Artificial Intelligence:

Ideally the skill of these participants should be tunable to allow you to find the right level for each situation. Considering aerial threats only for a moment, the scale runs from non-maneuvering bandit to instructor pilot and on to enemy aces.

This allows you to progress from familiarizing yourself with the new sim, though training against opponents good enough to teach you something useful and on to difficult and dangerous combat deployments.

No one enjoys getting shot down so fast you can't work out how to get a lock, much less shoot back. On the other hand lame AI can ruin the replay value of a game just as quickly.

These same principles of course need to be applicable to all areas of the sim where the paricipants should be "thinking;" everything from ground vehicles to pilots to the opposing "general" at the strategic campaign level.

Since everyone flies at their own skill level and each is different defining the threshold for this category is truly impossible. Perhaps then the best way to express a requirement for AI is to allow the difficulty presented by AI elements of a sim to be configurable to allow for a range of possibilities.

There is one thing though that has to be considered a must when it comes to wingman intelligence. Although accidents can and do happen in real life training and combat, a few sims released lately seem to have forgotten even to train computer wingmen in such basic techniques as how to fly.

You need only read the usenet newsgroup where posts detailing the latest misadventures of "wingbeciles"(TM) "LawnDarting" (TM) to know how frustrated sim pilots can be when their computer bretheren can't even keep a plane above the ground when flying to the next waypoint, much less provide support in a furball. This doesn't seem like too much to ask, especially for the perfect sim.

Sadly tales of wingmen flying barrel rolls around your aircraft are still rife in the news groups this year. And the efficiency of computer pilots in combat doesn't seem to have gone up much either. This still seems like an area for improvement until mutiplayer becomes more the norm.

ACMI/TACTS:

This is a great shame since such tools provide an excellent means to improve your flying, not to mention reliving the more (in)glorious moments of your most recent mission.

The threshold in this category is simply the ability to play back some or all of a mission that you have just flown. Once again, the facility should be available to record any flight, be it multiplayer or solo, campaign or single missions. Beyond this there are plenty of fancy features that can add a great deal to the basic package.

The perfect sim would have an ACMI system capable of saving mission recordings of fragments thereof. It should also allow file exchange of such recordings to allow pilots to share their escapades.

For strict simulation accuracy, the system should allow for wire frame display of the action, but given that this is one area where simulation can exceed the limits real life imposes, control of a full set of cockpit and external views would be a plus.

The recent Hornet 3.0 release has an excellent mission replay and save facility. It lacks fine control for use as an ACMI tool however. Su-27 also has a good mission recorder including a very nice feature allowing you to take control of a mission at any point during the playback. It too however was not really designed with ACMI diagnostic use primarily in mind.

Honours for perhaps the best ACMI system presented in a sim to date still probably go to the one in the Falcon3 series. This system has the ability to control a replay almost frame by frame coupled with extensive view control both inside and outside the cockpit. A more realistic wire frame mode is presented with the same kind of data cues the real pilots would use. Perhaps the kicker is an external view with a fixed camera that can be steered in 3D to any point in the action; this provides unparalleled ability to pick apart the flight and analyze what went well.

Take Su-27's .trk file flexibility in terms of exchange and ability to jump into the cockpit, add the basic ACMI from Falcon3 and marry it up to the other aspects of this sim and you would be very close to perfection.

DiD added ACMI recording to their ADF and TAW products this year. With stylized plane icons trailing across the sky it was possible to see the relative interactions of combatants and their weapons and targeting even. However, DiD left out much of the data (airspeed, altitude, heading and turn information) that you need for complete analysis of energy states in a battle. In this space then, we had some movement in the right direction but we're still not quite there.

Campaign Engine:

The perfect sim cannot afford to appear without some form of campaign engine. Which is to say, a sim supporting only pre-canned missions loses a lot in replay value and misses the basic threshold.

The basic requirement for a campaign engine is that it should present a sequence of missions that progress somewhat logically toward a more global initial objective. Within the sequence the sim pilot should have the sensation that their actions are affecting the outcome of the end goal and that missions while unpredictable are connected by a consistent thread.

Click to continue . . .

Total Air War.

The manner in which this is implemented is largely academic. A full blown real time wargame engine has been tried; EF2000 exemplifies some of the strengths of this approach. Yet one of the most successful "dynamic campaigns" was that in Falcon3. Implementation in this case was only dynamic in comparatively few aspects but it did follow some basic principles that differentiated it from sims that provide only canned missions.

Both are "dynamic" enough to satisfy most sim pilots; once again, suspension of disbelief is the key to it all -- do you feel as though you are in the middle of an on-going win-or-lose air campaign or are you following a sequence of clinical set piece studies in tactical airpower employment??

[footnote: The real key to this seems to be that some amount of the state of the theatre of battle should carry over from mission to mission. In missions, the set piece is always the same at the opening. In a "dynamic" campaign, the outcome of action from the last mission should be visible in the setup for the next. Part of the state is also bound up in resource management; fuel and weapon supplies are never infinite in real life but often they seem to be in sims.]

The perfect sim would probably have a full wargame implementation to complement the sim engine itself. However the threshold is probably nearer the semi-dynamic approach. Whichever is chosen, the perfect sim should allow for both solo and multiplayer play in the campaign setting. Also, some level of control should be available for the initial campaign set up and perhaps even for the objectives for the campaign. Such configurability will add considerably to the replay value of a single theatre campaign implementation.

One last plea for the perfect sim campaign engine: if we assume high fidelity squadron level operations are the goal then interaction with strategic assets such as AWACS and aerial tankers must be an important part of the mix. Such assets are a vital part of modern airpower and the perfect sim would be incomplete without elements like an obligatory stop at the "texaco" to top off the tanks.

TAW AWACS intercept.

DiD's TAW must surely rank as the most ambitious dynamic campaign implementation seen this year. iF-18 released with Imagic's updated TALON but TAW's realtime ongoing war is a bit more dynamic even than this high standard. TAW only falls short in allowing the player to really appreciate the wonder since it hides so much of the intricacy from the player: there is precious little chance to interact with the wargame engine on the strategic level. Players are left to marvel at the wonder of the war going by - if they actually stop to notice and I fear many will not even recognize it for the milestone product that it is.

This is one area where Jane's F-15 fell short. It boasted what is best described as a hybrid campaign; not really dynamic but much more than straight scripted missions. You could be forgiven for thinking that scripted missions was the implementation though after reading many of the debriefs. For a sim that poured such attention into the mission planning area, the debriefs were shockingly bare. This one misstep is a cruel blight on an otherwise excellent product, one that for me ruins the suspension of disbelief.

Multiplayer Support: