Post about artificial intelligence for implementing various offensive maneuvers (envelopment, turning, infiltration, etc.).

I've just uploaded my post about artificial intelligence for implementing various offensive maneuvers (envelopment, turning, infiltration, etc.). http://general-staff.com/a-wargame-55-years-in-the-making-part-3/

The elements you've described in this discussion are missing in virtually all computer war games I've played. The result is a very easy win for any human player unless the program cheats by bumping up the stats of the computer's side.

However, I think the real test of competent AI would be with respect to the defending side. Although defense is generally easier than the attack, choosing a good defensive position or moving to a better defensive position while not completely abandoning the battle objective, if possible, would be a much harder for a computer to identify. Or knowing when the attack has run out of steam and it may be time to launch a counterattack on some part of the field is a notion that humans intuitively grasp but would be difficult for a computer to handle algorithmically. That would suggest using a learning AI.

It's said that Napoleon would quiz his junior staff as they traveled about defensive positions utilizing the terrain they were passing through. I agree with everything you said, defensive positions should be easier, but...

I am looking forward to working on defensive AI in the near future. Also, think of Sherman and Johnston and the approach to Atlanta. Johnston kept taking up superb defensive positions and Sherman kept outflanking him!