JTCS Tactical AI (Full Version)

All Forums >> [Current Games From Matrix.] >> [World War II] >> John Tiller's Campaign Series



Message

umbro -> JTCS Tactical AI (9/30/2008 12:34:59 AM)

I recently jumped back on to the CS bandwagon with JTCS and started refurbishing some old scenarios. While testing these I was surprised to find that the AI does not seem to have been touched at all since the old days. I was further surprised to find the number of players that regularly play against the AI. So to provide a better game for these folks in the future I was thinking that we could run a thread that would note sub-optimal AI play that could be easily fixed.

Some examples might be:

1. Never move leaders without being stacked with a combat unit.
2. Never leave leaders unstacked (if, say, their accompanying units is destroyed).
3. Never leave HQs in LOS.
4. Do not move HQs within the command range of their subordinate units unless the enemy gets LOS.
5. Never assault a hex that has undisrupted units in it.
6. Do not shoot at units that you cannot damage and cannot damage you if there are other targets in range.
7. Never stack more than 12SPs in a hex.

Some counter-examples (primarily ones that would be hard to fix)

1. Play more like Guderian would.
2. Use combined arms effectively.

umbro



kool_kat -> RE: JTCS Tactical AI (10/2/2008 9:00:34 PM)

Also the AI tends to exhaust all air missions in the early stages of the game - regardless of appropriate ground targets.

Case in point: I recently played a RS scenario in which the AI had 15 air missions. Sure enough; starting on turn #1, the AI sent 3-5 air missions my way each turn - hitting any unit (combat, non-combat) in clear terrain. By turn #5, AI air mission capability was exhausted... and the scenario had 21 turns. Rest of the game, I could operate with no fear of a "hit" from the sky.




Warren -> RE: JTCS Tactical AI (10/3/2008 9:43:21 AM)


quote:

ORIGINAL: umbro

I recently jumped back on to the CS bandwagon with JTCS and started refurbishing some old scenarios. While testing these I was surprised to find that the AI does not seem to have been touched at all since the old days. I was further surprised to find the number of players that regularly play against the AI. So to provide a better game for these folks in the future I was thinking that we could run a thread that would note sub-optimal AI play that could be easily fixed.

Some examples might be:

1. Never move leaders without being stacked with a combat unit.
2. Never leave leaders unstacked (if, say, their accompanying units is destroyed).
3. Never leave HQs in LOS.
4. Do not move HQs within the command range of their subordinate units unless the enemy gets LOS.
5. Never assault a hex that has undisrupted units in it.
6. Do not shoot at units that you cannot damage and cannot damage you if there are other targets in range.
7. Never stack more than 12SPs in a hex.

Some counter-examples (primarily ones that would be hard to fix)

1. Play more like Guderian would.
2. Use combined arms effectively.

umbro


It seems at this point only the PBEM crowd is being catered to, or they are not capable of fixing it. [:(]



e_barkmann -> RE: JTCS Tactical AI (10/3/2008 11:41:52 AM)

quote:

It seems at this point only the PBEM crowd is being catered to,


and considering the PBEM crowd kept the game alive long after many others abandoned it, I think that's fair enough.

Any anyway, if you read the forums you would know that Jason has stated that 'Hal' will be looked at in the future.

cheers.



kool_kat -> RE: JTCS Tactical AI (10/3/2008 11:53:59 AM)


quote:

ORIGINAL: Warren

It seems at this point only the PBEM crowd is being catered to, or they are not capable of fixing it. [:(]


CS has always been at its best with human vs. human play. I believe that this "mode" is the most challenging, exciting and nerve racking CS! [:)]

The original TalonSoft folks; over 10 years ago, were well aware of the limited AI capabilities. Quotes from the original EF Playtester Notes:

"The A/I will sometimes stack up in Objective and fortification hexes to the exclusion of maintaining a good line around it..."

"When the A/I is defending it will tend to counterattack Objective hexes you control."

"The A/I will frequently expend Smoke and Air Attacks early in a scenario..."

"It [A/I] has been known to move HQ's too far forward..."

So; in order to have a more satisfying CS experience, either PBeM (the best way IMHO) or play with the A/I as defender. The A/I tends to make fewer stupid mistakes as defender, but it is still not brilliant.

I tend to get rather bored playing the A/I, so the majority of my CS games are PBeM.




countblue -> RE: JTCS Tactical AI (10/3/2008 8:42:32 PM)

I do my own scenarios most of them long and on big maps.
I usually give the AI at minimum twice the quantity of troops I have and sometimes more.
This gives HAL a bit of a chance.
1.04 seems also to improve the overall performance of HAL slightly.

CB




umbro -> RE: JTCS Tactical AI (10/3/2008 9:01:36 PM)

Mike:

Personally I never played against the AI as it was so flawed. I also have issues with PBEM as a venue. IMHO the only way to play CS is H2H online, but I am strange that way.

However, Jason did an informal poll on the site and ~50 responses showed that folks play against the AI 80% of the time, 20% PBEM and only 1% H2H online.

By generating this thread I was hoping to create an easy fix list that would make the AI less brain-dead without having to go to the expense of a complete reconstruction.

Jonathan



Jason Petho -> RE: JTCS Tactical AI (10/3/2008 9:10:05 PM)

HAL has a tendancy to try to recapture the first Victory Location taken regardless of the overall situation, especially in DCG play.

Jason Petho



Borst50 -> RE: JTCS Tactical AI (10/3/2008 9:31:17 PM)

Really??? Hmm...I always found that HAL seemed to try to recapture the HIGHEST VP hex regardless of the tactical situation...not the first....interesting.



Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
0.796875