A couple of random thoughts:
TJ, you were doing much better, and I almost found myself taking your input seriously. Then you had a relapse.
The pilot skill points are well made. I agree that a certain level of pilot exp should be necessary to fight the a/c to full stats. If the inexperienced pilot is in a totally superior a/c, he might be almost immune to attack, but conversely, the expert on the other side should be able to evade the attack as well. In summary, aces beat newbies, unless the newby is in a totally dominant a/c, in which case combat tends to inconclusive (rather than equal losses in any number).
I agree with some of the threads about the absolute magnitude of the losses on some of the examples. Thinking back to BoB type history, any squadron that takes say 3/12 losses in a single sortie is absolutely decimated, and this isn't that common. 1 loss per combat is far more common. This leads me on to a train of thought, that I am typing as I think (so forgive me if it rambles).
In WW2 combat, it was found that the a/c were much too fast to operate more than 1 (RAF) sqd as a coordinated formation in fighter combat once contact was made. The really hot late war a/c more normally could not hope to coordinate more than 4 a/c and sometimes 2. Thus you did not get 50 vs 50 (literally) fights. Even if they turn up together, you get a large number of 2, 4, or just possibly 6 or 8 vs similar numbers fights. Obviously these would interact, but only fleetingly.
Any formation of a/c (assuming trained pilots and good leadership here) would attempt to engage at an advantage, but once committed, the small fights break out. A brief period of combat (I don't know how long - 30 secs, 1 or 2 mins at most), and then the section would find itself in empty sky, and try to regroup. If it hadn't taken losses, it might re-engage any enemy it felt it could (at an advantage), but might not. If the section had lost someone, it would almost certainly not reengage, but either head for home (sweep), or try to catch up with the bombers (escort). I guess the CAP types would reestablish comms with the ground controller, and either call it a day, or seek vectors if good moral etc.
What the above means is that there ought to be a sort of break point on how the combat goes depending on losses. I would suggest a script something like:
Decide how the total formations meet (i.e. relative heights, missions etc), to determine the starting advantage. Split the combat into sections of 4 planes vs 4(if equal numbers) or 2 if seriously outnumbered, and evaluate the combat. The outline of the current combat system didn't seem to bad as a concept to me for this bit. Each section gets one 'combat round' to attempt to nail the enemy. Put severe limits on the chance of any one plane attacking more than 1 enemy. Each section is assumed to break away temporarily after this, and takes a moral check. If it has lost anyone (or taken significant damage), it needs a very good check to continue, but otherwise a more normal check. Evaluate how many sections are still interested in combat and have another round. Reevaluate the starting advantage, and repeat as above.
Some sections (if one side has a numerical advantage) would have a shortage of targets (all already engaged). I would give these 'spare' sections a chance of a free attack against an enemy section, at reduced effectiveness (you can't have 8 a/c all trying to shoot the same one - they get in the way), but with a much reduced risk of being shot back (the enemy is busy!). Alternatively, these spare sections get to have a chance to break through and hit the bombers if any are present.
What this would mean, is that it would be very difffcult to get high losses regardless of numbers involved, because any section with a loss, or a damaged a/c will most likely not engage again.
It would address the problem of one pilot getting too many kills in one mission, because he could only get one per mini round - so to get more needs multiple moral passes.
The starting advantage for each subsequent round would be modified by how the previous round went - so if one side gets creamed, it is likely to be at a disadvantage next round. This advantage should also be modified by leadership, and exp etc, so the in the case of the really outclassed experts against the newbies, they might get the advantage, and prevent the newbies getting effective attacks, but chose not to engage themselves, getting the advantage on the next round, or the escape etc. (fulfilling my theory for the losses in such a battle.
There is also a further feature of air combat in this game, that has occurred to me:
Each turn is 6 day hours. Thus, a single 50 a side battle (for example) may be 50 a/c in one formation (coordinated smaller formations), or smaller ones turning up for 6 hours ( or anything in between). The difference to an outnumbered CAP is huge. Similarly, the CAP is potentially fragmented, or coordinated.
I wonder if the a-a model includes details such as I have done above, or is a more 'purely mathematical' model?
Having said all of this, I might just end up with the same results as the current model!