View Full Version : 2005 - The Year of the Middleware?

09-19-2005, 05:55 AM
How did 2005 become the "Year of the Middleware"?

Before you had a few players with very recognizable engines (ie Unreal). Now, every other week there is a new killer middleware with not just great graphics, but also (semi-)real physics, non-scripted AI, and everything else a Devry drop-out needs to create great games.

But, I have a few questions.

1. How does real physics work online?

Its getting to the point where physics engines truly understand gravity, friction, and fluid dynamics at ever smaller scales. What is even more incredible is the "randomness" of the physical world is finally appearing in games. ("Chaos Theory" for Jurassic Park fans)

But how will this work online? As physics become more and more random (ie, explosions causing more explosions), it is impossible for two players to experience the exact same events. For one player, the explosion dynamics causes the house to fall at 45 degrees, but for player two, its 48 degrees. These small changes become huge over time. By the end of the match, the two worlds look nothing alike. How will this work?

2. No more originality?

These middleware engines will no doubt allow studios to do a lot more with a lot less. But there must be a price. Will studios alter their vision to make it work with the engine they just purchased for 40% of their budget? Will every game become a sandbox shooter simply because they are so cheap to produce? For example, which engine could produce Psychonauts or Conker and if none, which studio will take a chance making these expensive and risque games from scratch?

3. Real AI vs Script.

Up to now, its pretty much drooling AI bots trying to destroy you at all costs. But soon, bots will drop their premade scripts and instead have real personalities that allow them experience fear, hesitation, and perhaps even compassion. Beside the "random" effect this will cause online, what type of social issues could this cause? I can't wait for Republicans to show video footage of gamers killing aliens who are begging for mercy.

4. No more programmers?

What year will be the turning point for programmers. Right now, studios need the best and brightest but with these engines, almost everything can be made using simple scripts. As far as you know, are there any games in development that do not need any in-house programmers at all?

Any answers, specially to question 1 would be awesome!

09-19-2005, 06:29 AM
"everything else a Devry drop-out needs to create great games."

Where are these games?

The games biz is and will be for our lifetimes a completely technology driven industry. We are still doing kidstuff these days. Static worlds with dynamic actors performing a tiny range of actions.

The point we are at right now is nothing more than a similar milestone back in the days when you could display photorealistic images for the first time.

09-19-2005, 07:13 AM
The games biz is and will be for our lifetimes a completely technology driven industry. We are still doing kidstuff these days. Static worlds with dynamic actors performing a tiny range of actions.

The point we are at right now is nothing more than a similar milestone back in the days when you could display photorealistic images for the first time.

True, but I think there is a very serious "business" change in the video game business. Bigger than anyone is predicting.

Its kind of like the Porn industry. Before VHS, porn films had stories, characters, proper lighting, etc.. But after the launch of VHS, it didn't make any business sense to spend $1 million making a great film when a $25,000 amateur tape made almost as much money with less risk. Sure the content is poorer, but profit is what counts.

Middleware is like the VHS. No longer do you need professionals to make quality games. As long as you have good artists, you can license what you need make a game that is 80% as good at 20% the costs.

And imagine what will happen if art-assets become comidities. Now you just need cheap designers to drop things into place and add some style. Suddenly, you went from a team of 50 down to a team of 5.

Amateur game making is probably 4 years away. I can't wait to make the first game where a Canadian Jewish Liberal Nerd saves the world from the evil aliens and gets the girl. (Islamic-Jihad released a similar title, but trust me, its a bit of a downer) :)

09-19-2005, 08:07 AM
There is simply nothing new this year other than the term middleware has come into vogue. Game companies have been licensing tools or libraries of code almost as long as there has been a game industry. RenderWare or the Unreal engine are a small fraction of even the simplest game's code base.

Amateur game making has been going on since the first stored program personal computers were available.

09-19-2005, 08:15 AM
About the number of programmers dwindling as opposed to artists they are starting to use in games this next generation procedural generation (or some similar terming) which means that things like skies and fields can be generated through a complex algorithm and some art. The amount of art needed decreases significantly and the burden of how it looks is shifted to programmers.

I think that the number of artists will decrease much faster than programmers on some games or at least inrease much slower.

09-19-2005, 07:01 PM
#1 : I'd have to say, that a lot of the really random stuff will be relegated to things that don't matter as much -- your example of debris is one such case. Once the house has fallen, you don't really care about it anymore, so let the debris be wherever, and it's just trash at that point. For things that DO matter, you basically have to keep tabs on who holds the "official" information. On the other hand, things like throwing a grenade from point A at direction D, the client who threw it is considered the "official" information source, so whatever he says goes. As long as that information is the same, the reactions will be relatively deterministic, but for the actual simulation of the grenade bouncing around and when it explodes and does damage, it's generally normal for the server to hold official info on that. You'll generally want to save peer-to-peer type physics sims for games that only suppport small-scale multiplayer like 2-player co-op or 4 player deathmatches. Anything larger, the server is the physics.

#2 : Well, cost structures are basically the determining factor here. The fact that the industry is so hit-driven that there is rarely anything in between boundless success and tanking failure. That's why, if you listen to a lot of the statements from say, Allard, he keeps mentioning the casual gamer. They're basically a necessary component of the market if you expect the market to expand much at all. Cost is going up exponentially, while the market tends to grow linearly.

That's not to say that originality will die, but if it does, the US will be the first to kill it, I think. The US is not a real gaming culture, and I doubt it ever will be. The big sales genres in the US are mainly those derived from simulations. And the big problem is that originality always equals risk, and publishers are the powers that be -- putting power in the developers' hands is probably not going to happen within the next console gen.

#3 : Really good AI is still a ways off, I'd say. There are so many things that still remain a sort of holy grail. The more realistic you try to make it, the more scrutiny it will fall under. It's part of the problem when you have a market made up of stupid people. You can make a horror game where you can make NPCs act frightened and all, and the logical aspect of it may be acceptable, but the aesthetic aspect could be way off, and that's the really hard part of it. You can tweak it till you're fingers are bleeding from hitting the keyboard too much, but the problem is that you'll invariably tweak complexity onto complexity, and that breeds surprises, and surprises are never in your favor.

#4 : I've been through this a few times, but basically, I'd say this is probably the opposite of the truth. Middleware if you take it right out of the box, is always useless. If anything, I'd have to say this breeds demand for more specialization. If you license a physics middleware package, that does NOT mean you don't need a physics programmer -- in fact, it means you need a better one than the average pick of the litter, because it has to be someone who can pick apart a large library without breaking things. A physics package, even the very best ones, out of the box are typically trash, and really serve the common denominator (which for physics is pretty limited) -- to be usable, you'll have to rework all the damping, rework the collision (almost all physics middleware has wrong collision), rewrite every spot that needs refinement and bs where you need bs'ing -- all of these aspects are never the same from one title to the next. The same applies to rendering -- most render engines are slow as hell out of the box and need rewrites at so many levels. I could tell you volumes of everything that I had to rip apart in Lithtech just so it could render at a decent throughput. Or the fact that the base structure to the codebase was completely useless for the purposes the "Cinetragic"* team had at the time, and we basically had to take all the game-side interfaces and do them completely from scratch. That's how it usually works.

The real power of middleware is not that it saves you much work on the technology end or the game development end, but that it comes with a toolsuite and standardizes the art pipeline. For instance, something like Unreal has modding tools all over the place, and the idea of having the same tools in your own development house means there's so much more talent available to you who doesn't have to be trained. Similarly, you take Havok, whose true brilliance in the marketing end was getting Reactor tied into modelling packages.

* note : "cinetragic" is just a nickname for the now defunct company I was working at back then.

09-19-2005, 09:15 PM
Thanx cpiasminc!!!

That was a very thorough, well thought out answer! You being a developer, may I ask if you had any hands on with the new systems?

09-19-2005, 10:57 PM
360, yes... Beta kits only, so far -- our company isn't doing a launch title, and we're not a top-tier dev, and we're not doing a AAA title -- so we probably won't see final hardware in 2005. PS3... still not yet.

360 beta kits... not impressive. It's really quite easy to become pixel-fillrate bound on the beta R500. Still, it's been over 2 months since I looked at it because a current-gen title which is due out in about a month and a half is kind of taking precedence. I don't know what's going on with that project at all.