IGDA panel talks middleware

REDWOOD CITY, Calif.--Somewhere along the way from Pong to Peter Jackson's King Kong, game developers and publishers realized that creating games entirely from scratch with each new project was perhaps not the best allocation of time, money, and manpower. As the industry has evolved, it has grown more common for games to use assets, tools, and game engines lifted in whole or in part from other games.

These days such middleware is increasingly common. You can't swing a rag doll these days without hitting a game using the Havok physics engine, and the Unreal engine has been used on everything from Adventure Pinball: Forgotten Island to Harry Potter and the Sorcerer's Stone to, well… Unreal.

But is an increasing reliance on middleware a good thing? Aren't better results achievable by working from the ground up instead of repurposing other software for the job, no matter how adaptable it might be? Is the popularity of middleware going to continue growing?

Yes, yes, and yes, according to a handful of industry tech experts invited to speak at an International Game Developers Association panel last night at Electronic Arts' Redwood City campus.

The panel consisted of Craig Alexander, general manager of Activision developer Z-Axis; Kurt Busch, vice president of product development for Konami Digital Entertainment; Brent Iverson, chief technology architect for Electronic Arts; and Matt Wilkinson, director of technology for Activision. With the help of GameSpot executive editor Greg Kasavin as moderator, the lineup of seasoned industry veterans discussed the issue of middleware in game development for an hour before opening the floor for a half hour of Q&A with the auditorium crowd.

One recurring theme of the night was that middleware was going to be an inevitable key to cost-effective success in the next generation of game development.

"We're not bumping up against hardware anymore; we're just bumping up against time and money and the realities of development," Alexander said. "So smart, innovative ways of maximizing creativity at a fraction of the cost is a way to success rather than throwing 30 million bucks at a movie game title."

Alexander was not alone in believing that reusing code was a way to maximize creativity.

"I think that technology and code you're given, design decisions you're given…aren't really limiters; they're actually enablers," Iverson said. "If someone said, 'Make a great game,' and that's all the instructions you had, it would be pretty darn difficult to get started. But if somebody said, 'Make a great game about his movie that's coming out, or about dinosaurs, or pirates, that's a turn-based strategy game,' now all of a sudden you've got a lot of stuff that's laid out for you, and you don't have to think about those things anymore. Now you can say, 'How can I innovate in that space?'"

"I think technology a lot of times has that same effect… Now you can stop worrying about that stuff and you can say, 'Well, what do I do that's really interesting to the customer on top of that framework?'"

Even if developers aren't told to use middleware, they may choose to simply because the specialized tool gets the job done better.

"This is one of the reasons that Havok has been so successful," Wilkinson said. "First of all, most people don't understand physics, and second of all, who wants to write a physics engine? It's not very interesting, it's not very sexy, but it's great when it's in your game.

This middleware-friendly attitude hasn't always existed among developers. As Iverson noted, attitudes toward using somebody else's code have greatly changed over the years.

"A long time ago, I think reusing something was seen as a sign of weakness so it would be akin to, 'You wrote your game in BASIC,'" Iverson noted. "Now it's kind of interesting that you see in ads people brag about the middleware they used. If they used the Quake engine, they brag about that. It's a mark of quality knowing you're going to get a similar kind of experience."

Of course, not all middleware is a surefire shortcut to a good game. One well-known piece of middleware that came up several times throughout the night, usually followed by a smattering of snickering, was RenderWare, created by now-EA-owned Criterion Software.

"RenderWare's greatest asset was its own worst enemy too," Busch said. "RenderWare allowed you to get something onscreen very quickly, and it allowed a lot of people with really bad ideas to get something onscreen quickly and go around pitching them to publishers… It was dreadful stuff, the majority of it, but RenderWare itself was not bad."

To which Iverson replied, "People also use C++ compilers for evil, too. We don't blame the compilers."

Even when middleware is used by talented developers, the benefits have their limits. Havok may add rag-doll physics to a game, but there's nothing to keep competing products from incorporating it into their games, and developers will still need something unique to set their games apart.

"When you pursue a middleware solution, you're never going to beat the competition," Alexander said. "You're just going to match them. To advertise that you have rag doll in your game doesn't do much because there's 50 other titles [with it], but at least you're enabling a secondary or tertiary feature that brings you up to the level of the competition. If you're going to innovate in your category, you're going to have to home-grow it."

Companies are beginning to recognize the benefits of middleware and offset some of its disadvantages. Busch said that earlier this year, Konami Digital Entertainment began the process of centralizing its technology, essentially creating a corporate pool of different adaptable middleware solutions that wouldn't need to come out of a development team's budget and wouldn't be accessible to competing products.

Busch was all in favor of seeing it happen, but he noted that the idea had failed before. Busch and Alexander both worked for Sierra Games in the early '90s, when the company mandated that every game use the same interface, the same animation system, and other such technology. The company was known for its 2D puzzle-adventure games, and the technology Sierra had did that genre very well. However, it was less than adaptive.

"By the late '90s, everyone was playing Command & Conquer, Warcraft, and Doom, and that pretty much spelled the end of that genre," Alexander said. "One of the problems with standardizing too much is you better hope your genre doesn't go away; otherwise you're in trouble."

There was little disagreement among the panel members throughout the night, leading to the impression that the turn to middleware is inevitable in the near future, at least as far as the big publishers are concerned. As is usually the case, smaller development houses will be at a disadvantage here because they don't have the money to afford the premier middleware products and likely don't have the resources in time and funding to write their own. But for the big boys of the industry, the EAs and Activisions and Konamis of the world, middleware is an efficient way of reducing the costs, time investments, and risks associated with game development.

One caveat is that it has to be smartly designed and well implemented to be worth doing. Iverson stressed the need for modular design in middleware that allows for adaptability, lest developers find themselves mentioned in the same cautionary tales as Sierra.