So one thing I have noticed while playing many games, and keep forgetting to ask about, is AI fleet targeting priorities.
Often as I am moving my fleets around, I will find my fleets end up in a position that could be vulnerable to attack from AI fleets. However the AI fleets all seem to have specific targets they keep moving towards, completely ignoring potentially easy "targets of opportunity". Like I might have a small fleet moving to be stationed as a light defense force for a starbase, that ends up revealing an enemy fleet in the fog of war, and my fleet doesn't have enough movement left to get out of the enemy fleets strike range. Luckily for me though, the AI doesn't recognize this easy kill that just got dropped in it's lap, and during it's turn just keeps moving it's superior fleet off to whatever objective it was "originally programmed for" instead of wiping out my smaller fleet.
I certainly don't know anything about how difficult it is to program the AI, but this type of scenario is one of the HUGE reasons a player has an "easy" time against the AI. The AI never seems to be able to take advantage of "targets of opportunity" the way a player does. It seems the AI fleets are given a singular purpose, they won't do anything else until they complete that purpose; they don't make any "on the fly" adjustments.
This seems to be a pretty common thing in many games, not just GalCiv IV. If there is some way to "fix" this, it would go a looooong ways toward making the AI more challenging, without having to give it tons of extra resources and similar "cheats".
Happy Bug Squashing !
This is a big one, indeed!
And it should go hand in hand with the human players fleets/ships behavior - if we set a faraway destination, the ships will move there regardless of whatever gets discovered during their move! So sometimes you click on a destination far away and the ship starts moving - after one hex-field it discovers an enemy fleet! Now it should ditch the remaining moves and wait for the players input - but NOOO it expends all of its moves and the human player has no chance to react on the discovery of the enemy ship
There should really be a re-evaluation of a fleets destination every time it "discovers" a new ship - at least if it belongs to a faction that you are at war with (or have no open borders treaty)...
i think that the reason for this is that is that if the ai goes after all easy targets with their big fleets, then players can just make tiny ships to keep the big fleets occupied since they can only attack once per turn, and thus cause the AI to spend it's fleet time chasing around tiny useless ships while the player's fleet is actually doing something useful.
good idea, i agree
And on the flip side of that, the player can have 1 little ship go take over the AI colonies because the AI won't have it's big fleet passing by just squish that one tiny ship.
I think it would be better for the AI to crush "targets of opportunity", but there is always going to be some sort of "balancing act" needed.
One thing I was thinking of that could possibly work is for the AI to make fleets that are specifically "hunter killers", fleets that just go around looking for "targets of opportunity". Then the AI could also make it's fleets that have specific objectives like "go attack that world or starbase or shipyard" etc, and they wouldn't deviate from that task until it is completed. Once again however, I have no clue how the AI coding works, so I don't know how hard this would be to implement without using up lots of cpu and memory resources.
Just food for thought.
Well, I guess ideally there should be a recalculation of target/destination assignment as soon as a new enemy ship is discovered - which could be several times during the AI´s turn.
If this proves to be too calculation intensive, it would still be a big improvement if the recalculation happened as soon as the AI STARTS it´s turn - but of course it would need to "remember" that it saw a ship passing - I bet that right now it will forget the presence of an enemy ship that showed up briefly in scanner range as the fleet passed as soon as it is out of scanner range again.
That way a fleet that was going "somewhere" would get to rethink that assignment and possibly go back to the last position of a contact and try to re-establish it.
At the very least it should take a new target into account if it is still visible at the beginning of the turn - not only once it has reached its final destination.
This is probably much more complicated than aerez4546's idea about having "hunter-killer" fleets but:
Would it be doable to allow the AI to re-configure it's fleet to an appropriate level to attack that "target of opportunity"?
For instance, you have 15 Drengin ships in AI Fleet 1. A Terran 2-Lasers-1-Shield Fighter becomes a target of opportunity by entering the fleet's area. It's kind of stupid for the whole Drengin fleet to change course to take out that Fighter, but it would make sense for one (a Cruiser) or maybe two Drengin (Fighters) to give chase...
This is also a great idea!
I think we will - however - need a very simplified set of rules for fleet composition, forming and splitting and task allocation.
A human player will constantly - even during the course of a single turn - reevaluate the size needed for individual fleets to accomplish their task.
He will balance fleet size to achieve both maximum overkill (to limit HP/ship loss) and navy distribution to retain the ability to react to sudden threats.
In other words, don´t pull all ships together to chase a tiny hull - only to be wide open to a sneak assault somewhere else.
The AI needs to learn to allocate its forces smartly - where forces are needed strategically in response to potential threats (frontline/valuable assets) but also tactically in response to known threats (enemy fleets and their potential damage).
I would say that this part of the AI amounts to 50% of it´s potential against the human player - the other 50% is BUILDING its might - this 50% is EMPLOYING it.
Both abilities are multiplication factors. If the AI is great at building a navy, but does not use it or sends all of it to defend an asteroid mine...its resultant ability is 0.
Degrees of freedom are the bane of 4X AI developers. It is probably possible to program such a thing, but making it performant on the intended hardware is much harder. This is one of the reasons why I've been imploring the developers to simplify the decision-making process. It's bad for player (micromanagement) and bad for the AI (too many decision points). Striking that balance of features vs. actually being a strategy game is really difficult.
I fully agree, this surely needs improvement. I know it's tricky, but I'm sure they can make the AI smarter in this regard. The AI had me against the ropes in my current game. Those dastardly Manti had a much bigger fleet than I had, but I managed to sneak my way out of it by killing it's beelining fleets and sniping some ships here and there. Now I'm in complete control, why I should have probably lost this game.
There are many great features available to you once you register, including:
Sign in or Create Account
