Before we get too far into the simulated personality in charge type setup here let me say I had no such intention. I find the setting commanders and governors type gameplay be rather annoying...the user is either in command or not. Furthermore the flow of battle is far too fast to enable meaningful 'personality' behavior modifiers. Setting fleet behaviors does essentially the same thing while not tacking a face to it or forcing you into a particular option set.

Having to set and use avatars and sub commanders in that fashion makes sense in Total War historical type simulation or pure a pure simulation, but is cumbersome unless your going to limit yourself to a semi turn based play style.

In fact most of the limitations in scope mentioned so far in the discussion seem designed specifically to eliminate or substantially diminish such play style.

-However ...although I do not approve of this play for the user I do believe in behavior modifiers for the opposing agents. By adjusting the fudge-factors in the weighted random states of the sub-commanders would create stronger groups of behavior. This would allow stronger campaign setups as the user finds himself going against multiple personality command styles.

Actually, I feel I should explain the root of the behavior system as it is now-

As I said before, I use the concepts used in AI War as seen here

http://christophermpark.blogspot.com/2009/06/designing-emergent-ai-part-1.html

http://christophermpark.blogspot.com/2009/06/designing-emergent-ai-part-2-queries.html

When I set out to do this though I found I had a unique problem....Classic comparison does not work (or work simply) when all the ships are dynamic in nature....so i needed the ships to be able to self classify-

Each ship when spawned goes down a list and inventory before deciding on 4 major behavior modifiers-

Class - "Corvette" "Frigate" "Destroyer"; "Cruiser" "Battlecruiser" "Dreadnought" "Battleship"-

Weight Class - "Light" "Battle" "Medium" "Large" "Armored" "Heavy"-

Type - "Gun-Boat" "Anti-Ship Missile" "Missile Defence" "Command & Control" "Carrier Wing"

Speed class - "Medium Speed" "Quick" "Fast" "Fast Attack"-

Optimal-Range is then determined based on the min/max values of the decided primary role weapons systems(each weapons, defense system has a value attributed to it) .

During combat each ship will of course fire on targets of opportunity, but each ship will try to find the most appropriate available target to engage. During the threat level check each ship within action range is compared based on all values to include ship health and type, IE a command and control ship will not prefer to go toe to toe with heavy battleship.

This is where I want my fleet layer to come into play.... I plan to auto sort each fleet and group the ships into formation based on all these factors, with the primary capital ships at the heart, ships designated as escort to the edges ( or other choose able profiles).

By being able to select standoff profile, the ships in the fleet will maintain formation and attack from maximum range. Escorts could be set to break formation and engage when the enemy breaks into a certain radius( or not). Defensive ships would set their inbound engagement radius several time over to shield the most value ships.

A command structure as a UI does not have to incorporate persona elements, although from a player perspective the introduction of a UI system that takes characteristics from a command structure could afford them a degree of familiarity whilst incorporating newer gameplay behaviours. However given the high degree of automation you have built into your fleet behaviours I can also see that a lot of the basic strategy and tactical elements that a player might consider utilising (under such a structure) have been in essence removed from them as an option.

Some thoughts.

You did not mention whether ships will change targets if a higher value target falls within range or if they will continue to target their first objective. This could also extend into issues of where a ship is crippled does it still get targeted or do ships move on to new targets.

At about 1.37 into your youtube I watch a missile (active targeter design) rotate through an axis after the ship it was initially fired at blew up before it hit. What I found interesting is that its trajectory as it retargeted a new enemy ship ran it very close to hitting the ship that launched the missile in the first place. This put me in mind of similar problems experienced by submarine commanders who released active-search torpedos then being targeted by self-same torpedos. Do you have friendly fire built into the game? Also are the deployment of mines, chaff, jammers and other such elements planned for?

If all the tactical elements are AI automated does this mean your gameplay for this game (when it comes to combat aspects) is more about building your fleets and then watching the results of their engagements in what is virtually a spectator mode?

This may seem to be the case, but as the scale of play approaches my desired heights(small SINS scale, dozen planets+,200+ ships) it's simply the level of abstraction necessary to enable workable play. One of the things I always disliked about the C&C series was the units were not capable of acting on their own in any reasonable fashion...you could drive an army into an enemy base and they would just sit there after clearing the offensive units...ugh. Compare this to Total Annihilation - where you can ignore entire non critical battlefields after issuing your overarching commands. Setting simple behavior tags such as "Fire at will/Hold Fire/Return Fire" and "Hold Position/Maneuver/Roam" create surprisingly complex results.

That said, and to the question of the user being a mere spectator, Any user commands (to include setting target, where to move) override the behaviors system until that command is completed. If the user tell a ship to target a particular enemy, they will do so until they or their target is destroyed...same for movement. For super heavy weapons this is critical.... you shouldn't waste nukes/blackholes/asteroids /mirv weapons on small fry(the cycle-time/cooldown is too long) .

As for the AI changing targets mid way, a weakened/crippled ship's attack value is higher then a fresh one...they make a tasty target! Ships that would not normally risk engaging them may take the opportunity it get some shots in.

There is no friendly fire at this time because it's simpler to handle the collisions...If I have to start checking the fire lines of each weapon all the time that's allot of checks- On pc not an issue....on a single core android tablet...

That's said the decision AI does not run 1 to 1 with the stepping, in fact its semi random within a given range but typically it takes several seconds between a single ship target preference check. When there is not a preference or the preference has been destroyed or lost contact with they revert to targets of opportunity.

Those tracking missiles actually are targeting on individual parts of a ship, so if the part is separated from their parent the missile will continue to track it. If the part completely vanished before the missile impacts they go dumb until they time out, there is no redesignation. I like things to be able to miss!

Mines exist currently but do not actually have a AI deployment method....still working on that one. I hadn't considered chaff/decoy's which is an oversight on my part, I'll add those ASAP. There is a EMP missile that has the ability to interrupt a ships offensive/defensive targeting ability BTW (and another that cripples engines)... Also on the weirder end of the defensive spectrum there is a "return to sender" beam that reprograms enemy missiles and fighter drones sending them back where they came from...thinking of doing a special shield with similar reflection properties that acts on a pulsing behavior(I mean, if it was always effective they'd be invincible.)

EDIT - Chaff launcher in! Clouds break track lock of enemy missiles and some direction change, making then likely to miss. Good call on that one!