What's Killing Game Development? July 09, 2007 Jakub Wojnarowicz

Introduction

Page:: ( 1 / 6 )

It's intrinsic to human nature, but the idea of focusing on a single issue or event, or at least a single cause, is rather limiting and usually results in an incorrect analysis of any given situation. Take for example the question “why are there so many game sequels?” There are a half dozen answers to it and they're all correct, but people are apt to focus on one or two and dismiss the rest. Moreover, the very question itself is limiting – we're only asking about game sequels. There are other, equally important questions that are related to the original – why do so many games fail? Why do publishers have increasingly more control of development? Why isn't there more innovation in games? What happened to the great designers like Sid Meier, where is the new generation? Believe it or not, all these questions are related and are best asked with “why is modern game development the way it is?”

Ironically enough, there is actually a single issue that lies at the core of these questions, so pardon me while I paint myself a hypocrite and point it out – game development is a business and people are in it to make money. Once we accept this axiom, all answers will flow from it. Let's deal with the questions one by one.

Where are the great designers?

With the exception of Will Wright, few of the any old, great designers have done much lately. Sid Meier's name is on a bunch of games, but it's not actually Sid in charge of development or design. Chris Roberts, of Wing Commander fame? Let's just say he had a conflict of interest with the Wing Commander movie and his projects at Digital Anvil, and that Microsoft's sudden, detail-free take-over of Digital Anvil has quite a few details if you scratch beneath the surface. Lord British? He's heading a studio more than dealing with development itself. Warren Spector? I can't speak with certainty, but it's likely he passed on the reigns of Deus Ex 2 to Harvey Smith for a reason. And finally, there's John Romero, who in his own way inspired this article.

One answer to what happened to these big-name developers is that they simply moved on. That they did their time in development and then took higher-paying jobs further up the job ladder. Yet this doesn't quite answer why there aren't any big names now. Where are all the great ideas coming from? Who were the minds behind Allegiance, Sacrifice, World of WarCraft, Company of Heroes?

Which brings us to John Romero. Before it was fashionable to rip on the long-haired one, he convinced Eidos to fund ION Storm, which was to be the sexiest, most promising game studio ever. At some point between his departure from id and the jokes about Daikatana, John coined his motto: “Design is Law”. Even after it became a joke, the phrase does have some appeal. After all, doesn't it make sense? A great designer who has incredible ideas and legitimate ways to implement them organizes a team to make it happen. He says “this is how it should work”, and clearly communicates the ultimate vision to his subordinates, who, corrected from their erroneous interpretations, diligently get to work by making the nitty-gritty details fit together to make the Design a Reality.

This phenomenon isn't unique to game development. Sexy, simple, flattering ideas are all over the place, throughout history. In the 19th and early 20th centuries it was popular to think that it was the courage and ingenuity of the individual soldiers, combined with brilliant leadership, which lead the Great Powers to victory over one another. The idea was popular because the soldiers would naturally reflect the ingrained character of the nations at war. Sports teams, especially before free agency, also rely on this – the city feels good because its teams are good. Similarly, people like the idea of the great and noble game designer because everyone has a game idea and everyone imagines themselves to be that designer – not the peon slaving away in the trenches.

The truth is, however, that design isn't especially difficult and great ideas aren't hard to come by – not in the least. Everyone has ideas and most people with a college education (or capable thereof) can put them down on paper in an organized and clear fashion. Those with some experience in development, especially if they've seen the capabilities and limitations of a team, can usually make that document a good foundation to build from.

No, the difficulty is not in design in the least. Nor is it in programming – look hard enough and you can find programmers for all jobs. Art, sound, marketing, and PR staff are also readily attainable. Where real talent is necessary is in organizing all this staff. A typical team for an A-list game exceeds 100 members for most of its development time. The difficulty in keeping everyone motivated, focused, and working towards their assigned tasks is immense. This includes overcoming the occasional personality conflicts, drama, frustrations over progress, making sure people don't get bogged down on pet projects, and selecting the right people to manage specific departments. No employee is perfect and they all need to be motivated and managed in individual ways. It is the project manager and his immediate subordinates who involve the team in developing workable ideas and innovations, while excluding the ones that would distract from the game or are too difficult to tackle. It's about being able to keep milestones and satisfying the publisher, the marketing team, and the press. The task also involves that very difficult job of planning out the various parts of the game and bringing them all together as economically as possible – so that the engine (including network code), sounds, graphics, level design, story design, and AI can be worked on simultaneously and tested as early as possible for that ever-elusive “fun” factor. That's a lot of skills to ask of a single person, or a small group.

The great designer can still exist, but he also has to be the great manager. It's not enough to have a cool vision and get 10 or 20 people working together – that's within most people's reach. Where the early great designers set themselves apart is in having original ideas, being able to manage the teams of the day, and having the technical know-how to understand what was capable and what you could force computers to do. However, having all that knowledge and being able to keep 100 people working for two or three years on the same project, through delays, stress about milestones, technology changes, and revelations from the competition? There's the rub. Ultimately, the project management skills are more important than design skills, because they're harder to find. ION Storm's ultimate fate is a perfect example of that. It's not that John Romero was a bad designer, but as evidenced by the numerous delays and drama surrounding his team, he seems to have been a bad manager. Had Daikatana launched on time with its original feature set, it would likely have fared well against its immediate competition. Half-Life and the revolution of the FPS genre was still some time away.

More questions

Page:: ( 2 / 6 )

Why do so many games fail?

Games fail because the teams making them don't have the talent, the finances, the vision, or the marketing to succeed. Buggy games are often the result of either a rushed project (lack of money), or delays (poor management). Boring games are a consequence of a poor vision and/or an inability (for financial reasons) to take the time and perfect the little things that take a game to the next level – a better combat system, or more meaningful dialog, or crisper weapons control and movement. Even if a game is innovative, interesting, and bug-free, a sub-par marketing campaign can sink it. I can think of no better example than Interplay's Sacrifice. Whether through financial shortage or simply an incompetent marketing department, Interplay was unable to get the public excited about this excellent game. Of course, the public is also only willing to go so far to receive strange games, no matter how good they may be (ie, Giants: Citizen Kabuto, Sacrifice).

The rest of the questions

Before we address the rest of the issues affecting game development, let's look at some numbers first. These figures are what we came up with based on talking privately with several developers stretching back to last E3. Everyone has some variation, so we drew an average on the lower end of the scale from most of the numbers thrown our way. All sources have chosen to remain anonymous.

Let's assume for a second that you already have a game engine – whether you're developing a sequel to one of your own products or have licensed.

In the first year, your team will average about 20 people. Here you get your core programmers and designers learning what you can do with it, one or two artists of the various types (concept, texture, modelers, animators) to put the engine through its paces, and a rough framework evolves. Ideas are thrown around and the design begins to crystallize into a somewhat more realistic version of what was proposed in the design document.

Since these are your best people, they'll likely be earning more than everyone else. Total monthly cost per employee will be about $9000. Now that doesn't mean they get paid $9000, but after including the cost of office space, utilities, benefits, computers, networking, software licenses, and occasionally feeding these guys, that's a good estimate.

All numbers below are either conservative or very conservative figures. We generally took the lower end of answers to avoid the risks of exaggeration.

So, in your first year, your game costs you:

20 * $9000 * 12, for a total estimate of:

$2,160,000

Not bad, right?

Problem is, you have two more years to go, and if you've been paying attention so far in this article, that team is going to get a lot bigger. Up to around or over 100 people.

So for years two and three, let's assume an average of 100 people. It will be a little lower at the start of year two and quite a bit higher at the end of year three (when the almost inevitable crunch time brings in extra hands and QA), but the average cost will go down somewhat, to about $7500 per employee per month.

So, for year two and three, your game costs you:

100 * $7500 * 24, for a total estimate of:

$18,000,000

The grand total development cost is thus $20,160,000. At this time, we're not including engine costs (either development or licensing), or a marketing budget, or the overhead from the company itself (executives, support staff, etc.)

Let's just run with that number for the moment. Now, a game sells for $50 when new, and the publisher doesn't get all that. Hard figures on wholesale costs were hard to find, but we're told $35 is a good guess.

At $35 per copy, a publisher would have to sell just under 600,000 copies of a game to cover development costs. Add a $500,000 engine license plus another $4,000,000 in shaping the engine to suit your needs, and that figure jumps to about 700,000 copies. Then there's the marketing budget - $1,000,000. Add another $3,000,000 in ancillary costs (executives, support staff, smaller licenses, PR budget), and here is your grand total estimate:

$29,040,000

Divide that by $35, and you get almost 830,000 copies sold to break even.

Does that sound easy? Let me put it this way: Crackdown was one of the most exciting and enjoyable games released on the Xbox 360 (the most numerous of the next-gen consoles), it got rave reviews and good word-of-mouth. In February, it sold 427,000 copies. In March, it wasn't even on the top 10 list. Now, it wasn't quite listed as a AAA title (in the vein of WarCraft 3, Gears of War, etc.) for most of its development, and the dev team probably averaged 60-70 people over the last two years. It was announced in October of 2005 and had likely seen a year and a half of development by that point. There were early rumors of delays, which usually means that these aren't the usual expected delays associated with development.

Due to the smaller team, it cost less to develop but also had less support than AAA titles. Most likely Crackdown simply broke even or made a modest profit for the publisher – which makes no guarantees for the developers. The great achievement of Crackdown was to establish a solid new IP with some interest from the audience at large in a meaningful sequel.

More on the horizon

Page:: ( 3 / 6 )

Of course, we hear about games that sell 3,000,000 to 6,000,000 copies over their shelf lives. As we all know, however, that doesn't happen every often. The profits from these games usually simply cover the losses incurred by funding other titles. A publisher might fund 5-10 games to AAA status per year, and maybe 2-4 will actually achieve those sales. These will pay for the costs of the other games that bomb or sell, but not enough to cover their expenses.

So everything is good, right? The industry has a system that rolls along, right? Wrong.

Just over a decade ago, around 1995, a game might cost $300,000-500,000 to make. It was a “hit” if it sold 300,000-500,000 copies, at $25 per. So if you had a hit game, you needed only one dollar per sale to cover your costs.

Nowadays, a game costs $30,000,000 to make. It's a hit if it sells 2,000,000 copies, at $35 per. If you achieve this hit, you take $15 per copy just to cover development expenses.

Also in 1995, a game cost $50 in 1995 dollars at the store. Now, games are still $50 in 2007 dollars (which, if you've been following the dollar's decline against all other currencies, means a lot less money). Margins for publishers and retailers are both thinner by default. A low dollar helps bring in extra income from abroad, but it also makes developing in foreign countries expensive, and of course puts inflationary pressure on the dollar internally, which leads to higher living costs and thus more pressure for increased salaries and higher licensing costs.

Developers and game engines

Every wonder where all those independent developers went to? Why most dev houses are actually owned by publishers (a la Raven, Relic, etc.)? If things are tough for publishers now – which they are, unless you're EA (and to a limited extent Activision), they're horrible for developers.

A developer gets an advance to work on a game. Let's say the team is experienced, has good leadership, a proven track record of finishing games (more on that later), and a killer idea. The problem is that no engine out there for license can do what they want, and they need to make their own. This is going to be a Big Project. We're talking something on the scale of The Sims, Neverwinter Nights, or WarCraft III.

The first thing the team needs to do is develop an engine. Forget what you think you know about engine development. It's not John Carmack or Tim Sweeney sitting down in their office for a year and grinding out a wicked fast renderer with physics, object loading, animations, terrain, hooks for the latest shader effects, networking, content creation tools, sound, speech synching, conversation trees, item values, and so on. There's a whole team working on this, usually for at least 18 months and sometimes in excess of two years. The biggest developers out there have a dedicated engine team keeping an engine up to date for their game development group(s) to work with (who further customize it to their own needs). Eventually, however, every engine reaches its limits.

A new console generation comes out. Or the old engine, built on the “graphics cards are going to have more memory and faster memory and cores and CPUs are going to be faster” paradigm, suddenly doesn't work. You can hack in shader support and better lighting and maybe normal maps, but support for multiple cores and multiple GPU paths is going to take a complete re-start.

Cost? $5,000,000 on the low side, $10,000,000 on the upper end. Think about everything that the Source engine does. Lighting, networking, scripting, physics, a superb animation system, excellent sound, it loads huge levels into relatively limited memory and does this quite quickly, it has support for complex scripted events and state of the art AI. One guy doesn't do that any more. Ten men can't do that any more. You're looking at hiring around 20 excellent programmers with a good work ethic and good communications skills to get all that done, along with the actual development tools.

Add that $10,000,000 to that conservative $20,000,000 estimate for $30,000,000. How many developers have that kind of cash lying around? Pandemic and BioWare thanks to Elevation Partners. Id Software because of its long track record of success. Perhaps Epic, though there are rumors that they had some rough dealings with Atari, and Microsoft did co-fund Gears of War.

So you get your funding

Page:: ( 4 / 6 )

So you get your $30,000,000 from a publisher. They're sold on your idea, they're sold on your team being able to create the game engine, they believe in you. That $30,000,000 is the advance. You have a primo team so you get a very high royalty rate – 25%. Low-end teams can see as little as 7-10%.

Here's your income formula:

(Gross – License) x Royalty – Advance => Net.

Gross income on a 3,000,000 seller = $105,000,000. Let's say you don't have a License, you're not relying on a movie, book, or sport, or a licensed engine. You multiply the remainder by your Royalty rate, or 0.25, and you get $26,250,000. Then you subtract your Advance, or $30,000,000, and ... you've lost $3,750,000. That's right, red ink baby.

Why is this formula so unfair to developers? For starters, it can be. Publishers control the purse strings, they have the money, they have the stronger negotiating position. They're the ones taking the risk with $30,000,000, not the developer. As we've all seen with games that bomb spectacularly, that risk quite often doesn't pay off. The publishers are hedging their bets by racing 5 horses and betting that two or three of them will make enough money to pay for everything.

That doesn't help the developer though, does it? Well, here is where sequels come in. But before we go there...

90% of game development delays are not because of bugs or late artwork or anything like that. Mostly, it's about capturing that elusive “fun” quality. Fun can't be quantified, you can't formulate fun. You can try, and design documents try to bring the various parts of development – design, sound, art, technology – together.

Due to the costs and risks related with prolonged development cycles, companies try to get everything done at the same time and then put it all together. It's like having a thousand piece puzzle, giving 200 pieces to different people and trying to assemble the parts and then put them all together. In the end, a lot fits, but everyone's going to have left-over pieces which keep it from being perfect. That's where the delays begin, trying to fit in the last bits of the puzzle to make it truly complete. All this is because of the cost of development – the more that can be done simultaneously, the cheaper it is and the sooner the game gets to market with its new tech features that no one else has seen before.

Now, when you're making a sequel, you're just doing a variant of the puzzle. You've got the core elements down. Anyone who's played Call of Duty and Far Cry knows how far superior the weapon feel, control, sound, and balance in CoD was. That's something that probably took months if not years to perfect. Far Cry weapons and movement are sloppy and feel laggy by comparison. So when it's time for a sequel, you can build off what you already did right – the movement code, the engine, the sounds, some artwork, the animation system – everything is there. The content is easy to create by comparison, and that is what sequels usually do – they have new levels, some new art, a few new enemies, and a couple of cool new technology/gameplay tricks.

This is why we see so many sequels – it's the only way for companies to make money. Did you just sell a hit game? Then you've got a built-in audience dying for more and it's very easy to make “more”. Even if the game wasn't a run-away hit (ie, Crackdown), if the core gameplay is solid, you can elaborate on it and work off that core fanbase to build on.

That assumes that the publisher has the patience to do this. Sometimes they'll have a great IP and potential franchise on their hands, then they'll want a sequel very quickly, typically in time for the holiday season, just to earn the quick bucks. They may be short on cash, or they may underestimate the difficulty of creating a AAA game, or perhaps they just don't see the purpose of continuing the franchise (ie, their 2-game movie license deal is going to be done, why risk a lot of money on a sequel?)

Can this continue?

Page:: ( 5 / 6 )

As you can see, it's not a healthy situation. Despite protests to the contrary, gamers want flashy graphics, they want amazing sound. Of course, they also want all those things they insist do matter – AI, level design, openness, cool scripting, great stories, and uniqueness. Then there's everything that people can't quite put a peg on that falls under the “fun” category. Weapon feel in Call of Duty, the intricacies of the S.P.E.C.I.A.L. system from Fallout, movement code from Quake, the announcer in Unreal Tournament, the degree to which player actions change the world in an RPG, the pause/unpause combat option in BioWare games.

Let's look at that last example in particular (we're just going to make this up, I have no idea if the following happened and I doubt it did). Imagine Knights of the Old Republic had initially been conceived as a fully real-time game, with no pause. It's well into alpha or even beginning beta, and while the game has a great story, dialog, good graphics, a cool skills/powers system, when you get to the combat, it's too chaotic. It's not too hard – the game has difficulty levels and the combat has been balanced for real-time, it's just that dropping grenades and using force powers in real-time combat is so fast that it's hard to enjoy it when it happens. The team, being ambitious, isn't satisfied with this. It's a good game still, but it's not worthy of that AAA Alberta Beef rating that BioWare is famous for.

So, hypothetically, they begin to tinker. They slow combat down, but that results in fights that are too long and there are now a lot of “attack” animations that don't actually do damage, instead waiting on player input. So the team goes to the next logical step and tries increasing hit points and putting the abilities on timers, but making them more powerful. That feels artificial and stagnant though, not to mention requiring a lot of re-balancing. Finally, they decide to take an old trick from their old games and include pausable combat. It works! The game is fun. That one little change made it all better. The downside is that you still have to re-balance it. The skills need looking at, the enemies need to be tougher, the boss encounters have to be re-thought, and development extends by 3 months. BioWare tweaks some more “not fun” issues along the way, and the game is delayed 6 months total, adding millions to the development cost.

Game development is inherently unpredictable. Even with an elite developer like BioWare, for every Knights of the Old Republic there's a Jade Empire (good, not great), or MDK 2 (decent, but way too hard and didn't sell). Possibly the only company that always succeeds with their games is Blizzard, and that too comes at a cost – tremendous development times, constant “delays” (though the games rarely have a set date), high expenses, huge expectations, maintaining a massive support structure (battle.net, Blizzard forums, PR, marketing, and patch teams), and so on.

Almost every publisher is running at the ragged edge, EA excepted. The risks are high and the rewards, while great, are not so great as to cover all the bases. Activision is the second-strongest publisher out there, but a year where their major titles bomb would put them on the ropes, and if the following year was bad as well, the company could be in real trouble. The safety margin is slim for them, and for publishers like 2K, THQ, and Ubi, it's almost non-existent.

Publishers are upset with Sony for bombing with the PS3. Their plans to develop games for it, even exclusives, have been shelved or altered. Suddenly the Xbox 360 and Wii need serious looking at. The market for next-gen entertainment is much smaller because the PS3 isn't selling anywhere near what it was expected to, and certainly not along PS2 lines. Yet nobody can afford to alienate Sony, because they still need the PS3 and maybe PS4 will be huge again.

Remember how developers would talk about hitting milestones in order to see further funding? That's not that big a deal any more for two reasons. First of all, publishers own many of the development houses out there. Secondly, if a company is funding five major games in a year, and one developer is having trouble which results in delayed milestones, what's the publisher going to do? Sink the $20m investment so far? Bite the bullet? It's better to power through and simply overlook development more, maybe demand a new project lead, changes, or feature cuts (S.T.A.L.K.E.R. comes to mind).

Where to from here?

Page:: ( 6 / 6 )

Remember when I said that organization, not design talent, is key? Put that together with the concept of how games are built – simultaneously in pieces. Every single piece has to be as close to perfect as possible as the game is slowly put together, and then the flaws need to be ironed out. Someone has to be able to spot the flaws, figure out how to fix them, and how to get everyone on board to do that – and there will be team members who think it's fine the way it is, or there's another flaw. Strong, intelligent, focused leadership is key, and that's a rare talent in any industry, especially a creative one like game development.

You can throw people and money at a problem but eventually you'll hit the law of diminishing returns. Even the best managers can't oversee 200 people working on one huge project. It's too much. Games hit 200 people during testing now, or if the publisher is desperate for a Christmas release. The new manpower can help but it needs to be trained and educated and brought into the community.

If you try to “crunch” through, paying overtime or bonuses, people burn out after a few weeks. They might be 100% productive at 8 hours per day, but only 80% productive at 12. Then you'll have team members who can't handle it, they have families, lives, they may be sick of the project or maybe they just weren't that passionate in the first place. This is why a developer or publisher will forgive many shortcomings in a new hire as long as he has a finished game on his resume.

The problem is that the art, programming, networking, and other demands are so great now that the next generation of games might well require 150 or 200 people. This means even more costs in a generation where console sales are less than spectacular. Thus we can expect this console generation to last longer than the last one, perhaps 7 years. Publishers will need to be able to recoup costs over several sequels, not just one, and build up a cash base to fund the early round of development for the next generation of systems (which promise to be even more complex, if trends hold true).

Tools development is lagging. Incremental increases in the power of tools are being made as the demands on artists and designers are becoming exponential.

This isn't an apocalyptic scenario, it's just that the game industry is in the middle of a major shake-up. EA bought exclusivity from the NFL, partly out of a legitimate fear of competition (because their games suck in comparison to 2K's), but they overpaid simply to rid themselves of a competitor. Without sports sales, 2K is even deeper in the hole. It's doubtful if EA will ever make back the money they paid the NFL, but they're guaranteeing a huge revenue stream that's denied to 2K.

Atari is down on the mat and the referee is beginning to count. THQ, Ubi, and Activision are one or at most two bad years away from looking like Atari did last year.

On the bright side, casual gaming is booming. Everyone plays PopCap games. Steam offers a distribution alternative that's attractive to low-end developers. Darwinia, Sam & Max, Defcon, and Eets are all examples of good games made on a reasonable budget, and distributed online.

It may be that we will need to accept paying higher prices for games. Games have cost $50 or so for roughly two decades now, while inflation has slowly chipped away at the value of those $50. Even then, having more money will not necessarily fix the problem – you run into the problem of managing that 200-man team that the extra money pays for.

Finally, there is evidence of gamer fatigue. Gamers are becoming more choosy with games, the pressure for more innovation is there. However, innovation is hard. It was easy in 1993 when the risk/reward ratio was absolutely there (remember that $300,000 cost and 300,000 copies sold being a “hit”?), and when new ideas were relatively simple and plentiful. It's also much easier to communicate a vision to a 20-man team, which is more flexible and less prone to inertia or falling into side projects than a 200-man team will be.

Perhaps the solution is going Wii – simpler, cheaper, smaller games.

The next few years should be fascinating, if a bit scary.