An interview with Gilman Louie, Chairman of Microprose.

Let's move on and talk about the campaign itself. One thing not really appreciated yet is that F4 completely integrates two independent and fully dynamic wars: an air war and a ground war. Tell us about the thought and method behind this.

GL: As I said earlier, what we started off with was a ground war. The basic philosophy is that the air force is used as a means to support the ground effort. But it's the ground war that wins or loses the war. Basically if you don't have an army, it's very difficult to win a war.

So what we did was we had one programmer, Kevin Klemmick, who is the lead engineer on the entire campaign side, he hates flight sims with a passion. Even though he grew up around airplanes and his Dad has a plane and he used to fly with him, he likes strategy games. He was the perfect guy. We didn't want a guy who loves flight sims, we wanted a guy who loves strategy games.

And what Kevin did was build an entire campaign system around the basic five circles doctrine. That is basically the political structure, the infrastructure, moral, your army and air force and all these assets. And he built a complete strategy war game, on par with the best you find from SSI or any of the other classic war games by Avalon Hill.

And that is what generates all the things in the war. We generate physical analysis of outcomes of different kinds of units. We actually have a tool.. say, take all these F16s and MiG 29s, and put them in statistically, what would be the results. And he feeds those datapoints back into the campaign. And the campaign runs the complete war. So that is number one. He writes the ground war, then figures out how the air force would support that ground war. Then he figures out what are the things you would do in an air war to counter a ground war. So he builds that up in layers. And then he bridges that with the simulation.

On the simulation team we have these two terms: aggregated units, and de-aggregated units. Aggregated units are Kevin's problems (laughs); de-aggregated units are the sim players problem. So, you're flying along and you will see a label out there in the distance, and then it will jump a few pixels. When it's jumping it's probably de-aggregating into multiple vehicles, depending on the vehicle type, it will hand off between Kevins units and the sim units.

But what is cool about the ground war is that it is running as long as you are in the game. Unless you turn off time. So the moment you get in and click on campaign and the UI comes up, the war is running. And it's running even as the game is loading a particular mission. You can be sitting there in the load screen and you'll hear an explosion, because your base just got bombed. And you're tossed back out to the interface, and the UI will ask you to choose another mission because your airplane was just destroyed on the runway as you were waiting to fly.

It's the most amazing thing. Time matters in this game, because it was written as a strategy based ground war first, the air war second, and the simulation as a layer on top of that.

And this is the first military simulation on the PC with a full blown ground war integrated with the air war, and both running dynamically in real time.

GL: Yes. Now people will look at the ground war and see some of the flaws, because we had to do trade-offs. For example the ground war doesn't do collision detection, so you may see a tank drive through another tank. But if we had to do collision detection on 100,000 vehicles it would bring your computer to it's knees. You would need a Cray just to run the campaign.

The other thing we gave up was specular lighting. It took too much processing power, so we fake it. We did that because we want Kevin's campaign - outside of the flight dynamics of the aircraft - is the most important thread. And he wrote it on a separate thread than the rest of the game. So when I say you can run this on NT 5 beta 2, you are actually getting the campaign on a separate thread. And that thread will run under NT on another processor. So you can get a 20-30% increase in performance on that system.

Wow, sounds like the basis for some incredible expandability.

GL: Oh yeah! I'm sure when this game comes out that .. on combatsim.com, in the bugs and the features column.. you'll have to keep adding more pages. There is so much in this campaign, so many combinations and permutations of different events, that we just can't track them all. And we believe we caught most of them. But there will be idiosyncrasies and things that people bring up, and our job now is to be as responsive to our customers as we possibly can.

With a game of this complexity that's to be expected, so all we can ask of you guys is that you deal with the issues that come up. The responsiveness of yourself and your team so far is as good as we've seen.

GL: It's funny. We've had a lot of calls about the testing process. The way we used to do it is play the campaign a hundred or a thousand times and then you kill all the bugs. Then you just keep doing this until there are no more bugs.

The problem with Falcon is that to win the campaign you have to fly one successful mission every hour (sim hours). A short war to win will take you six days. A typical war will take you twenty to forty days. So, if the human player actually played the whole game, that's 24 times forty, at an average an hour a mission, that's how long it will take you to complete one campaign. So it was not humanly possible to test out all the combinations. We did statistical analysis in order to test the different components and we had a couple of guys run through the entire war, but it's just not possible to test under the old methodology here.

With that awareness of how long a campaign plays out, we want to make an official announcement that combatsim.com will move into the pizza and coke delivery business wherever Falcon4 is sold…

GL: (laughs) and my Engineers say, "We need some sleep, cause as soon as the game comes out you're not gonna let us go anywhere…"

Gilman, you've kept close tabs on our forum, so what are your current priorities for enhancements?

GL: First is to ensure that any instability in the game is dealt with. Because this is a multi-threaded game, issues of stability will be because of the thread manager. So we're very aware of what those issues are and every rev will make the game more stable.

Click to continue . . .

Shots from the ground war.

GL: Issue number two is multiplayer. It's pretty good on the shipping product but not to the point where I would like it to be. So over the next forty five days, every edition will improve multiplayer. We're doing things like compressing packets better, better smoothing algorithm and so on , and we'll put lots of energy behind multiplayer and multiplayer features that people want.

And the third part will be the campaign. Anything that involves improving the campaign, AI improvements, and features that people want.

The most requested improvement at the moment is the high res cockpits, and multiplayer is our big priority.

Another one we are hearing a lot is the whole clouds and weather and wind thing. Will we see cloud layers half way through '99?

GL: The cloud debate in this company is probably the most furious debate among the graphics guys. There's about a dozen different ways to do volumetric clouds, and what I said was… in fact, some of our test suites actually have volumetric clouds, but it's a bit slow.

So what I told the guys to do on the first rev of the game was, "Look, I want volumetric clouds as much as you guys. But.. I want the clouds to have a tactical effect in the world, both in how the AI uses the clouds, and how it effects weapon types…you know, firing heaters through the clouds.. being able to use the environment as part of the context of the game. That was the highest priority.

So what we have is the flat layer thin clouds with about a couple hundred feet of fogging effects on either side of that. But you can still shoot and fly directly through the holes in the clouds, and shoot it clean. And that works very nice, because you can actually spot a plane zipping out underneath the cloud layer through one of the holes, and zip through it and get a missile lock and fire on him as he's going through the hole. Whereas once the bank is between you you won't be able to lock on that target. There are some cool things like that.

Volumetric stuff, we'll have a team messing around with the clouds, and that will be a feature that will probably come up in the summer when I believe that we will have machines fast enough to really support it.

Cool.

GL: The problem with too many of our end users is that they turn everything on max. And then they complain that the game is only running at 2 frames per second!

Too true. The positive part is that the survey we did showed the average system was a PII 300.

GL: Absolutely. The current weather has the following things: there is a wind pattern and the clouds will move, there is wind in the world, there is a contrail layer in the world. And all that affects the game. If you're up there in the skies the enemy planes spot you at fifteen miles, but if you're down in the dirt they have a very small eyeball range because you get lost in the ground clutter.

Ah, that's why my HUD is drifting all the time! I thought it was my F22 Pro acting up again!

GL: (laughs) Yeah, your Tm stuff shouldn't drift. We wrote code to take out the spikes!

Back to multiplayer for a second, when is the planned release of the server module?

GL: Undetermined as of yet. We're in the design phase. We have to do a costing analysis also. But it's one of the priorities, and it will take a lot of work to do right. Stay tuned. As soon as the design plan is finished we'll get you more information.

What is the current priority for add-on aircraft?

GL: The MiG 29 and the F18. MiG 29 we're looking at having both the SMT, kind of the F15 variant that has the FastPack gas so it has more range, it has all the avionics, MFDs, vs the old A model. And with that pack we'll also do the A model, so you can fly either a NATO (German) MiG 29 or the Soviet block MiG 29.

When I talked to Tom Nichols a few months back he talked about naval AI expansion along with the MiG 29 game, can you tell us more about that?

GL: We were looking at that, but we're going to reserve that for the Bosnia affair, and we might do a navalized version. But that will probably be a patch that ends up in our F18 game.

What's your guess on schedule for MiG 29?

GL: Under a year, hopefully much shorter but it depends on how we have to allocate resources to support Falcon 4, and any features we add for Falcon 4. We'd like it out before Christmas.

Is F18 concurrent in development with MiG 29?

GL: The Falcon team will split up into two teams: one team will do MiG 29 and one team will do F18. F18 will follow MiG 29, maybe come out three months later.

F18 will be very cool, it will be the Taiwan scenario. It takes place four years from now where the Chinese have been building up a bunch of Su 27s and there is a conflict over Taiwan. Taiwan flies F16s and the US Navy will be heavily involved, maybe even Australia will be involved, both of them fly F18s.

Are we going to see full scale carrier ops with F18?

GL: Yes. And we want to make sure that the carrier stuff is done right.

Would you do the two seat naval version?

GL: Possibly. We're in discussions right now. A lot depends on what data we get. I don't like doing games unless I have solid data on the aircraft. Real data, real pilots, preferably somebody who has been in that aircraft, who continually flies the aircraft so we can check and make sure the declassified stuff is right.

Go to Part III

Falcon 4.0 is produced by Gilman Louie and Steve Blackenship.