As folks on this and other forums have griped, the AI is terrible at defending against units with moves greater than 2. One approach to solving this is boosting the AI -- but I suspect there is a very good reason for the 2-move limit, which is that each additional move becomes exponentially more complex. Further, even human players have complained about the impossibility of defending against 6-move naval units.
A realistic solution that would NOT require (significant) AI reprogramming is to enable some variant of "Interception" for non-air units. This would effectively recreate the concept of "Zone of Control," and for modern units it makes a lot of sense. For example, a 6-move destroyer could have a 3-tile "Interception" radius, so that spacing destroyers 6 tiles away from each other gives you at least some screen against invaders.
This concept makes most sense when applied to units normally reserved for counterattacking. A bomber should also have the option, like a fighter, to engage in "Interception" missions, but directed against land domain units rather than air domain units. (Whether, then, there should be counter-interceptions leads us down a dangerous path... )
I am not familiar with how interception behaves (presumably it's controlled at the C++ layer), but any initial thoughts on whether this would (a) be at least a partial solution to the problem identified; and (b) be workable?
A realistic solution that would NOT require (significant) AI reprogramming is to enable some variant of "Interception" for non-air units. This would effectively recreate the concept of "Zone of Control," and for modern units it makes a lot of sense. For example, a 6-move destroyer could have a 3-tile "Interception" radius, so that spacing destroyers 6 tiles away from each other gives you at least some screen against invaders.
This concept makes most sense when applied to units normally reserved for counterattacking. A bomber should also have the option, like a fighter, to engage in "Interception" missions, but directed against land domain units rather than air domain units. (Whether, then, there should be counter-interceptions leads us down a dangerous path... )
I am not familiar with how interception behaves (presumably it's controlled at the C++ layer), but any initial thoughts on whether this would (a) be at least a partial solution to the problem identified; and (b) be workable?