As for semi-trivial battles (no chit choice, no reinforment and no guard commitment possible) those would require only steps 1, 2, 3 and 4. Steps 1 and 2 to setup the "handshake" and predetermination of dice that avoids reloading and steps 3 and 4 to show the die rolls and take the loses. So it would be 3 mails. Sure, its slower than just letting the attacker run the battle alone - but not exactly impossibly slow either.
I was talking about the battles that weren't officially trivial, but the computer decides they are close enough, and picks one's chits for one. If it picks chits for both sides, that would be fine (except, not secure), but it frequently allows one side to pick a chit while choosing one for the other side. But, those battles could then be replayed.
Personally, I think the game should completely ditch these kinds of battles. Battles where the computer picks both chits (or, uses standing orders) would be OK. But, battles where only one side picks are notoriously insecure.
Still, even battles where the computer picks both sides have a problem: On a separate instace of the game, the person can play it out. Then, reload everything to before the land phase.
I think your scenario requires one other thing to cover all "computer-to-computer" or "computer-to-player" battles: There has to be the same kind of exchange occurring (battle file, for example) between the land movement phase and the land combat phase. This file wouldn't contain any battle information, but WOULD tell the other guy which battles are about to occur (thus preventing a reload from working).
If I'm not mistaken, the model you use throughout your post can be applied to this "extra" file exactly the same way (except, no dice and no encryption needed).
The current (game) model has the person shipping both the movement phase and the combat phase at the same time, after everything is done. It really needs to be required earlier. It doesn't require a response, because the non-phasing player would be seeing all of the upcoming battles before any were fought.
However, even without code change, we can do something like what you suggest: Make a house rule that you must turn in your land phase file before any battles are initiated. It wouldn't help when all battles are done on the same system, but it would help if any one of those battles required battle files be sent. If the phasing player did not send out his land movement file, and then all of a sudden is sending out a battle file, well, there's a problem.
Now, a dedicated cheater can get around part of this by simply playing until it asks to send a battle file. Then, he knows he has to use "safer" choices. But, if nearly all battles (except true "trivial" combats) required battle file intervention (by your model or Marshall's), then in almost all cases, it would not be profitable to try this tactic.
What do you think?
At LAST! The greatest campaign board game of all time is finally available for the PC. Can my old heart stand the strain?