When to fire and when not to and the TAC AI

[Juardis]

There have been discussions on whether or not you intentionally target something or let the TAC AI target it. There have been discussions regarding hiding and ambushes. There have been discussions regarding SOPs. All that relates to my concern. But first, the background.

Zook team, hiding, behind a ridge. There are 6 Jadgpanzers on the other side of that ridge, 5 of which are out of LOS. The 6th was 150+ meters away. The chance to hit was 3%, the kill probability was rare. Now, I couldn't set an ambush because there wasn't an HQ unit nearby. And even if there was and HQ unit nearby, there were too many areas that the tanks could appear at. I could've left my zook team hiding, but then I lose the initiative if any of the tanks come over the ridgeline. So I unhid him wondering whether he would fire with only a 3% chance to hit and if it did, only a rare chance to kill. He did, he's dead. 3 shots he got off, none of them close, before he took an HE round in the kisser.

Now, do you guys/gals think that the TAC AI should fire with only a 3% chance to hit? Do you think the fire order should be given if there is only a rare chance to kill? Personally, I don't think so, but I can be swayed.

I would think that the hit % would have to be a minimum of something like 10% before giving the order to fire and only if the kill probability is OK or better. I know, this could be solved with SOPs, but in the absence of SOPs, we're left with code tweakage to decide when our units fire and when they don't. This could also be solved by keeping your units hidden and manually targetting the next turn, but then you lose the initiative and may miss a text book shot.

Any thoughts?

BTW, it was a veteran zook team not under duress.

------------------

Jeff Abbott

[Robert Mayer]

Jeez, that's tough. In a laboratory setting, 3% with rare probability of kill sounds uninviting. But in "real life," whatever that is (I've never fired a bazooka at a Jagdpanzer, nor have I been shot at to my knowledge!), I suppose the zook team might not anlyze things that closely.

But that brings up another question. If we, as players, can make decisions based on discrete bits of info like hit percentages and kill chances, shouldn't the AI do the same? In that case, if we as the human would most always eschew the shot, shouldn't the AI do the same? Admittedly, this is a pretty easy example: few if any humans would take that shot. But what about 10% to hit? 20? With slightly greater chances to kill?

Any trigger level the AI uses will be arbitrary and will tick someone off ; goodness knows I've yelled at the computer for firing off the last 'shreck round at a halftrack half way across the map before. But I'm not sure we could ever get a formula that would satisfy everyone...

...short of being able to set at least general engagement parameters for units; i.e., SOPs.....

[Kingfish]

Zook team, hiding, behind a ridge.... Now, I couldn't set an ambush because there wasn't an HQ unit nearby

AT teams can set their own ambushes.

------------------

"What do you know about surfing, major, you're from God damn New Jersey"

[joeski]

Yes AT teams can set there own abushes but usually nothing farther than 100m.

Juardis: What do you mean you would have lost the initiative? In that scenarnio those tanks would have toasted the zook no matter what. Unless you hide until they are 30m away and then you probally only get off 1 or 2 shots. I don't normally let the AI chose my AT's target.

------------------

"If you're in a war, instead of throwing a hand grenade at the enemy, throw one of those small pumpkins. Maybe it'll make everyone think how stupid war is, and while they are thinking, you can throw a real grenade at them." - Jack Handey

[Leonidas]

In this case, the decision is pretty complex, because it depends on what else is happening. I can think of a few situations in which I wouldn't mind my 'zook firing at long range. For example, suppose the enemy armor is in LOS of some AT guns that will be firing next turn, and maybe some of my armor will come over a nearby ridge during the early part of next turn. If there's a big armor battle brewing, I might as well throw in the 'zook too.

Like Joeski, I'm confused about the problem with hiding the 'zook. On defense, I hide everything, except for armor. Hiding is especially important for AT teams and guns, because they make such poor decisions on their own.

Hiding is the big advantage that the defender gets in CM to overcome the attacker's raw force advantage. If the defender doesn't hide until he can get off a devastating first shot, then he's just fighting an attrition battle that the attacker is likely to win.

[Juardis]

The ridge line was about 50m away. If the tank crested the ridge while my zook was hiding, my zook would not fire. If, in that same turn, that same tank continued on behind some trees, then my zook would have lost a golden opportunity to fire at close range. Thus, he lost the initiative that turn. Now granted, if that tank stopped in clear LOS, then I could easily unhide him next turn and plink away, but there is no gaurantee that I'd get that 2nd chance.

Yeah, I had forgotten about zooks setting ambushes, but like I said, there was too much ground that the enemy tanks could have appeared, and any appearance was well within my kill distance so I did not want to miss an opportunity.

As for what else is happening, I don't think the TAC AI cares. All the TAC AI sees is a target and a chance to kill it, however small and however remote. The ideal solution is to let me decide what hit percentage is high enough to give my zook team a fair shot, but that's prolly not gonna happen in the current engine.

As for what it should be, that's the purpose of the thread. I think it should at least be 10% chance with OK probability to kill. Now granted, even with NO chance to kill, a zook can still immobilize or gun damage a tank, but those chances are small and I for one would not want to rely upon that happening consistently.

[Leonidas]

I say let the player decide, to simulate the complex decisionmaking that the 'zook team leader would be making. Any single standard would be unrealistic and frustrating.

[jasoncawley@ameritech.net]

You don't *want* to set an AT team's ambush longer than 100 meters. You won't *hit* anything.

If you want to make the "open up" decision, then stay on hide until you know it is a decent target. If someone runs super-close, chances are the team will panic and fire anyway, even on "hide". If something stays in LOS, you un-hide manually. And in the meantime, *guess* which target is going to come close (just like the real guy would have to) and *aim* that way, with an ambush marker. Close, because you really want to fire at 50 yards anyway.

Yes, it is annoying that AT teams expend their ammo at silly to hit ranges. But the solution is in front of you - the "hide" and "ambush" commands.

[CMplayer]

Originally posted by Juardis:

Now, do you guys/gals think that the TAC AI should fire with only a 3% chance to hit? Do you think the fire order should be given if there is only a rare chance to kill? Personally, I don't think so, but I can be swayed.

No of course not. It should be possible

to order to zook team to hold fire

and remain hidden until a certain range.

This is basically like setting out an

ambush marker, except that in some

positions, like you point out, the

enemy armor might appear from several

different points.

HERE IS AN ELEGANT SOLUTION:

Why not let units target SEVERAL ambush

markers at the same time? This would

provide lots more flexibility, solve

a lot of the problems which otherwise

would require SOPs, and above all, do

it within the confines of the existing

game engine's stock features.

regards,

--Rett

[Juardis]

Originally posted by CMplayer:

HERE IS AN ELEGANT SOLUTION:

Why not let units target SEVERAL ambush

markers at the same time? This would

provide lots more flexibility, solve

a lot of the problems which otherwise

would require SOPs, and above all, do

it within the confines of the existing

game engine's stock features.

regards,

--Rett

Yes, that would work. It is similar to the ambush zone someone proposed earlier. I like it.

[Guest Michael emrys]

Originally posted by CMplayer:

HERE IS AN ELEGANT SOLUTION:

Why not let units target SEVERAL ambush

markers at the same time? This would

provide lots more flexibility, solve

a lot of the problems which otherwise

would require SOPs, and above all, do

it within the confines of the existing

game engine's stock features.

But is it actually all that elegant? A two man team can only watch so much acreage at a time. Heck, even a 10-man squad can only keep so many sectors under adequate observation at once. Therefore, if you permit them to set more than one ambush marker, they'd better pretty close to one another in azimuth.

Michael

[CMplayer]

Originally posted by Michael emrys:

But is it actually all that elegant? A two man team can only watch so much acreage at a time. Heck, even a 10-man squad can only keep so many sectors under adequate observation at once. Therefore, if you permit them to set more than one ambush marker, they'd better pretty close to one another in azimuth.

Michael

I disagree. It's not too hard to notice

a tank at about the 75-100 meters that

a zook's ambush marker would be at,

regardless of whether it takes a left around

the buildings, or a right around the

buildings.

In practice, in CM, ambush markers are

often used against units which are

already spotted, and known to be

moving towards the marker. The point

of laying out the ambush marker is

just to set the range at which the

unit opens fire. This is not the same

thing as waiting all day for some enemy

to maybe show up.

Suppose we imagine an equilateral triangle,

with sides 100 meters long. At one corner

is the hidden zook team. At the two other

corners are ambush markers. Between the

ambush markers is forest, and on the other

side of the forest is a tank. The angle

between the two ambush markers is 60

degrees. Do you think it is unreasonable

for the zook team to observe both points?

Suppose a zook team is in a long but

narrow stretch of forest with LOS out

of both sides into open ground. Would

it be unreasonable for the two man

team to observe tanks at 100 meters

in the open ground on both sides of

the strip of trees they are in? Then

you would have two ambush markers at

a full 180 degrees from each other.

Still it seems quite reasonable to me.

These are the kinds of practical situation

where I would like to be able to target

multiple ambush markers.

Especially if the alternatives are to

either: 1- simply unhide the AT team

and watch in horror as they open up

at 200 meters an manage to shoot off

two misses before getting cut to pieces,

or 2- target one ambush marker and have

the tank maybe go the 'other' way, which

they then ignore for the 10 seconds while

they have a flank shot at 80 meters.

regards,

--Rett

[This message has been edited by CMplayer (edited 02-16-2001).]

[CMplayer]

woops, replied instead of editing

[This message has been edited by CMplayer (edited 02-16-2001).]