Maybe adding AA modifiers to doctrines and a sizeable boost to the effect of fire control and radar could be increased to better simulate reality?
I'm not sure that increasing the amount of AA you get, for a given investment of IC, fixes this. Or at least, it doesn't do it without significant side effects. Raising the raw AA numbers also increases the number of planes shot down. To an extent, some of this is desirable, but in practice the sheer absurdity of the imbalance of IC when bombing ships requires something more, else you have to shoot down
so many planes that people start wondering how this would even be possible with WW2 tech. There's a real danger of breaking the
appearance of simulation for some people.
If the problems with AA-based damage reduction are that a trivial investment in AA is too effective, and that impactful amounts of AA are functionally and/or
completely unobtainable, then a solution could be found entirely in the defines.
1) Reduce the coefficient, ANTI_AIR_MULT_ON_INCOMING_AIR_DAMAGE
2) Increase the exponent, ANTI_AIR_POW_ON_INCOMING_AIR_DAMAGE
3) Raise the damage reduction cap, MAX_ANTI_AIR_REDUCTION_EFFECT_ON_INCOMING_AIR_DAMAGE
This is what it looks like when the coefficient is reduced to 0.1, the exponent increased to 0.5 (a square-root function), and the cap raised to 80%. With these particular values, the AA damage reduction cap of 80% requires a weighted AA sum of 64. A single ship really can put a dent in this, for one. That not-so-hypothetical heavy ship with 3x DP secondaries and AA3 would provide around 15 on its own, requiring only a total of about 200 from the rest of the fleet. A modern(ized) strike force should be able to muster this without resorting to gimmick ships, even without other game changes. (Admittedly, dual-purpose secondaries and lights are probably more expensive than they should be in vanilla, but it can be done.) 80% is a 5:1 reduction factor, which would make a deep impact on the currently outrageous returns of naval bombers. Of course, not every task force is going to have nearly this much. There are still patrols, surface raiders, or underdeveloped/depleted strike forces that wouldn't get nearly this much benefit, and submarines still have zero AA.
There's still some other things that could be experimented with. For example, screening efficiency (and/or capital ship screening efficiency for carriers) could provide an AA bonus. This would, of course, mostly only benefit proper battle fleets (and/or carrier escorts for the floating airfield concept.) As mentioned previously, any boost to AA values also results in more planes being shot down, so it should be done with care.
If I had access to the game code, I would change the whole form of the AA damage reduction calculation to one approaching 100% asymptotically, e.g.
y = 1 - exp(-kx). To me, it makes sense that the value of a given percentage point of damage reduction increases asymptotically in this way, so that's the function I would have used, or something very much like it.
P.S. For those not familiar with log-log plots, it's no accident these functions look like lines here. The game's calculation for AA damage reduction is a power function, and power functions look like lines on log-log plots. Power functions have the form:
y = b * x^m. If you apply a logarithm to both sides, that translates it to
log(y) = m * log(x) + log(b). When you plot log-values, as on a log-log plot, that gives it the appearance of a line. The logic of lines also mostly applies. To raise or lower the function, increase or decrease the "intercept," which was the coefficient of the power function. To change its slope, alter the exponent.
