Shots from Janes Longbow: Flash Point Korea. Click for a larger image.

"Suspension of disbelief" is a hot phrase in PC simulations. Some key elements of suspension of disbelief are communications and interaction with other elements, object modelling and graphics detail, and the campaign structure. Most developers have moved in the direction of a dynamic structure to give the illusion of a real time environment.

One example of this structure is the WARGEN engine which Digital Image Design developed for EF2000 and TactCom. WARGEN oversaw every element of the virtual battlefield, with two levels of AI: WARGEN and SMARTPILOTS.

In the virtual battlefield, stuff is happening all over the place, and it is NEVER predictable or repeatable. This feeling of dynamism, of randomness in a larger community of interaction moving toward a specific goal, has become the great diamond in the sky for sim developers.

In EF2000, the entire battlefield is alive with activity, and all of it generated by WARGEN in accordance with four AI levels: the Grand Strategic level, where all major decisions are made politically; the Strategic level, which is responsible for organizing everyone and everything in terms of where they are and where they are going plus need for reinforcements or supplies; the Operational level is the player: missions are sorted out in terms of priority and who will do what where; and finally the Tactical level where a sophisticated set of rules resolves combat that takes place elsewhere and generates comprehensive statistics for use by the higher levels of AI. Smartpilots looks after all the computer controlled planes (CCPs) in the arena.

But many players have commented that Janes Longbow achieves a very similar FEEL in spite of the fact that it is something of a hybrid, using a structure that is essentially predetermined, but with enough randomness and interactivity to give a SENSE of dynamism. The initial release of Longbow lacked this, but in Flash Point and Longbow GOLD the AI kept track of campaign progress between missions, allowing for an ongoing flow of missions toward a predetermined goal with a sense of participation in a larger picture. Destroyed objects stay dead between missions, adding to realism and "suspension of disbelief."

Flash Point has too many added features to describe them here, but suffice it to say that it was not a random act of kindness that PC Gamer awarded Longbow the best sim of 1996! In Flash Point objects had simple mission goals: go to this point, attack this object. With this added dynamism, in Flash Point Longbow achieved new life. This is essentially a "semi-dynamic" structure, and may be the direction of Janes in sims in general. I recently interviewed Andy Hollis for his thoughts on campaign structure:

Csim: Andy, would you agree that a fully dynamic campaign structure would be the ideal for a modern military simulation? Why or why not?

Andy: I suggest that a fully dynamic campaign *experience* is the ultimate. That can be achieved many ways. You can do it by taking a pure engineering approach and simulating everything as you mention above. Alternatively, one can take a more "artful" approach, providing the appropriate cues only where people notice them. Think of it this way. If you have a black box, and whenever you put certain things in, you get the expected things out, who cares what is in the box?

One more point: It is possible to do a good job and a bad job of either approach. A good job of either will beat out a bad job of the other any day. And therein lies the rub... Folks love to debate the pros and cons of the methods, judging each by the relative examples that are on the market. They fail to consider the quality of the implementation which, in the end, matters more than anything else, in my opinion.

Csim: What do you see as the essential elements in a fully dynamic campaign structure?

Andy: To answer this, its best to first look at why campaigns came about in the first place. In the first flight sims I was involved with (MiG Alley Ace, Solo Flight), single missions were the norm. We strove to create believable worlds and dramatic situations (all on an 8-bit microprocessor :-). Players were given strong payoffs for success and specific feedback for mission failures. With that accomplished, we then looked for something to connect it all together and give a stronger sense of purpose and consequence. That begat "the career", with medals and promotions (Gunship, F16 Stealth Fighter, F15 Strike Eagle II). In a way, a career was the "lifelong campaign" and satisfied the need for the player to identify with his pilot character, and not want to lose him.

With F15 Strike Eagle III, we added in our first basic campaign elements to give an intermediate sense of purpose, somewhere between a single mission and a full-length career. Though campaigns have progressed significantly since then, the fundamental elements are still the same:

1) A series of missions that lead to an ultimate goal. The missions must make sense as to the way they contribute to the ultimate goal (e.g. take out the GCI & SAM sites first, then the enemy airfields, then the local AAA, then bomb downtown HQ).

2) The series should feel like the results of one mission dictate the next mission type based on the new situation. This happened in SE III by keeping track of priorities of remaining targets and selecting new targets based on weighted priorities.

3) The resource situation should make sense from mission to mission. In other words, destroyed objects that are not easily replaced should stay dead.

4) The player should get clear feedback that describes how his previous mission affected the overall situation so he feels good about his successes and reticent about his failures. A strong sense of how the overall situation is evolving will play upon a player's sense of team commitment and evoke a stronger range of emotion. That's a good thing.

5) Suspension of Disbelief. Everything you do must be believable. If you give the player even small one bit of information that does not make sense, then it becomes "just a game" again. Dumb enemy AI, dumb friendly assets, unrealistic campaign flow, and bad presentation (hokey acting/graphics, gamey text descriptions), can each ruin an otherwise excellent sim experience. Its hard enough to get these right in a single mission, but very hard within the campaign.

Csim: Why did you decide against a fully dynamic campaign for Longbow?

Andy: What people typically mean by a "fully dynamic campaign", is one where *everything* is algorithmically generated on-the-fly. That means having sophisticated AI figuring out modeling what every unit is doing throughout each mission, then how the entire situation has then evolved, analyzing it to give new orders and figure out what new appropriate missions are for the player. We haven't decided not to do this. In fact, we are doing it. Its just that our methodology is not as algorithmic as some others might employ. Why?

A human will always be able to do a better job of handcrafting any particular mission than a computer will. Now, before anyone gets all excited, I am not saying that a human always *will* do so, only that they are more capable. We want our missions to have plenty of interesting, dramatic, and unexpected situations. This is tough to do algorithmically, since the computer will typically generate missions that fall into a specific set of templates, flushed out with some variations. That has been borne out by the many attempts at the algorithmic approach, which always seem to degrade into variations on the same set of six or seven mission. In Longbow, and especially in Flash Point Korea, we were able to handcraft a long series of captivating missions and link them together in a way that satisfies the basic tenets of a campaign system as described earlier. Going forward, you will see even more elements handled to add variety to the system for the sake of replayability.

Csim: What is the direction for Janes future simulations with regard to campaign and mission structure?

Andy: We are going at it two different ways. For one of the Origin SkunkWorks product lines, we have taken an approach that expands on the method used in Longbow, adding in much more variety of mission possibilities at each campaign tree situation node. So, even the first mission of the campaign will typically be different.

Additionally, within the missions themselves, there is significant variability in which enemies will show up and when, and the AI that controls them will have varying parameters. Also, which the decision as to which situation node you go to after a mission has many more parameters than just the simple win/lose of the preceding missions. In general, no two campaigns will play out the same.

The other product line is going at it more algorithmically, but not to the extent that some sim developers have. For this line, we are keeping track who controls what ground on the battlefield and the results generated by the ground forces themselves (which theoretically you are supporting) dictate how the territory changes hands. Mission selection and availability depends on what potential targets you know about in the enemy's area (maybe some that you or your wingies detected during previous missions), and also what assets you may need to protect. All of these things are prioritized and mission selection comes from these lists. In addition, we have special missions that show up when certain special circumstances are noted. These are handcrafted. In the end, we get the best effects of the algorithmic approach and the handcrafted approach.

Csim: Can you explain what is meant by the term "semi-dynamic" campaign?

We use this to describe the above systems since they are not entirely algorithmic in nature. That seems to be what folks want to label as "fully dynamic". So if its not "fully algorithmic", it must be "semi".

Csim: What are the advantages of a semi- dynamic campaign engine?

You can get the wide range of mission variety that only a human can generate, yet still provide a seemingly endless series of campaign progressions. It combines the best of both worlds. The best missions on an individual basis, all combining to provide an infinite variety of interesting campaign sequences.

Csim: What are the limitations of a semi dynamic campaign, and how will you address them?

The only limits are the creativity of the people creating the campaign sequences, adjusting the parameters, and crafting the missions. The tools are there, its how you use them that makes it great or not.

Making great sims is more an art than a science. Its very straightforward to take the engineer's approach, but it does not give a great game. Great gameplay is an art. That is the advantage that the experienced Jane's staff brings to the table.

Flash Point offered a full screen FLIR white hot or black hot mode...

Csim: Janes Longbow was a great simulation. One thing I especially liked was the incredible variety of missions: rescuing trapped assault teams, escorting other teams in, etc. And the fact that there was a ground war and an air war progressing at the same time, with even foot soldiers becoming a threat! After such a great sim success, where do you go?

Andy: One of the two methods described above. And maybe some more variety in what you might be doing in those missions.

Csim: What do you think of the virtual battlefield concept? How far away are we from seeing this reality? What will it mean for sim lovers?

Andy: Different people see this as different things, so its hard to say when we will get there. In general, though, two things have to happen first. One is that on-line gaming must become economically feasible.

Right now, there are a number of first generation systems out there, but no one is making as much money doing this as for traditional package-goods retail gaming. On-line is still considered an additional feature. To justify the huge development expense necessary to do it right, it will take a comfortable expectation of a reasonable return on investment. (Geez, I sound like a banker :-)

The second enabler is having some standards emerge for communication between "modules". Again, this ties back into the fact that products are currently designed to be self-contained first and foremost, with on-line being an afterthought. Only when this changes will developers be willing to compromise on functionality in the interest of generality.

That said, I believe we are already making some steps in this direction. You see some different products (always from the same company, though) already beginning to work with each other. This will continue. Soonafter, you'll see on-line "theaters" where this action can be setup and managed. Little-by-little we will get there.

The other thing to consider is scope of play. Remember, this is for entertainment value. There are scales to things in combat that do not lend themselves nicely to fun. "Virtual Battlefields" need to combine elements that have similar scope so that players have equal chances of having fun. My favorite example is talking about combining Attack Helicopter games and Tank games. This does not lend itself to a good game. If the thing was even close to realistic, I'd hate to be the guy in the tank since he would almost always lose, . Even more obvious is combining individual infantry with fast-moving jets. Just doesn't work.

Csim: How will hardware changes impact coming projects: 3d hardware, force feedback interfaces, dual monitor setups....?

Andy: The great thing about this industry is that about half the innovation comes from just learning to do better gameplay, and the other half comes from hardware enabling. Faster CPUs get you better graphics, physics, AI, etc. More monitors, better joysticks, force feedback, all combine to give new user interface cues. Basically, all contribute to a more immersive experience. Any new technology that adds significantly to the experience will be supported. Its just a question of installed base (how many people have them) versus cost to support. You'll see Jane's doing whatever makes sense here (even if it doesn't seem like it makes sense until you see the products).

Csim: What would your next project be if you had all the talent and resources you ever wanted?

Andy: Personally, I like to be involved with sims of current equipment. In addition to being modern hi-tech and sexy, it is important that the equipment be near-operational so that the appropriate info can be gathered on how it really works. When we can get a set of operator manuals or pilot interface design docs, the real pilot workload experience can be given to a player. Until that stuff is available, it is all guesswork (albeit intelligent guessing). Certainly there is a market for sims of equipment that is not yet even close to being in service (thus, no detailed docs), but those developers can do whatever they please, because no one can prove them wrong. There's a whole flood of games out there like that. To me, though, you may as well be writing a spaceship game. It is someone's perception of the reality of the interface, not the real thing.

Csim: What was your favorite non Janes/Origin/EA product released in the last twelve months?

Andy: Command & Conquer. Very immersive. Great balance, great variety. Nice presentation. Why not a sim? I have a hard time playing other people's sim products because I look at them too clinically. I keep seeing things that could have been done better and it spoils the game experience for me.

My favorite sim is always one that is currently under development at Jane's.

Csim: Will Janes do a Comanche simulation?

Andy: Maybe. Again, see my notes above about docs and detail.

Csim: Can you say anything else about present projects?

Andy: Why yes. We are doing a @&@!*^!% and !%^@...and then &!%*^

Seriously though, we generally do not like to get in the business of announcing products and content early in the development cycle. This would be a disservice to our loyal customers as it raises expectations too early and may lock us into commitments that we may later want to change. I've seen too many other developers get buried in their own hype. Besides, it wouldn't be the Origin SkunkWorks if we told you everything, right?