Fleet threat logic

[mordachai]

I know that this is a can of worms - that no solution will work adequately for all situations - that, inherently, in war, there are literally impossible situations where you have to simply sacrifice some objectives for others - and we ain't all gonna ever agree on which get sacrificed, and which take priority.

That said - there is clearly a great deal of room for improvement...

One scenario I see constantly is that I have a defense fleet IN THE SAME SYSTEM as a raider fleet from an enemy empire I'm at open war with...

Does the defense fleet (automated) let go of worrying about some space monsters (which are innocuous - just a nuisance to be dealt with "eventually"), to deal with the immediate and far more pressing threat of the enemy fleet? No. never.

It continues to attack that pointless monster while its colony or bases are being attacked, destroyed, invaded, whatever. Because "one order is all you get" or something... the logic doesn't seem to support the concept of priorities once a mission is taken?

I think any and every human player will agree that the enemy fleet is a vastly higher priority, unless the defense fleet is just committing suicide or similar extraneous reasons why it literally CANNOT help defend the local targets from that fleet.

I can see having human-only type orders like "defend X at all costs!" which prohibits the fleet from abandoning its post despite other obvious threats...

But the normal case - and the automation case - really ought to be "defend X but also defend the local system if available to do so"

Or more generally - when we get the alert that a fleet is incoming - any defense fleets (or attack fleets) near enough to respond, that aren't UNABLE to do so, should do so (with some limit to only responding enough to deal with the threat - not causing every fleet in the empire to be pulled in every direction at once).

It's a super hard set of code to write, for sure - I don't envy whomever is writing it. Maybe even a case for machine learning or fuzzy logic -- but even hard-coding some degree of threat-assessment priority-list seems doable (albeit I admit I'm not in that code, I don't know how many devils are in these details)...