The Bad(?) AI

[madgamer2]

I was wondering if the new Matrix versions of this great series will improve what I think is the worst feature in the series, the AI. It was so easy to win against the computer even if you gave the computer lots of advantages. So I was just wondering if the matrix versions will address this problem or am I alone in this feeling.

Madgamer

[berto]

I, too, am tired of beating up on the weak AI. Command Control (?) is unappealing, since (a) with it, I have the worst of both worlds (computer opponent AI, human AI "assistant"), and (b) it removes me too far from the low level action and details that make these ACW games so appealing. PBEM is not an option for me.

I'm going to give Hot Seat solitaire a try, with some faux Fog of War (e.g., pretend I don't know the enemy's full disposition and plan of attack/defense, etc.).

If improving the AI is a matter of giving it more computing time, I'd be perfectly willing to wait several minutes (up to five or more) to give the AI more time to "think" through each move.

Sigh. So many otherwise wonderful games fail so badly in the AI department.

[madgamer2]

Yeah I feel the same way. I play with the computer getting a big boost but I still win Gettysburg by the end of the second day. (SIGH) I think that today the game programmers can write a little better AI code, I am hoping so. I have no real reason to buy the new package as I can play them just fine on XP but will most likely get it anyway as I am hopping that internet play will be possible.

madgamer

[acwbuff63]

Do you play PBEM or CC? Do you use the advantage slide?

[berto]

ORIGINAL: acwbuff63

Do you play PBEM or CC? Do you use the advantage slide?

But doesn't the advantage slide just affect the computer AI's lethality? What about intelligent movement?

[acwbuff63]

True, but it will allow the computer to be more aggressive and a human to be more cautious.

[berto]

ORIGINAL: acwbuff63

True, but it will allow the computer to be more aggressive and a human to be more cautious.

For straight-ahead, frontal attacks, maybe.

I have witnessed many situations where the AI was inert, passive, not responding (at all, let alone intelligently) to my moves. If I adjust the slider to extreme AI advantage, will this really goad the AI into action, moving more aggressively? Or will it be guided by its normal, predictable, seemingly aimless movement routines? Be just as inert and passive, but have truer aim, bigger guns, greater fire rate, unlimited ammunition, untiring marksmen, etc.?

If I make an "unexpected" flanking attack or encirclement, I want to see the AI respond to it, not just continue looking forward, or aimlessly dance around a fixed defensive position.

Expecting the AI to conduct intelligent attacks, not to mention creative, unexpected ones--that's probably too much to expect.

Does anyone successfully, satisfyingly play these games by, mid-game (e.g., not being constrained by an option or mode(s) set at the start of the game only), at the appropriate times (to be determined by the human player), and in every possible combination:

--switching sides (human-controlled Union, AI Confederate for X1 turns; then vice versa for Y1 turns; then back again...) (for example, human-controlled Confederate, AI Union, Shiloh Day 1; vice versa Day 2)
--playing Hot Seat (for X2 turns, then not...) (for example, human-control both sides at crucial points in the battle)
--playing computer-control, both sides (for X3 turns, then not...) (for example, computer-control both sides through the night)
--playing total Command Control (for X4 turns), then turning it off, micromanaging instead (for Y4 turns), then back again...
--playing partial Command Control--human-control some sub-commands, computer-control other sub-commands--adjusting the level and extent of Comand Control on the fly and at whim throughout the course of the game (imagine an army commander moving around the battlefield and giving more or less attention to his current sector, being more or less oblivious to other sectors)
--playing turn-mode (for X5 turns), then phases-mode (for Y5 turns), then back again...
--any other playing mode(s) or option(s) not mentioned here

Hmm, I think I'll begin exploring this hyper-modal style of play.

Wouldn't it be neat for scenario designer notes to recommend doing this for X turns, doing that for Y turns, doing this other thing for Z turns, ... recommending a modal mix for the most satisfying game experience?

In the final analysis, when "playing" against the AI, I wonder if all that we can hope for is, not a real contest, rather just a tool to learn more about a particular battle or combat situation. If it's a true "game" (i.e., competition) that we want, maybe playing against a human (hot seat, PBEM, network) is the only viable option.

[berto]

Just started playing the Belmont scenario of HPS Shiloh in the hypermodal style and had the game crash on turn 2. [:(]

Oh, well...

[Luciano B]

I stronlgy suggest to fight PBEM battles ...it is the best way to play BG games. I'm sure that in the communities on the web there's the right opponent for each player.

[Richard III]

These Games have scripted AI`s. So do most, if not all new wargames developed in the past 10 years or so.

That means that the AI`s ability to respond to many different human player moves are limited by the game engine and the amount of time the Scenario designer puts into " scripting" the AI responses for each turn in each scenario.

Obviously, after a few turns, the possible " AI " responses needed to be input becomes a geometic, not arithmetic progression.

That`s the way they work.

That is why they have a limited re-play value, if all you want to do is beat up the AI.

Because you are able to " beat " the AI, as you so modestly state above, after playing a Scenario 3 or 4 or 5 times, does not make it " weak ".

It means you have observed the way the AI always responds to your move and you take advantage of that by making a different move in your next Game that the AI is not scripted to respond to.

Since you obviously have a lot of good ideas on Wargame design, here`s a thought.

Pick up any new Tiller Game design. They all include a very user friendly Editor/Scenario designer and Scenario design Manual, as does Glenn Saunders Pz.Campaign Site pages. Then modify a simple existing Scenario.

Who knows, you might even design a complex and effective " AI script " and have fun doing it......[8D]

[madgamer2]

Does the AI have to be scripted? Is this the only way to do it? Perhaps it is asking to much of game designers to to be able to have the computer react to human movement rather than having them carry out some set of scripted moves. I am not a expert on such things as this as you can tell. I am just wishing that the computer could give a human player a run for there money thats all.

Madgamer

[berto]

The AI may be scripted at the strategic level, but at the tactical level, it is algorithmically driven, I believe.  I infer this because of Commander Control, where you command "attack this" or "defend that," and the AI carries out your orders for you.  How can arbitrary command actions be scripted?
 
I have a very busy schedule and time commitments, which is one reason why I don't commit to PBEM.  (The last thing I need in my life is to feel guilty about not responding in a timely fashion to my opponent's moves.)  I am intrigued, however, by the suggestion to try my hand at modding and developing my own AI.  I am a programmer/sysadmin by trade, although not well versed in the art of computer AI.  I do like a good challenge, though, so maybe one of these days...

[acwbuff63]

In addition to the scripted A/I (or instead of), the HPS series has a dynamic A/I. No scripts. A flag system is used to influence moves. But it is limited in larger scns. Also. the new HPS Atlanta and HPS Chickamauga each has 5 or so scns that are specifically designed for fighting the A/I. The are titled "A/I challenge." I forgot the scns numbers.

[Ashantai]

A general system is used with the AI.
 
Believe me, the AI becomes suicidal if set to extreme attack and total force flags. It's...impressive to watch the AI's total aggression when that happens.
 
Set it through the scenario editor.

[berto]

I've never had the pleasure of toying with the scenario editor.

Is it possible to edit canned scenarios, are the standard, stock scenarios hard-coded, unmoddable, or is the scenario editor only good for creating new scenarios?

If we can edit the standard scenarios to our own liking, hmm..., this is definitely something I should look into.

[Ashantai]

You can edit existing scenarios freely in both TS and HPS games,

[SittingDuck]

ORIGINAL: Richard III
Because you are able to " beat " the AI, as you so modestly state above, after playing a Scenario 3 or 4 or 5 times, does not make it " weak ".

It means you have observed the way the AI always responds to your move and you take advantage of that by making a different move in your next Game that the AI is not scripted to respond to.

Dude, that is one HELL of a good point. It changes the way I perceive game AI.

[berto]

I invariably beat the AI on the first try, not the 3rd, 4th, or 5th times.

Anyway, I did briefly check the Scenario Editor, and printed and read the manual. Very interesting. Now if only I can find the time to begin experimenting with this...

[madgamer2]

Well thanks to all who posted about my question but I was hoping to find out if the AI in the Matrix release will be different/better. I think the art of writing AI code has gotten better now than when the game was first released. I was hoping for a comment from the developing team. I will buy it anyway but I would like to know if they are changing it.

Madgamer

[berto]

Yesterday, I was rereading Shelby Foote's account of the battle of Shiloh, particularly how the Rebels took several days to stumble into attack position (where, according to plan, it should have taken them no more than a day), how they almost lost the element of surprise, then how tangled and confused the whole affair developed (with Rebel soldiers, at a crucial point in the battle, stopping to gorge themselves on the abandoned Yankee breakfasts, just to give one example).

I then reflected on the many other battles that didn't go according to the commanding generals' best laid, and detailed, plans.

The more I think about it, the more I think that playing Commander Control is the way to go.  With the "help" of the bungling AI "assistant", it helps neutralize the advantages the otherwise unencumbered, unfettered player has over the AI.

BUT, I wish there were a convenient way--with Commander Control still active, without deactivating it first (and reactivating it later)--to browse the map, check the morale of the troops, their supply, etc, and many other details of the battle.  With Commander Control, I feel too distant from these sorts of details, details that help bring the simulation to life.

That is, with the Commander Control dialog box still up, I'd like the ability to browse the battlefield (and access all menus) as if I were playing in the normal manner--but without the ability to move or attack, functions that would still be under the direction of the Commander Control box settings.

I'd also like to see more Commander Control actions beyond weak and strong attack, weak and strong defend, and do nothing. Like, for instance, move to, forced march to, reconnoiter, etc., also more varied forms of attack and defense, somewhat like TOAW's--tactical reserve, strategic, reserve, etc.

Matrix, how doable would these modifications be?