RE: Proposal of improvements list
Posted: Sat Oct 28, 2006 11:53 pm
Tim
I think that one thing that gets overlooked on AI's any more is that Scripting is not entirely bad. Especialy in certain situations I recall several times when HnM AI had a bunch of its troops in close proxcimity to each other all they would do is mill about countermarching in the same area disrupting the heck out of themselves. It seems like an AI that reconizes such situtations and forces groupes of battalions to adopt some sort of higher formation such as a line oriented toward the enemy, maybe splinting the difference between closest foe and most numerous would be better than standing around and getting slaughtered after all are disrupted.
By the same token Grand tactical scripts need not be bad if, the scenario designer can can generate several of them. That then are selected either random or by a probability scale of occurence. These could even be layered by events. Such as
Script 1a
apply game start
Army refuse left flank advance
script 1b
Apply if hostiles on left = less than xnumber, on turn xnumber +or- xnumber (or %chance or random number)
advance left flank
Potentially there could be a large number of possible starting scripts each with a % or random possiblity of being the one the AI uses initialy. With a potentialy large number of event or random or % dependent subscripts for each starting script.
one of the primary questions of course is at what level the script starts and leaves off on ther AIs behavior. Since HnM armies are already diveded in to commands under a particular officer this is maybe the place to make that cut.
There were a lot of games that used path following scripts. The major problem with wich, is that units tend to clump, or if they hold lines they do not react to flank threats. Or under or over value flank threats. Clearly a major problem is projecting possible flank threats. A failing shared with many human oponents. Implmenting super formations (that is above the unit level) could go a long way toward fixing this problem. EG a command of 8 battalions, if designated to do so keeps a foward facing line of 4 battalions while staggering back the remaning 4 on wich ever flank the bias is set for.
I have never played any HnM morph PBEM or Hot seat (The turn phasing format kills the game for PBEM for me)but did a while back play a Lot of East Front and West Front Pbem and Hot seat. After I had picked up much of the technical part of the way those games work. I noticed quite quickly that many Human oponents that were otherwise much better at playing the games than I was, had serious flank security problems. I note this also playing minatures or board games in wich deployment has a fog of war element. One way to make Make AIs stronger, it follows might be to emplement order action lags. In other words when the player issues orders to units the unit action on the orders is not instantaneous. Probably talking about a whole diferent engine there. But the advantage is that it gives a scripted AI a better change to shove a battle plan down the Human players throat. Even in Human to Human play the guy that has a plan vs the guy that doesent is at an advantge.
Now Mostly off of AI discussion another thing that can greatly add to any games fun level is some type of free deployment within some specified area of the map. The way that relates to AI is that given multiple AI scripts available to a scenario designer the AI deployment need not be the same from game to game in the same scenario.
I think that one thing that gets overlooked on AI's any more is that Scripting is not entirely bad. Especialy in certain situations I recall several times when HnM AI had a bunch of its troops in close proxcimity to each other all they would do is mill about countermarching in the same area disrupting the heck out of themselves. It seems like an AI that reconizes such situtations and forces groupes of battalions to adopt some sort of higher formation such as a line oriented toward the enemy, maybe splinting the difference between closest foe and most numerous would be better than standing around and getting slaughtered after all are disrupted.
By the same token Grand tactical scripts need not be bad if, the scenario designer can can generate several of them. That then are selected either random or by a probability scale of occurence. These could even be layered by events. Such as
Script 1a
apply game start
Army refuse left flank advance
script 1b
Apply if hostiles on left = less than xnumber, on turn xnumber +or- xnumber (or %chance or random number)
advance left flank
Potentially there could be a large number of possible starting scripts each with a % or random possiblity of being the one the AI uses initialy. With a potentialy large number of event or random or % dependent subscripts for each starting script.
one of the primary questions of course is at what level the script starts and leaves off on ther AIs behavior. Since HnM armies are already diveded in to commands under a particular officer this is maybe the place to make that cut.
There were a lot of games that used path following scripts. The major problem with wich, is that units tend to clump, or if they hold lines they do not react to flank threats. Or under or over value flank threats. Clearly a major problem is projecting possible flank threats. A failing shared with many human oponents. Implmenting super formations (that is above the unit level) could go a long way toward fixing this problem. EG a command of 8 battalions, if designated to do so keeps a foward facing line of 4 battalions while staggering back the remaning 4 on wich ever flank the bias is set for.
I have never played any HnM morph PBEM or Hot seat (The turn phasing format kills the game for PBEM for me)but did a while back play a Lot of East Front and West Front Pbem and Hot seat. After I had picked up much of the technical part of the way those games work. I noticed quite quickly that many Human oponents that were otherwise much better at playing the games than I was, had serious flank security problems. I note this also playing minatures or board games in wich deployment has a fog of war element. One way to make Make AIs stronger, it follows might be to emplement order action lags. In other words when the player issues orders to units the unit action on the orders is not instantaneous. Probably talking about a whole diferent engine there. But the advantage is that it gives a scripted AI a better change to shove a battle plan down the Human players throat. Even in Human to Human play the guy that has a plan vs the guy that doesent is at an advantge.
Now Mostly off of AI discussion another thing that can greatly add to any games fun level is some type of free deployment within some specified area of the map. The way that relates to AI is that given multiple AI scripts available to a scenario designer the AI deployment need not be the same from game to game in the same scenario.