when does or "should" the game (AI) take control away from the player?

[aka_tom_w]

I think this is the heart of the Borg Spotting issue

Jason C says:

"Tank cower is broken. Fix it please.

As a general rule, the Tac AI should ignore the player's orders to substitute its own, on the fly orders only if the SOP it then follows is right at least 9 times out of 10. The present tank reversing behavior is instead wrong 9 times out of 10.

Incidentally, infantry going to "sneak" toward the "nearest" cover when shot in open ground is another example of a broken Tac AI SOP. It is OK if the nearest cover is within 20m, perhaps. (Advance might work with cover within 50m). Otherwise hitting the deck completely is a far better SOP."

At some point there must be some agreement that the game (IMHO) will need to take EVEN MORE control away from the player in some instances to make Relative Spotting work even better.

Other folks here have suggested that EVERY unit out of C&C should come under the control of the AI. (i.e. no control by the player)

Well we have seen that in Jason's example PERFECTLY fit Tanks that are VETERAN status and have GOOD C&C (in good morale) do things out of the player's control and he does not like that. ( at least I think that is what he means is his post)

I am not so sure there is a real problem there, but I started this thread because the REAL issue is WHEN "should" the player LOOSE control of the unit. The way I see it the loss of control of any unit is and should be some form of appropriate consquence or AI punitive action for some dumb behaviour or negative result the player caused or let happen (or was simply the victim of)

If your tank crew is forced to bail out of the tank I think the TAC AI should take over that tank crew, they should be unable to SPOT ANYTHING and should do what they need to stay safe, (that means retreat if they can or just sneak to good cover and hide) but the player should have no control over them.

There seems to be agreement that when units Break panic or route it is completely OK for the game to take away control. NOW in CMBB if tanks break, route or panic the game also takes away control. I think is a good thing and so far it seems to work pretty well.

The tank cower behaviour however is an interesting issue. I have never seen it or been inconvienced by it so it is hard for me to comment on or judge Jason C's observations. :confused:

I would suspect that the way the game IS in this regard is the WAY Charles intends it to be ( I am guessing) so I would not call it a bug and if it is in fact, really broken, then Charles and Steve will have to want to "fix" it.

My guess is they don't want to fix this tank "cowar" behaviour.

(just my personal opinion only)

-tom w

[ May 10, 2003, 12:13 PM: Message edited by: aka_tom_w ]

[aka_tom_w]

no comments? :confused:

-tom w

[Tankist]

There is no problem with routing or broken tanks or squads. Even in 'classic' wargames, routed units are controlled by the 'system', and this is fine.

But one have to be subtle here; the more you take away control form the player (the single most important person), the more you make him an observer, instead of a 'guiding spirit'.

After all, people want to play the game, not observe the computer playing itself.

Sometimes more realism does not mean more fun...

Tankist

[Vanir Ausf B]

After all, people want to play the game, not observe the computer playing itself.

Sometimes more realism does not mean more fun...

Very true.

[JasonC]

If a tank crew goes on panic, as a morale state, fine let 'em run around like a headless chicken. When infantry has had all the fight shot out of them and routs off the map, great that is morale modeling and real friction in war.

But armor cower and cover panic are not morale states. They are SOP routines the game has the Tac AI impliment, in the belief that they are proper procedures to follow in unwelcome tactical situations that commonly arise in the middle of a minute, sometimes unexpectedly. They happen even with tanks in "OK" morale state, because the AI thinks they are sensible, not because the crew has lost its nerve.

The player has not yet had a chance to react to the information, presumed to be new, that a superior enemy AFV has LOS to this moving vehicle location, or an enemy MG covers that spot of open ground. The assumption is that the player would not want to send his infantry over open ground the MG can see, or his vunerable AFV across ground seen by an enemy AFV it is outmatched by.

In CMBO, the procedures associated with these cases were (1) infantry accelerates to run and "stiffens" morale, with the idea of "pressing on to cover". And (2), AFVs pop smoke and reverse 50-100m, with the idea of breaking LOS, preserving the presumably vunerable AFV, and living to try something else the following minute.

The former was based on the idea that fire into open ground is exceptional and to be avoided, and that the critical thing is limiting time of exposure. However, it encouraged unrealistic "rushes". It was also far more suitable as an SOP in western European terrain than in steppe.

In CMBB, it has been replaced with "cover panic", my own term for it. I call it that not because it is actually trying to model panic, but because "panic" is the in game morale state at which the player loses control of the unit, and the SOP kicking in has a comparable effect, simply because the present SOP is so often the wrong thing to try.

Still it is an improvement compared to CMBO - they aren't accelerating to run, they are hitting the deck at least. And heading to the nearest cover would be a sensible thing to do, if (1) that cover were reasonably close, (2) the movement rate used would actually get them there (which sneak does not do, beyond about 10m). There is also a seperate problem that the cover selection ignores lay of ground (no doubt it is just too hard to figure for all enemy shooters).

Being shot at in the open is not an unusual occurance in steppe terrain. Range must substitute for cover in steppe. In pure FP terms, there is little to choose between scattered trees at 200m or steppe at 400m.

The SOP too strongly distinguishes between them. It is not suicidal to fight from open or steppe terrain, when done right (in range, coordination with other arms, etc), and the SOP is too strictly based on the assumption that no sensible player wants his men to try.

The extra morale stiffness the "advance" command gives can partially counteract this, when the firing FP is low enough because the range is long enough. But units hit the "cover panic" SOP threshold long before they hit actual panic.

The simplest SOP when shot in the open is to hit the deck and stop moving. This is right for most ranged fighting, but can be frustrating in its own right in the cases where it is the wrong thing to do. E.g. a short gap of open ground between two bodies of good cover, both near enemies. If you drop and sit still then, you get shot at in poor cover needlessly.

A better SOP but a more complicated one is to head for the nearest cover at a speed that will reach that cover within 30 seconds. If there isn't any close enough, hit the deck and sit still. Or, in pseudo code, cover within 10m "sneak" toward it, else cover within 50m "advance" toward it, else "halt".

With armor, the problem is the SOP is designed to be right for "hopeless" match ups, especially ones entered into "by surprise". So the instant LOS is established, a threatened AFV decides to try to break contact instead of dueling. The intent is simply to stay alive long enough to reach the next orders phase and get some sensible instructions from the player.

But there are definite tactical situations, quite common whenever superior enemy AFVs are anywhere on the field, where the SOP is the wrong thing entirely, and completely misunderstands what is going on. A new LOS from a superior enemy AFV does not always mean "danger Will Robinson, get out of Dodge". Sometimes it means "perform a cresting drill to get off 1 shot while he is distracted". Sometimes it means "spring out of cover from an unexpected angle". Sometimes it means "everyone engage, hail fire, disable him with gun damage and track hits". And the Tac AI cannot be expected to know such things.

Since the player *can* be expected to know such things, the player must be trusted to know when it is sensible to order a tank into LOS of an enemy, and when it isn't. The presumption that a new LOS line means the player did not expect it or forsee it or intend it, simply is not true.

Intentionally breaking and creating LOS is the principle tactical variable used in armor combat. And must be used to keep down losses, whether the chance of a single round from a single AFV killing the enemy is high or low.

If an SOP is to override such decisions, it must be right 9 times out of 10, and it won't be if it always treats the new LOS as a threat. I have suggested the 2 major "decision items" to tell when the movement into LOS is intentional and when it isn't.

One, when a round is ready to fire, fire it first and then decide whether to break LOS. This covers all the cresting drills, allows hail fire, and handles some "spring from cover" situations where the primary issue is teamwork, "tag team" tactics.

Two, don't pop smoke and reverse unless the enemy is more or less facing you, or make that decision only once that becomes a threat. Gun facing within 45 degrees is a sufficient criteria. This handles all the cases where the reason the player established LOS was to "get the drop on" an enemy, to get in a "free" lick or two.

The present SOP has added problems because it does not read terrain well for LOS differentials, certainly not compared to human use of ground. It often heads for the wrong cover. It often wastes time slowly rotating, which is the most vunerable thing you can do. And reversing tanks do not fire.

But all of these would be livable if it were only going into the routine at all, in cases where the friendly AFV in question was genuinely and immediately in danger from a superior enemy AFV, while in the process of reloading or after already having taken one shot.

For what it is worth.

[aka_tom_w]

Hi Jason C

interesting post

BUT the real question we are curious about is are you using V1.03c to test out your theories?

(the latest public beta)

I have read your post and wonder if it is your expectation that your suggestions will be adopted for either the final 1.03 patch or the SOP game AI for CMAK?

I can perhaps see where some or your proposals might make it into the upcoming CMX2 , but I don't think we will see any real improvement for then.

-tom w

[ May 11, 2003, 12:21 PM: Message edited by: aka_tom_w ]

[JasonC]

I realized that some may not understand why in ranged fighting the right procedure for infantry shot at in open ground, with no cover nearby, is to halt and sit still. It can seem counter-intuitive, viewed in too narrow a focus on just the unit under fire. If the presumption is that being shot in the open is an unusual and disasterous event, and the unit must do something about it ASAP, then it might seem like "halt" is the last thing to try.

If you understand the role of an individual infantry sub unit in a larger attacking force, at range in open terrain, then it makes sense. Units advancing over open ground do so covered by "overwatch". Other assets are waiting to reply to anything that shoots at the moving units. The moving units are also taking turns; there are lots of them. Their role in the overall assault is to approach the enemy, draw fire, ID targets, get close enough to threaten the defenders, then finally to suppress them by their own infantry fire, allowing more units to close.

When a unit comes under fire in such an assault, it is usually because it is closer than others, and moving. When it stops moving it stops getting closer. The enemy firepower that can bear on it stops rising, because the range stops falling. In addition, other units continue to close. They attract fire. As the first unit is passed, fire shifts away from it, giving it time to rally.

By the time its turn to advance comes again, it has had a "breather". Enemy firepower is spread over many movers and many minutes of rally. All this is happening at relatively long range. Rally goes on continually, while the defender is under overwatch fire by heavier weapons and his ammo is limited.

In addition, as under fire it is probably one of the closest units to the enemy. If it still does not have full IDs of enemies, it can do little directly. But if it does have such full IDs, it is probably within effective rifle range. It can therefore reply to incoming fire. The suppression it can achieve will make it easier for other units advancing in the meantime. But to reply, it needs to remain stationary, not sneak sideways.

It is still possible for a unit to get caught in open ground, more than 50m from cover, and under fire strong enough that it can't simply absorb all of it and continually rally away its effects. But if that happens, all that occurs is the defenders keep shooting until they drive that small group of attackers into actual panic or rout. *Then* they run away, but not before.

As long as it is possible to continue playing their assigned role in the overall assault, it is far superior to trust the rest of the force to "shoot them free", or to pass them by, as they cover the rest of the team with their own fire. Which they do by halting.

Before the fire gets hot enough, say when the morale state is still just "alerted", they should continue their orders. "Move" might go into the routine at "cautious", while "advance" only goes into it at "shaken". That way, if the fp is low enough because the range is long enough, men won't panic just because the terrain is steppe (at 400m) rather than scattered trees (at 200m, just as dangerous objectively speaking).

I hope this clarifies why the SOP of "better cover 10m sneak, else better cover 50m advance, else halt" is the right sort of SOP to have when shot at in "open ground".

[ May 11, 2003, 01:00 PM: Message edited by: JasonC ]

[JasonC]

I am currently using v103, the public beta. All my observations are not made on that "build", however. I have had 102 much longer than 103, obviously, and have much more extensive experience with it. If major changes were coded in the 103 beta, some of these issues might be better than most of what I've seen, because I simply haven't seen all situations in 103 yet.

I haven't heard that such changes were made in 103, but I'd be happy to be wrong about that. As for whether SOP changes are something for CMX2, or for DAK, or for 103 - that is above my (for free) pay grade. I'd love to see them in 103, but wishes and practicalities are two different things and I'm not the judge of the latter. I can suggest tactically sensible SOPs and try to put them in pseudocode. That's all I can do.

[xerxes]

The only issue I have with infantry moving under fire in the open is that from reports I've read, the most likely behavior of a unit starting to panic is to "freeze". Moving towards the enemy, even if it's the "right" thing to do in the circumstances appears to be an unusual reaction.

But, this doesn't have anything to do with the original question.

I think control should be taken away whenever it makes sense to. Crews that have bailed from tanks that aren't in C&C should have control and spotting info taken away. Apart from that situation I'm not sure what other situations it would make sense and be playable.

[mchlstrt]

JasonC, glad to hear you're checking out 1.0.3. If you get any sense of where these behaviors are at in this build please post it. I generally find your thoughts on these sorts of things pretty enlightening, with the added advantage that you actually come up with some 'what would be rights' & not just a big pile of 'what is wrongs'.

As to what's possible, or even practical, yeah that's ultimately Charle's call, for all the right reasons. These discussions are useful for trying to sort out what it might be worth BFC's time to consider , & making people like me feel a little less tactically incompetent.

strt

[aka_tom_w]

Originally posted by mchlstrt:

JasonC, glad to hear you're checking out 1.0.3. If you get any sense of where these behaviors are at in this build please post it. I generally find your thoughts on these sorts of things pretty enlightening, with the added advantage that you actually come up with some 'what would be rights' & not just a big pile of 'what is wrongs'.

As to what's possible, or even practical, yeah that's ultimately Charle's call, for all the right reasons. These discussions are useful for trying to sort out what it might be worth BFC's time to consider , & making people like me feel a little less tactically incompetent.

strt

yes I agree

Good point

-tom w

[Jim Boggs]

After playing Boo_Radley in a PBEM, I think in his case the AI should have taken over before the set-up phase.

[MikeyD]

How much control you feel the AI should have has a lot to do with how far up the chain of command you imagine yourself to be at. If you see yourself as 'Battalion Commander' issuing sweeping orders and watching how your decisions impact the tide of battle, then a lot of AI control would feel right to you (as long as the AI operates in a logical manner).

But if you're always on the deck imagining yourself as the individual squad leader or tank commander any deviation made by the AI would naturally be more frustrating, like someone has taken your weapon from your hands.

The AI can cause frustration for both views of the game, of course. If you tend to conduct your battles like a French general in WWI you could get frustrated when your men appear unwilling to die like sheep. Or if you're playing like an individual tank comander you get frustrated when "your" tank doesn't act as bravely or rashly as you would've done (I have a mental image of Rambo running between mortar blasts while firing his squad mg from the hip).

But remember, whatever powers you get to push your men forward the AI would get as well, and any battle where both opponents show disregard for life and limb could make for a very short nonsensical battle. If you want to combine total control with senseless destruction get Quake.

[ May 12, 2003, 01:28 PM: Message edited by: MikeyD ]