Page 1 of 1

Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 3:03 am
by jchastain
As I have discussed elsewhere, after only a few days I am already able to routinely thrash the AI in detailed combat. Here are a few suggestions.

The single biggest issue I have with the AI is that it doesn't seem to understand the difference between offensive and defensive posturing. The game design itself is smart in that the attacker loses morale each and every turn - time isn't on their side. And therefore while a defender can afford to sit back and wait (and only improve their circumstances doing so) the attacker must take the initiative and strike. Great design, but the AI doesn't seem to understand that distinction. Whether on offense or defense, they throw themselves at their enemy. Unless we start within melee range at the onset, I can always find favorable terrain and establish a good defensive position knowing they will come at me.

The AI should know when it is the defender and should initially establish itself as such and force me to blunt my armies against their lines. I know that will elongate detailed battles a bit, but it makes a huge difference in their intelligence level. It also means they will have to be much more effective at responding to flank attacks and shifting their positions to compensate. That is a rather big addition to the AI, but it would make a huge difference. And then, once the attacker is disrupted beyond a certain threshold, they might maintain their defensive positions or they might pursue (which could then use the current offensive AI routines).

Beyond that big request, I would say that the artillery AI is actually decent. It manages range well and makes any charging attacker pay a huge price for getting too close. Even against the AI, I have a healthy does of respect for artillery. The cavalry is hit and miss. While they delight in finding lone units and make them pay a heavy price for being out of formation (as they should), at other times they are completely wasted charging lines where they will accomplish very little. Variety is good so that we don't know what to expect, but I would reduce the frequency that they just ride straight into fixed positions. Perhaps as importantly, I would delay them doing so until the infantry is a bit closer to take more advantage of the ensuing formation change.

The one area where the AI is overly aggressive in attacking detached units is in its attacks on garrisons. They waste a huge amount of effort charging towers that have literally no impact upon a battle. I suppose this is an attempt to provide a defensive bonus, but I believe the discussion above describes a far more effective manner for doing so. I'd have them bypass forts unless they are in close proximity to the main army or have visibility into an area where it is absolutely critical that it remain concealed (and that's a very, very short list of places).

The AI is also a bit too generous in simply charging the middle of my line. While they may want to engage the main line to hold a large number of forces in place, they should collect enough forces to try to overpower a flank at the same time rather than just all marching right into the killing zone. Those attacking the main line might also wish to abandon the column formation just a tad earlier - all too often they are parading into the range of my artillery despite the barrage that was just giving to the last 2 units to try the same thing. Which leads me to add that I get the impression that units don't learn from the experiences of their peers. If they outnumber me and are trying to overwhelm the lines, fine - it isn't my favorite plan but it is a plan. But running single file into the killing fields just isn't good generaling.

As I have stated repeatedly, the game is a lot of fun just as it is. Every battle is different and I do enjoy the detailed combat immensely. But I think continued evolution of the AI for this aspect of the game could make it even better.

RE: Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 3:17 am
by sol_invictus
I've only played a few small battles for a few turns just to get familiar with the mechanics so I am not familiar with how the AI performs, but your suggeastions sound both reasonable and I would think doable in a patch. I am certain that when the patch is released, it will not only fix the few little bugs but also include some AI tweaks.

RE: Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 3:24 am
by Ralegh
Great ideas - right on!

Additionally, while the AI occasionally uses its fortified garrisons in combat, it should do so HEAPS more - I certainly use mine a lot more than it does.

The AI should also place more attention to screening its supply caissons.
Broken supply caissons should not affect the morale of their side.

I can't wait to play some of you guys in TCP/IP games, so we can try detailed combat against a human!

RE: Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 12:52 pm
by Sonny
ORIGINAL: Ralegh

Great ideas - right on!

Additionally, while the AI occasionally uses its fortified garrisons in combat, it should do so HEAPS more - I certainly use mine a lot more than it does.

The AI should also place more attention to screening its supply caissons.
Broken supply caissons should not affect the morale of their side.

I can't wait to play some of you guys in TCP/IP games, so we can try detailed combat against a human!


??? You never played TCP/IP while beta testing??

RE: Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 1:05 pm
by Ralegh
I am in Australia. People in the same time zones as each other did all the TCP/IP testing. Sob. I woulda, but for some reason they are all in bed when I was playing. My PBEM tests didn't even get very far. But now we can PLAY the game, rather than trying to make it break...

RE: Detailed Combat AI Suggestions

Posted: Mon Jul 04, 2005 4:10 pm
by jchastain
The strategic game AI is generally pretty good. I did see one oddity today though. With a war raging with France, Austria had an opportunity to choose their first military advance. And the AI selected... Naval Repairs. That seemed an odd choice for a nation with a huge army and practically no navy in the midst of a land based conflict. Perhaps the AI could evaluate the number of land units a nation has versus the number of naval vessels and use that to influence the likelihood of an army versus naval selection?