Scourge of War Boards

I assume from the NSD team forum threads I've read that the AI is pretty much finalized. Can one (or more) of the testers give us their perceptions of how the AI compares to the TC2M AI?

Amish John wrote:

I assume from the NSD team forum threads I've read that the AI is pretty much finalized. Can one (or more) of the testers give us their perceptions of how the AI compares to the TC2M AI?

Here's my perception. They are definitely more aggressive than TC2M. They will try and flank you any chance they get (just check out the AAR I posted a couple months ago). If their orders are to attack, they will do so without backing down. There have also been improvements to keep them from piling up on top of each other which was a common problem with TC2M. Another big improvement over TC2M; when a regiment has retreated, they are more likely to rally and come back to the fight. Many times I thought the battle was over, the enemy regroups and comes back for round 2 or 3 or 4..

I stumbled into news regarding this game at the Wargamer forums. I dug out my copy of TC2M to familiarize myself again with the system. I presume Scourge of War will be a similar. I have to admit that I haven't had a chance to read up on it yet. I just created my account recently at the forum and saw this timely post.

The one aspect of TC2M that drove me crazy was the friendly AI. Whenever I would put my AI commanders into a stance, they would almost always behave without logic. More specifically, it seemed they would always march lone regiment smack into a flanked position (not a flanking position, mind you), march smack dab in front of another friendly regiment and block line of sight, or march off to nowhere special. I basically had to "take command" the entire time and mirco-manage.

Again, presuming I will be "taking command" or letting my AI commanders lead in this new game, will there be an improvement? Thanks!

What about AI in accordance to defense? In Tc2M they seemed to always willingly abandon the best positions such as fortified positions.

We have worked on the AI, most of our work has been the pathing, making sure that units do not walk through each other or through lines of fire. We have worked on the orders that are similar to what you saw in tc2m and we have added a few different types that we like. I'm not sure how much testing has gone into these yet, but it will be an improvement no matter how it goes.

Please remember that this AI was never meant to be some sort of RTS, meaning that units were never supposed to 100% follow your orders. It was this that caused us to realize the need for the tc button. They are supposed to act based on your orders and their own experiences. But I agree that it could be better.

Thank you for the insight. I certainly didn't mind when the AI commander would take its own initiative. It was more the little things like you mention...crossing the line of sight of other friendlies, walking directly into being flanked or wondering around. I also echo the other poster regarding leaving high ground or positions behind a wall for no apparent reason.

That all said, I liked the game enough that I didn't mind the micro-managing too much. I basically took command until the game was basically decided, then I would let the AI have its fun. However, it was certainly challenging on the larger maps and unit scale. Therefore, I pretty much stuck to the smaller scenarios which sort of reduced my overall satisfaction.

I believe that AI is something that we will work on forever. I still get bugs and it's one of those things that we discuss before we fix. The implications are far reaching.

For some time I've felt the need to defend the TC2M AI, just a little bit.

On page 63, of the TC2M Game manual, it reads, quote......."Each commander in the game has many factors that influence his decisions on the battlefield. As such, your subordinate commanders and units will not always do what you want them to do. It is possible that while your command is being pounded, you may see reinforcements arrive , but not engage the enemy. This was an intentional design decision to simulate what actually occured during Civil War battles. It is the representation of the possibility of delayed orders, confusing orders, or no orders at all. It also represents the timidity or cowardice of a commander, or his limited awareness of what is actually going on around him.
The AI considers each commander's personality and his current orders(stance). Based on these various elements, there is a chance that a commander will deem an attacking force too powerful for his men to handle and will choose to stand fast-waiting for the enemy to come to him. In these cases, the commander will not commit his men until he is attacked or an enemy comes to within range of his rifle-muskets. This adds greatly to re-playability of TC2M Senarios-commanders may react differently each time you play."

Now I'm not saying this applies to all the situations we experience with the AI. And I'm certainly not suggesting that some of the criticisms of the AI aren't warrented, because I share many of the complaints.
But I know from personal experience, some of my frustrations with the game, has a bit to do with the in-game camera view. I get so used to keeping the in-game camera view as high as possible, and from that view point, I'll see one of my brigades marching into a position that'll get them flanked, and I'm yelling at them ..."what are you doing!!", then lower the view to ground level, and realize that the brigade simply couldn't see the danger due to terrain.
And the enemy AI has continued to suprise me on many occasions, as to how clever it can be.

Again, please don't misunderstand me, I'm not trying to play the devils advocate here. But I think at least 15 to 20% of the time, what seems to be faults with the AI, may be our on misperception. Well, at least for me anyway.

And multi-player is bound to give us levels of game satisfaction that the AI could never do for us. Mainly, act like a human.

Good point Ephrum and I'll never change that. This is not an RTS, it's a simulation. So even when you tell someone to hold, they might not. That's the way we designed it and that's the way we like it. Sure it hurts us on reviews because people don't understand that unless they do the research.

The thing that I feel needs to be improved is that I want a greater chance that the officer holds. Right now I feel they leave too easily. It's improved, but needs to even be better. But it's something that's very hard to test and will be something we work on for a long time until we get it right.

During a defensive order or hold order, will the officer find good defensive or holding terrain and keep it's regiments there (behind a stone wall, etc). One thing that I hate about the AI is they always abandon a perfectly fine stone wall for some reason when an enemy gets close, then moves forward into a field to attack!