to the same company, submitting identical final-year projects, does not make a good impression.
IB: The industry doesn't have the highest opinion of Digipen graduates, but that's changing. I did hear a story about a designer who had students bringing in magazines with him in so he could talk about how he was famous...
NJ: Lack of specialization is a problem.
Q. Industry and university shouldn't mix - it puts too much pressure on skills acquisition rather than on thinking. Thoughts?
JW: Universities won't let that happen.
DB: Thinking is part of the skill base.
A very welcome contrast to a day of sitting and listening, the GDTW organisers set up a joint party with the IGDA Northwest chapter - the kind of party where they take over a pub and put somebody's credit card behind the bar. A great way to end the conference. I got through a number of ciders, played some pool, and spent quite a long time chatting to assorted people.
Kaseem and Richard, IGDA Northwest committee members, and doormen for the night
Rocking out with Guitar Hero
Ross, a LJMU student, shows me his Marathon tattoo.
It's always tempting to walk away from an event allowing the final part of it to leave the biggest impression, but looking back over the whole two days, my opinion of GDTW as a whole - at least for the 2007 conference - is not high.
I'm left with a distinct feeling that there's very little here for people who are actually doing professional development right now. Much of the content seems to be targeted at Liverpool JMU's own students - certainly they must have made up at least 75% of the attendees, and Ernest Adams' workshop on the second day appeared to be laid on exclusively for them; a good chunk of what's left is too buried in academia to be useful, full of things that are too niche or too unrealistic to be applicable, and anything that is worth paying attention to already appeared at GDC or another similar conference. It's pretty telling that for the first day, even a load of the students didn't turn up; someone told me that apparently Wednesday is their free day, so they didn't want to spend it at the conference, only turning up on Thursday when they'd otherwise have to be working anyway.
Maybe there's a bleeding edge in there for somebody, somewhere, but you'll have to sift through an awful lot of pointless waffling to find it. I don't need to go to a conference to sit in a room, have someone tell me I should have a system for tracking where my textures came from, and then show me an XNA starter kit. Now maybe this really is a problem with the state of games academia - maybe the papers presented there are the best that the universities of the world have to offer. If so, it's very depressing. However, I doubt that it's the case - and one thing I'm suspicious of is that the papers were selected with an audience of academics in mind. I can easily imagine that what was covered will be more valuable to other people writing papers than anyone else. Looking at the website for the conference, I found the technical programme committee - I assume the people responsible for reviewing and selecting papers - and there are exactly two names, on a list of 46, that aren't from a university (and of those two names, one is from the ACM and the other is from the British Broadcasting Corporation's R&D department). With a selection committee composed so entirely of academics, it's not surprising that the result is a set of papers that appeal strongly to academics.
Networking opportunities, also, are practically non-existent; it's unlikely that you'll bump into somebody you know or want to know unless they're giving a session or are going to JMU. (Granted, this isn't something they can easily fix - they can't exactly hire people for you to network with - but still, it needs to be said). The IGDA party was the high point from this angle.
To be fair, there's only so much that they set out to do; rather than evaluating them purely on my own criteria, I should probably hold them up against their own standards as well. The goals listed on their website are:
- To provide a forum to discuss state of the art games design and current and future games technology with the specialists. The presented research may have been "state of the art," but in many cases, it wasn't my art. "State of the somewhat related art," maybe. And of course, discussing things with specialists requires that those specialists attend the conference, which - beyond the people presenting the sessions themselves - didn't seem to have happened very much.
- To enable academics and researchers in computer game technology and computer entertainment to present their work during the research sessions, and to seek opportunities for collaboration with the industry and for academic research partners. Again, this mostly happened aside from many of the presentations only having loose links to either "computer game technology" or "computer entertainment," and collaboration with the industry will again have been difficult with an industry that wasn't present.
- To facilitate relationships and collaboration between academics promoting Computer Games Technology courses, the UK games industry and supporting organisations. Er, ok. I'm not an academic promoting Computer Games Technology courses (might even go so far as to say I disapprove of them), nor was I asked to be involved in facilitating relationships and/or collaboration with an academic promoting Computer Games Technology courses, so I have no idea whether this happened. If all it means is "a chance for LJMU staff to shake some industry hands," then yeah, it happened.
- To provide games companies and related industries the opportunity to present their companies, products, job opportunities, and to discuss and discover alternative technology for game development. I can see how companies like Craft Animations were able to take advantage of this.
- To allow students studying Computer Games Technology to make early contact with key players in the industry, enabling them to learn first hand about games development, and the challenges of working with and for a fast moving technology. Not much in the way of "key players" in attendance. Also, stop me if I'm wrong, but wouldn't learning "first hand" about games development mean... actually doing it? Learning by being told about it by a developer is learning second hand. Maybe I'm just being picky.
Of course, it doesn't have to be like that. There are two directions GDTW could go in for the future (and no, you smart-alecs, those directions are not "down" and "out").
Its first option is to try to become a full-blown developer's conference, and live up to the image that its name conjures. Pick sessions that are more accessible - and more useful - to people who will have the opportunity to go back to work the day after the conference and start applying what's been presented. Cut the JMU students loose a bit so that they don't overwhelm the attendance record. To be clear, I don't think this is an option they should take - they're not the best people to do it, and besides, GDCE shut down for a reason. We've got things like the Lyon GDC and the Leipzig GCDC to attend already.
The alternative is to put even more emphasis on the presentations from academia, get people from industry onto the selection committee to ensure things stay relevant, and end up hosting a conference that is explicitly about academia presenting papers to the industry. An opportunity for academics to present their f