Brain-dead TacAI?

[womble]

...I think the arguments would tend to favor a tight arc as being better in most situations.

I think I'm agreeing with you, and GaJ, when I say 'most situations' doesn't include "being fired upon by an enemy which can hurt you". In the described situation, "self-preservation" was sufficiently urgent to initiate movement/retreat, yet not to fire back at the enemy which was (I gather) spotted and not very far outside the arc. You're not preserving any sort of ambush if you know you've been spotted (because there are rounds pinging off your turret) and doubly so if you leave your ambush position. Might as well engage, even if it's because you're backing off.

Now, on the other hand, as has already been mentioned, if it'd been a Sherman, say, or any other vehicle with a decent turret traverse speed, it might've gotten shots off before it felt it needed to retreat, and the same could well apply here too: fast traverse might've meant the Puma could fire while evading (and we'll leave aside the discussions about hit chances in that situation, eh? ).

[A Canadian Cat]

When I give a vehicle a covered arc command I typically make them huge 180 degree arcs, often covering the entire portion of the map in front of the vehicle.

What I do in situations like this is give a Target arc command that is a quarter of an elipse.The problem with covering a field with a pie shaped arc is too much of the field is not covered. So I do this: set the first point of the cover arc far away and to the right so that all but a small near corner of the field is not included then select the other target point much closer and 90 degrees ish to the left. thus creating a quarter of an elipse that covers most of the field.

I am writing this on my phone, if you want further clarification I can post a screen shot later.

[JonS]

I am writing this on my phone, if you want further clarification I can post a screen shot later.

Nope, I know exactly what you mean because I do this too.

[BletchleyGeek]

Quoting from the manual.

From the tutorial

Sometimes you will want your units to focus their fire in a specific area. The Cover Arc command was made for this purpose. A unit with a cover arc command will usu-ally attack only enemy units that are located in the designated area, unless it feels immediately threatened by an enemy

outside its arc.

Page 84, where the command is described in depth

The Target Arc command orders the unit to only fire at enemies within a cer-tain target area and/or range.

[...]

Any visible enemy units that are located inside this area, or that move into this area, will be fired upon. Any enemy units outside of this target arc will be

ignored (until self-preservation takes over and the Tactical AI decides to

override player orders; e.g. if an enemy unit suddenly pops up at extremely

short range).

[...]

This Command is also useful to keep a unit’s “attention” focused on a specific

part of the game map while it moves. If, for example, you want to keep a

close eye on a bunch of buildings (where you suspect enemy activity) while

driving down a road, you could assign a target arc to several units covering

this area. The target arc increases the chances that units will recognize and

engage an enemy threat within the target area quickly.

After placing an arc, the unit will rotate its main gun turret - if available - to

face the center of the designated target arc, to minimize acquisition delays

and maximize spotting abilities. Infantry units will shift their facing accord-ingly.

No mention of the Morale and Training effects on this, of course. But I do see where 76mm expectations might have originated.

[GreenAsJade]

If an AFV taking canon fire feels "immediately threatened", then it sounds to me like CAs are currently exactly what I'd want

I can certainly live with "doesn't continue to shoot at after it leaves the arc".

GaJ

[Battlefront.com]

To confirm what the Manual says about Covered Arcs and Spotting... there is a bonus for spotting things within an Arc. The size of the Arc matters, however, as a HUGE Arc is effectively no different than saying "pay attention everywhere!". And the bonus isn't enormous to begin with, but enough to give an edge sometimes. Edges are always a good thing even if the ultimate result isn't what you want to see. Nothing in war is guaranteed!

The speed of engaging, however, isn't different. A unit without a CA that spots a target right in front of it will engage just as quickly as one with a CA. Keep in mind that engaging can only happen after spotting! Which means, all else being equal, a target coming right in the middle of a CA will get engaged faster than if there was no CA. Though tons of circumstances can change that +/- for any one given circumstance (e.g. a non CA unit probably will spot something faster if the target is in the open than a CA unit in dense terrain).

The other advantage of a Covered Arc is having the unit optimally facing the area within the Arc. This can be mimicked without a Cover Arc by using Face, but depending on how each is used the CA may produce better results.

Units are reluctant to disregard their Arcs, because (as JonS pointed out) they are assigned for a reason that very well may be integrated into a larger plan. Self preservation, however, is something that may cause an override. As I said earlier, it's very difficult to tell what your friendly TacAI is thinking at any given moment, such as you seeing the incoming round from x direction but the virtual crew does not. It's also impossible to say what the AI controlled units are doing at all, obviously.,

To sum up our position, I'll simply quote black_prince:

As covered arcs currently exist, they are straightforward and predictable and allow the player to use more sophisticated tactics than would otherwise be possible, in other words 'it aint broke'.

While we never rule out a case to be made for a tweak, we do think that Arcs are behaving pretty much as they should be within the constraints and variable conditions we've discussed here. As such, I wouldn't expect Covered Arcs to be noticeably different any time soon.

Steve

[Vanir Ausf B]

Self preservation, however, is something that may cause an override.

I cannot say for certain about infantry, but vehicles will never do that.

[GreenAsJade]

Vanir: how do you know that they don't? I mean, how sure are you?

Steve: if AFV's never return fire for self preservation, is that "behaving as they should?"

It strikes me that AFV's do need to retaliate?

GaJ

[Vanir Ausf B]

I've done a lot of gunnery range testing of tanks in CMBN. And by a lot I mean many hundreds of turns of tests that typically have 10 tanks in 10 separated rows getting pummeled by something at around 100-500 meters away, usually another tank but sometimes snipers. When doing these tests the way I keep the target tanks from returning fire against their tormenters is to give them a covered arc (one that does not include the unit firing at them, obviously). It is 100% effective. They will just sit there and let the enemy unit shoot at them turn after turn. If they are penetrated and take casualties they may bail out, or they sometimes pop smoke and retreat, but I have never seen one disobey the CA and shoot back.

[GreenAsJade]

Self preservation, however, is something that may cause an override.

I wonder if this needs a tweak, then for AFVs?

GaJ

[GreenAsJade]

A light globe has just gone on for me about this. I haven't been able to understand why some of my infantry are so bad at spotting guys in the distance. Now I can see why: these inf have close covered arcs on!

At first glance, this seems like a real problem. I am not saying to these guys "concentrate your gaze at the field immediately in front of you", I am saying "don't open fire on things until they get close".

This are two very different things. It seems to me that the coupling of covered arc with spotting focus has put me in the position where I can't spot things well in the distance unless I'm willing to fire on those things!

I just discovered this because I have a whole field of inf looking out over an oppos approach. The only units without covered arcs are mortars - and only because I'm not in the habit of giving these guys covered arcs (they're not deployed, and seem to pose no thread of opening fire wrongly).

I couldn't for the life of me work out why none of my inf were seeing the oppo moving in the middle distance, yet the mortars could see them. Now I understand: and it feels "wrong".

Surely we need to be able to say to inf "don't open fire" without loosing the spotting ability? This "focussed spotting" effect is really quire pronouced: the guys with covered arcs really are not spotting the oppo when the oppo forces seem to be quite obvious.

To ask it a different way: how do you spot into the distance without risking opening fire?

GaJ

[womble]

A light globe has just gone on for me about this. I haven't been able to understand why some of my infantry are so bad at spotting guys in the distance. Now I can see why: these inf have close covered arcs on!

Steve, can you answer this? Is the CA spotting bonus only within the CA, or does the CA angle determine the direction?

I just discovered this because I have a whole field of inf looking out over an oppos approach. The only units without covered arcs are mortars - and only because I'm not in the habit of giving these guys covered arcs (they're not deployed, and seem to pose no thread of opening fire wrongly).

Have any of the other infantry units got binos in them? What range are we talking about here? I know I've had units spot things outside their covered arcs; does this mean they would have spotted sooner without 'em.

Steve, can you elucidate whether distance beyond the arc makes any difference to whether an infantry team will override their CA? I've had teams fire on MG nests that were about 10% outside their (circular) CA even though the MG hadn't fired on them yet.

I think Vanir's evidence that vehicles don't override their TAs for self-preservation purposes is quite convincing. Might there be cause for some change there?

[BletchleyGeek]

To confirm what the Manual says about Covered Arcs and Spotting... there is a bonus for spotting things within an Arc. The size of the Arc matters, however, as a HUGE Arc is effectively no different than saying "pay attention everywhere!". And the bonus isn't enormous to begin with, but enough to give an edge sometimes. Edges are always a good thing even if the ultimate result isn't what you want to see. Nothing in war is guaranteed!

The speed of engaging, however, isn't different. A unit without a CA that spots a target right in front of it will engage just as quickly as one with a CA. Keep in mind that engaging can only happen after spotting! Which means, all else being equal, a target coming right in the middle of a CA will get engaged faster than if there was no CA. Though tons of circumstances can change that +/- for any one given circumstance (e.g. a non CA unit probably will spot something faster if the target is in the open than a CA unit in dense terrain).

The other advantage of a Covered Arc is having the unit optimally facing the area within the Arc. This can be mimicked without a Cover Arc by using Face, but depending on how each is used the CA may produce better results.

Units are reluctant to disregard their Arcs, because (as JonS pointed out) they are assigned for a reason that very well may be integrated into a larger plan. Self preservation, however, is something that may cause an override. As I said earlier, it's very difficult to tell what your friendly TacAI is thinking at any given moment, such as you seeing the incoming round from x direction but the virtual crew does not. It's also impossible to say what the AI controlled units are doing at all, obviously.,

While we never rule out a case to be made for a tweak, we do think that Arcs are behaving pretty much as they should be within the constraints and variable conditions we've discussed here. As such, I wouldn't expect Covered Arcs to be noticeably different any time soon.

Thank you for the clarification, Steve.

One of the things that sold me into CMx2 were that units perception was modeled in a realistic way (tank crewmembers have to actually look into a given direction to spot something, and that might lead them to not spot something else). However, having them "forgetting" about their goal (they track to kill the target when they get a fire solution, which is mediated by crew experience, training and psychological status) when the target gets out of the arc is at odds with what one would expect from actual people. Therefore that people raise their hand and say "this AI is braindead".

[GreenAsJade]

There is an argument that says that the unit is _obeying its orders_ when this happens. They didn't forget, they just did what they are told.

There's also another argument that says whatever the case (of what people expect), that the current implementation is workable, and the best it can be.

I think it'd be better to focus on concrete questions like "should this current implementation be tweaked so that AFVs return fire when they're under dire threat?" than exploring why someone might say something stupid like "brain dead".

The current concrete questions are:

How do you spot into the distance without risking opening fire?

Is the CA spotting bonus only within the CA, or does the CA angle determine the direction?

Should AFVs return fire when they are under dire threat (because it seems that they don't)

GaJ

[BletchleyGeek]

Some people in the thread weren't understanding the OP, I think I did, and just offered my interpretation.

Is that really bad, GaJ?

The current concrete questions are:

How do you spot into the distance without risking opening fire?

Is the CA spotting bonus only within the CA, or does the CA angle determine the direction?

Should AFVs return fire when they are under dire threat (because it seems that they don't)

1) I don't understand what do you mean with this one.

2) From what I gather from playing the game, reading Steve answers and reading the manual there's no spotting "bonus" - i.e. there's no such a thing as a magical +3 to a "die roll". I think it depends mostly on how wide is the arc. My guess is the following: The "spotting" guys are scanning his environment, that is, rotate head to bearing 270º, look for some time, rotate head to bearing 290º, look for some time, etc Each of these actions or activities take some time to do (simulation ticks). With no arcs, in order to scan the whole 360º arc, he takes a number N of ticks (which might well dependant on training, morale, being under fire, etc.). With a restricted arc, say 90º, in ideal conditions (highest morale and training, not being under fire, etc.) it would take him N/4 ticks to complete the "scan". Or in other words, he would be able to fully scan the arc 4 times, increasing the chances of spotting anything accordingly.

3) Are we sure they don't ever? Steve is not going to post here the TacAI routine, but this would involve certainly how the unit in question is perceiving its environment. Does the unit know of other enemies - either by visual contact or sound contact - in the vicinity? Does it perceives these other as more important threats than the one we know he can spot? Does the unit believe it's going to get a shoot good enough to kill/supress the threat? Is it low on AP (or HE) ammo? Motivation surely also plays a role.

Some of the discussions we have on threads like this one contain a lot of useful info for new players (and certainly, not so useful discussions). I noticed recently that someone setup a CM:BN wiki

http://combatmission.wikia.com/wiki/Combat_Mission:_Battle_for_Normandy

perhaps we should start using it.

[YankeeDog]

Should AFVs return fire when they are under dire threat (because it seems that they don't)

GaJ

Under 1.01, AFVs will definitely ignore a cover arc and return fire against a "dire threat" at least sometimes, because I've had them do it.

Now, whether this needs to be tweaked at all, I couldn't say. I have only encountered this type of situation a few times personally, so I wouldn't say my experiences represent a reliable sample size.

[womble]

Under 1.01, AFVs will definitely ignore a cover arc and return fire against a "dire threat" at least sometimes, because I've had them do it.

And yet Vanir has had 100s of tanks being pounded for thousands of aggregate turns without returning fire while under CA restrictions. I'm not disbelieving' but

summat's squiffy here.

[Michael Emrys]

And yet Vanir has had 100s of tanks being pounded for thousands of aggregate turns without returning fire while under CA restrictions. I'm not disbelieving' but

summat's squiffy here.

Was Vanir testing within 1.01?

Michael

[aleader]

Under 1.01, AFVs will definitely ignore a cover arc and return fire against a "dire threat" at least sometimes, because I've had them do it.

Now, whether this needs to be tweaked at all, I couldn't say. I have only encountered this type of situation a few times personally, so I wouldn't say my experiences represent a reliable sample size.

As they obviously should. I don't understand this whole argument quite frankly. If I'm in Tiger and a Firefly starts firing on me, I don't care one bit what my 'orders' are, I'm firing back and\or getting into cover until I find what fired on me, unless I am completely brain-dead. If the game doesn't do that, it needs to be fixed, period.

If someone can find me an actual account in a real war (other than the suicidal nuts in Japan) where an AFV crew knew they were being fired on by an obvious threat (i.e. not just an MG or sniper) and they still just sat there because the army told them to, please share.

If it is a limitation of the game and can't be fixed, say so. Justifying this behaviour in a game that purports to be realistic makes one look foolish.

[c3k]

If someone can find me an actual account in a real war (other than the suicidal nuts in Japan) where an AFV crew knew they were being fired on by an obvious threat (i.e. not just an MG or sniper) and they still just sat there because the army told them to, please share.

If you were in a Tiger, how would you know a Firefly was shooting at you? The sound of the shell whooshing past? The sight of one looking at you? Or the impact of the shell? Is he shooting at YOU, or your wingman? Are you jussst about to squeeze a round off at him? Or, is that just a Sherman?

It's a tough call...

Having said all that, yes, there are definite situations where there is a known threat and the unit should abandon its covered arc and possibly run away. There are conflicting reports about how the game is reacting to these situations. That needs to be clarified before anything else gets done.

[Battlefront.com]

The spotting bonus is only within the colored area of the arc. Outside of that there's no bonus. And remember there's a difference between spotting and engaging. Cover Arcs only limit engagements, not spotting. If something would be spotted without a Cover Arc it should be spotted with one. Again, engagement is a different thing.

A lot of TacAI, or AI for that matter, behavior is criticized by players because they wish to have their God perspective be the benchmark for what should or shouldn't be known to their units. As I said in my previous posts, that is absolutely NOT what we strive to do in CM. Especially in CMx2 with Relative Spotting.

I think C3K summed it up well:

If you were in a Tiger, how would you know a Firefly was shooting at you? The sound of the shell whooshing past? The sight of one looking at you? Or the impact of the shell? Is he shooting at YOU, or your wingman? Are you jussst about to squeeze a round off at him? Or, is that just a Sherman?

We have three choices of what to do in such a situation:

1. Have the Tiger know, with instant and absolute certainty, exactly what the situation is no matter what the variables are. This means knowing that something is in fact shooting at it, what that shot is, where it is coming from, where it might impact (e.g. rear armor much more threatening than front), what are the chances of significant damage, what are the chances of causing damage to the shooter first, etc. The Tiger would then make a perfect reaction tailored to this precise information.

2. Have the Tiger not know, at all, what is going on. It just sits there, brain dead, because you haven't told it to do anything else other than just sit there.

3. Have the Tiger receive information about what's shooting at it roughly modeling how in real life that particular type of tank, with that particular quality of crew, in that particular tactical terrain situation, with other things factored in on top might react. And just as in real life, this might mean correctly understanding that there's a Firefly firing at it, or incorrectly thinking it to be a mundane Sherman, or not being aware of the firing at all. Better still, this information may change as the seconds tick by and new information comes to that specific Tiger.

We always choose #3, but far too often players expect #1. Sometimes they get so frustrated with #3 that they seriously suggest #2. Yeah, I've seen people argue that if the TacAI can't "get it right" then give players absolute micromanagement authority. Which, of course, would produce far less desirable results on a regular basis, but that argument is usually not accepted in the heat of the moment.

Where there is a definable, repeatable shortcoming in the #3 logic we can (as best we are able to) make changes to the TacAI to better reflect the realistic responses one could reasonably expect for a specific situation. Short of that... we're not making any changes because 9 times out of 10 they would get us going towards logic #1. And we have no intention of purposefully doing that as it runs counter to CM's core philosophy.

Steve

[Childress]

The spotting bonus is only within the colored area of the arc. Outside of that there's no bonus.

Steve

Then what does the player do when he wants effective spotting, for example an FO, nose pressed against the bocage, but at the same inhibiting the spotter from firing at targets?

[sburke]

Then what does the player do when he wants effective spotting, for example an FO, nose pressed against the bocage, but at the same inhibiting the spotter from firing at targets?

I use a 360 degree short spotting arc. It has worked for me no problem. My FO only needs to see the terrain not specific units.

[Childress]

I use a 360 degree short spotting arc. It has worked for me no problem. My FO only needs to see the terrain not specific units.

But Steve just posted that units enjoy a spotting bonus *within* the Cover Arc. IOW, the FO could open fire on enemy troops exposing his position.

[JonS]

If someone can find me an actual account in a real war (other than the suicidal nuts in Japan) where an AFV crew knew they were being fired on by an obvious threat (i.e. not just an MG or sniper) and they still just sat there because the army told them to, please share.

Oh, actually. I've got one of those ... somewhere. It's of a British Sherman squadron (or regiment?) drawn up on the forward face of a hill, waiting for something (an attack to begin, IIRC). Whatever it was, the timing kept getting slipped, and they got left out there on the hill for most of the afternoon. Tanks were gradually getting picked off by a German tank or gun they assumed was hidden in trees somewhere on the opposite hill. Eventually the commander though 'screw this' and pulled what was left back through the hedgrerow behind them, then went off to swear at his brigadier.

There's similar stories from GOODWOOD.

And there are oodles of stories of infantry stolidly holding their positions under perpetual artillery fire.