Steel Beasts vs Combat Mission t-72 visibility test

[holoween]

54 minutes ago, BeondTheGrave said:

Given the fundamental assumptions and limitations of the CM system, I feel like these are pretty good and historically accurate, to the period, results. Even if there are some slight issues, @TheCaptain's several tests seem to suggest that they push the Soviet player to behave historically (closer engagements with plenty of backup) anyway. To reduce it to simple nerfs and buffs, I dont know that the Soviets need their long range spotting buffed really. If youre having trouble, the solution ought to be to keep pushing in, and to bring more tanks, which is Soviet doctrine anyway. As an American once said, "grab their nose and kick 'em in the ass." Just my .02$

Not bad spotting doesnt push the soviet player to play soviet doctrine. Lanchesters square law does that plenty. Look at any tank attack and youll see that they all look practically the same. Currently the Americans do Soviet style attacks better than the soviets because they actually see their targets.

It may very well be that soviets tanks were worse at spotting but im not convinced that was inherent to their vehicles rather than the training level.

You could also blame it on doctrine and have a modifier for it for each nationality but that would be difficult to find out, model and leads to strange results like M1A1 being more comparable to a loe1 than a leo2

https://mcoepublic.blob.core.usgovcloudapi.net/library/ebooks/Canadian Army Trophy Book_2018.pdf

see page 240

 

Edited October 20, 2021 by holoween

[holoween]

delete pls

Edited October 20, 2021 by holoween

[Artkin]

38 minutes ago, holoween said:

Not bad spotting doesnt push the soviet player to play soviet doctrine. Lanchesters square law does that plenty. Look at any tank attack and youll see that they all look practically the same. Currently the Americans do Soviet style attacks better than the soviets because they actually see their targets.

It may very well be that soviets tanks were worse at spotting but im not convinced that was inherent to their vehicles rather than the training level.

You could also blame it on doctrine and have a modifier for it for each nationality but that would be difficult to find out, model and leads to strange results like M1A1 being more comparable to a loe1 than a leo2

https://mcoepublic.blob.core.usgovcloudapi.net/library/ebooks/Canadian Army Trophy Book_2018.pdf

see page 240

 

Nice. Leo1 outperforming M1A1. How did they do in 1991 after cutting down the radio chatter? 

Nevermind we will never know! They werent present due to Gulf war. 

Edited October 20, 2021 by Artkin

[Centurian52]

Ran my earlier test again, but with the M60A1 trying to spot the T64. I remounted the crew of the M60 and dismounted the crew of the T64. Lighting and conditions are identical to my earlier tests (clear, 1100, dry). Because the T64 crew in the earlier tests were regular with +2 leadership I set the M60 crew to regular with +2 leadership. The results were as follows:

(?) spot      firm spot   (in seconds)

18               18
213            213
133            133
50              50
98              98
22              29
81              81
58             72
92             92
17              34

average time for (?) spot: 78.2 seconds (a lot of the spots on this run resulted in instant identification)

average time for firm spot: 82 seconds

Overall the results for the M60A1 were about the same as the T64A. So I don't think it's a matter of American tanks spotting better than Soviet tanks. I think it's a matter of the M60A3 and M1 Abrams specifically spotting better than every other vehicle in the game, American or Soviet (although maybe I should test the M60A1 RISE). Tbh I was expecting the M60 to do slightly better. Not because it has better optics, but because it has one more crewmember to help spot.

[BletchleyGeek]

9 hours ago, The_Capt said:

Not definitive but those tricky lads at BFC appear to have linked spot time to scan distance, or at least this is a working theory.  

I think either Vanir or akd (or someone else I can't remember) discussed a bunch (read in the 100s of samples IIRC) of exhaustive tests on this very same kind of issue, different matchups, like 5 or 6 years ago. It was quite clear that indeed, the area to cover by the spotter was a critical parameter. There is also some probabilistic model in place that allows to account for "better optics", "better situational awareness", and probably some combat psychology (how overloaded is attention of the crewman doing the spotter). Hence why you need very high (probably astronomical) sample sizes to reverse engineer the actual process programmed in the game. So unless Charles pops in to give us a one hour lecture on the subject I think we will all be long gone before anybody can actually reverse engineer CMx2 spotting.

I don't think BFC is doing raytracing plus physical modelling of the human eye and/or optical devices to determine that an object has been spotted.

Let's say that spotting isn't precisely the strongest suit of CMx2 simulation capabilities. I am personally content with knowing that 1) the scan area matters a lot, 2) comms matter almost as much, 3) equipment (both optics and design of turret) matters somewhat, and 4) psychological stress matters a bit (under fire, crew suffered casualties).

My intuition is that the OP just came across the heavy tail of the to-spot-chance "die roll". 

[domfluff]

We actually know how this part works.

"Clutter" isn't really a thing, although there was mention of an increased change of things like wire-guided ATGM failure, but we know that the underlying terrain tiles along the firing line degrade spotting.

This is analogous (and presumably directly analogous) to the Hindrance rules in Advanced Squad Leader and derived titles.

[BletchleyGeek]

1 minute ago, domfluff said:

We actually know how this part works.

If you have a link to the pseudocode feel free to share it. Peeps here will appreciate it.

[domfluff]

I can describe how it works in Combat Commander, since that's immediately in my head:



(Combat Commander is also derived from Squad Leader, which is the genesis of Combat Mission, so although it's not 1:1. you can see a lot of the same DNA).

Above is an American rifle squad. Stats are Firepower 6, Range 6, Move 4, Morale 6 on the unbroken side, and these mostly go down on the broken side. Squares are special things with other triggers (in this case, automatic fire and smoke grenades).

Generating fires would be comparing the firepower of the lead unit, in this case 6, and then adding one for each additional unit/weapon participating in the attack, so a full platoon firing at a single squad might start at 8.

This is then modified by hinderance. The line of fire is drawn, and each piece of terrain this line crosses over in between firer and shooter will modify the shot - but only the highest modifier is counted. This might mean -1 for some rough terrain and -2 for some high wheat fields, meaning that the total hinderance would be -2, for a resultant firepower of 8-2 = 6.

(This part is the similarity, and we know from previous communication that a very similar mechanism is part of the spotting/firing solution in CM.)

The firer then rolls 2D6 (in combat commander's case this is drawing a card with dice printed in the corner, but it's 2D6 nonetheless), and adding their modified firepower. The target is the enemy morale, and they're trying to roll great than that to break them. This morale is modified by the cover they are in (so, a building would add +2 to this target number.

Succeeding against an unbroken unit breaks them, against a broken unit destroys them.

[The_Capt]

11 hours ago, holoween said:

Are you removing outliers? At 10 rounds at 1000m i already have one taking 177s to spot and one taking 255 yet im not seeing such results for your tests.

Nope, 10 random tests as they came, if you are using exactly the same conditions then we are probably talking dice rolls.  Those are good documents but I am still having a hard time accepting that a tank built in 1979 has a worse backup targeting system than a Soviet tank built a decade earlier...but stranger thing have happened.  That, and competition range conditions is going to work very differently than wartime tactical friction.

AKD is correct, we would probably need to do 100 plus run throughs to really see the curve but these conditions are very controlled if you load and run the exact same scenario 10 times.  Scan arc looks like it is definitely playing a role here...and I am wondering how it handles broken terrain now.

My results of a single tank at 2000m look very much like Centurian52s with mean time to clear-spot being only 4 seconds faster for the T64.

I am pretty much past the whole "spotting is broken!" thing at this point.  I am not really keen on comparing SB to CM because we might find ourselves in an awkward position; SB is SB and looks like a really good game, CM is CM and is a good game.  Both are wargames with different emphasis/game engines and wargamers should stick together.

What I am finding interesting is Tank Platoon behavior at the moment.  So when the target gets spotted, not all 4 tanks immediately spot it.  Normally one does and then the rest follow up, particularly after the first tank start firing.  This might give us a sense of how C2 works.  Communicating a target to a friendly unit is incredibly hard to do (it is there...right there!  Where is there?), we have developed a lot of procedures and tactics to do it, simplest being "follow my tracer".  CM looks to have something under the hood going on but I have not pinned it down yet.

I can see why UK Mod got onboard with CM Pro because if you can simulate effective tactical C2 under various conditions repeatedly, you are moving past a training support tool and entering into operational research territory.

Edited October 20, 2021 by The_Capt

[The_Capt]

7 hours ago, Centurian52 said:

Ran my earlier test again, but with the M60A1 trying to spot the T64. I remounted the crew of the M60 and dismounted the crew of the T64. Lighting and conditions are identical to my earlier tests (clear, 1100, dry). Because the T64 crew in the earlier tests were regular with +2 leadership I set the M60 crew to regular with +2 leadership. The results were as follows:

(?) spot      firm spot   (in seconds)

18               18
213            213
133            133
50              50
98              98
22              29
81              81
58             72
92             92
17              34

average time for (?) spot: 78.2 seconds (a lot of the spots on this run resulted in instant identification)

average time for firm spot: 82 seconds

Overall the results for the M60A1 were about the same as the T64A. So I don't think it's a matter of American tanks spotting better than Soviet tanks. I think it's a matter of the M60A3 and M1 Abrams specifically spotting better than every other vehicle in the game, American or Soviet (although maybe I should test the M60A1 RISE). Tbh I was expecting the M60 to do slightly better. Not because it has better optics, but because it has one more crewmember to help spot.

I would be interested to see what the T80 can do.

[The_Capt]

2 hours ago, BletchleyGeek said:

I don't think BFC is doing raytracing plus physical modelling of the human eye and/or optical devices to determine that an object has been spotted.

CMx3..next year in the holy land?

[holoween]

9 minutes ago, The_Capt said:

Nope, 10 random tests as they came, if you are using exactly the same conditions then we are probably talking dice rolls.  Those are good documents but I am still having a hard time accepting that a tank built in 1979 has a worse backup targeting system than a Soviet tank built a decade earlier...but stranger thing have happened.  That, and competition range conditions is going to work very differently than wartime tactical friction.

The backup sight from Leo2 is comparable to the main sight of T72. Seems reasonable to me. It is a backup sight so no point massively overbuilding it.

9 minutes ago, The_Capt said:

AKD is correct, we would probably need to do 100 plus run throughs to really see the curve but these conditions are very controlled if you load and run the exact same scenario 10 times. 

Plot the results on a graph and youll see a pattern. So far with with 60 tests under 3 different conditions i can certainle see a trend emerging.

9 minutes ago, The_Capt said:

I am pretty much past the whole "spotting is broken!" thing at this point. 

It isnt broken. For the most part it works great.

9 minutes ago, The_Capt said:

What I am finding interesting is Tank Platoon behavior at the moment.  So when the target get spotted, not all 4 tank immediately spot it.  Normally one does and then the rest follow up, particularly after the first tank start firing.  This might give us a sense of how C2 works.  Communicating a target to a friendly unit is incredibly hard to do (it is there...right there!  Where is there?), we have developed a lot of procedures and tactics to do it, simplest being "follow my tracer".  CM looks to have something under the hood going on but I have not pinned it down yet.

I can see why UK Mod got onboard with CM Pro because if you can simulate effective tactical C2 under various conditions repeatedly, you are moving past a training support tool and entering into operational research territory.

Run 4 tanks without command links and see if they have worse results. The time between no contact to any contact at all should be exactly the same between 4 tanks in and out of C2. Only if there is a noticable differece is there a chance platoons working together is modeled.

[Vanir Ausf B]

C2 can make a big difference in spotting times. Units with a contact marker on an unspotted unit will spot that unit around 50% faster.

[domfluff]

41 minutes ago, The_Capt said:

What I am finding interesting is Tank Platoon behavior at the moment.  So when the target gets spotted, not all 4 tanks immediately spot it.  Normally one does and then the rest follow up, particularly after the first tank start firing.  This might give us a sense of how C2 works.  Communicating a target to a friendly unit is incredibly hard to do (it is there...right there!  Where is there?), we have developed a lot of procedures and tactics to do it, simplest being "follow my tracer".  CM looks to have something under the hood going on but I have not pinned it down yet.

We've got a fair idea of that too.

Partial spots will resolve into full spots faster than no spots. This time isn't constant (so, dice rolls), but there is a clear tendency here.

If a platoon was four tanks with C2 links, but dispersed, then a spot from Tank 3 will spread to the Platoon HQ, and then down to the remaining tanks. You will see the HQ pick up the spot first, then the others in the platoon. Again, this transit time isn't constant, but will depend on C2 quality and other, presumably random elements. This is all assuming valid C2 links, therefore unbroken radios, actual LOS comms and unsuppressed crews.

If the tanks are within horizontal sharing distance (4 squares) they will share to their neighbours as well.

That means that a platoon will be expected to spot faster than an individual tank in real terms.

[The_Capt]

10 minutes ago, domfluff said:

We've got a fair idea of that too

This is fantastic stuff but who is "we"? Did someone write all this up before (very likely)?  Do you have a link?  I am not surprised to be honest, we have saying "amateurs talk tactics, pros talk logistics, masters talk C2...God holds the dice" and it would appear that this road is leading there.

[domfluff]

That's the big thread, but it's really key to "how to play CM" in a lot of ways. The C2 sharing is sophisticated, and it has some obvious consequences.

As a simple one, one of the reasons why you want recon elements embedded at the battalion level, and not added as single vehicles directly into a lower formation (Czechmate is an example of that), is that the correct units at the correct place in the C2 chain will share their information much more efficiently, and they can do their job much better.

This is especially important in the hyper-modern scenarios in CM, because winning the information war is often more important than winning the physical one (or should be, in a scenario which really understands what CMBS can be about).

[domfluff]

As far as I'm aware there haven't been extensive tests on exactly how much a partial spot can improve spotting, but it's such a significant upgrade that it's really self-evident in practice.

You really don't ever want to send in a tank platoon into an engagement without letting them know what they're expecting. 

[The_Capt]

17 minutes ago, domfluff said:

That's the big thread, but it's really key to "how to play CM" in a lot of ways. The C2 sharing is sophisticated, and it has some obvious consequences.

As a simple one, one of the reasons why you want recon elements embedded at the battalion level, and not added as single vehicles directly into a lower formation (Czechmate is an example of that), is that the correct units at the correct place in the C2 chain will share their information much more efficiently, and they can do their job much better.

This is especially important in the hyper-modern scenarios in CM, because winning the information war is often more important than winning the physical one (or should be, in a scenario which really understands what CMBS can be about).

Been playing this game a long time and still learning stuff about it...nice.

[chuckdyke]

7 minutes ago, The_Capt said:

Been playing this game a long time and still learning stuff about it...nice.

BN CW My Honor is Loyalty another example. the Aufklarung is embedded with the Panther Kompanie but they share their links off map by radio. 

[holoween]

46 minutes ago, domfluff said:

That's the big thread, but it's really key to "how to play CM" in a lot of ways. The C2 sharing is sophisticated, and it has some obvious consequences.

Agreed:

But then you realize all militaries have different doctrine on how exactly to deal with this which also changes with time and CM doesnt(as far as i can tell) model any of that. Its just a generic system.

[A Canadian Cat]

25 minutes ago, holoween said:

Agreed:

But then you realize all militaries have different doctrine on how exactly to deal with this which also changes with time and CM doesnt(as far as i can tell) model any of that. Its just a generic system.

They do model who gets a radio and who doesn't and who reports to whom. That clearly has an effect. But otherwise doctrine is not in the CM model - its in our orders.

[domfluff]

9 minutes ago, IanL said:

They do model who gets a radio and who doesn't and who reports to whom. That clearly has an effect. But otherwise doctrine is not in the CM model - its in our orders.

Yes, it absolutely does.

The most obvious effect is in Cold War. The US pushes combined arms down to the company level - if the M150's spot the incoming armour, then the tanks will hear about it very quickly.

The Soviets instead mostly do not, and their combined arms layer is at the battalion level (e.g., that ATGM formation is probably attached at the battalion level, so has to go through more steps, and often worse steps, to spread to the equivalent tank platoon).

That has a lot of consequences, some of them subtle but it does mean that the Soviets will tend to suffer at the smaller unit level, comparatively speaking.

That's quite aside from differences in optics, number of crewmen (all else being equal, a three man tank crew will spot worse than a four man tank crew, since the autoloader can't tell you what it sees), and quality or quantity of radios.

That TO&E structure is entirely due to doctrine, and has a strong effect on the outcome of how information is shared, and ultimately how spotting will work.

Edited October 20, 2021 by domfluff

[Vanir Ausf B]

1 hour ago, domfluff said:

As far as I'm aware there haven't been extensive tests on exactly how much a partial spot can improve spotting

 

[domfluff]

Awesome, thank you. Will digest.

[holoween]

14 minutes ago, IanL said:

They do model who gets a radio and who doesn't and who reports to whom. That clearly has an effect. But otherwise doctrine is not in the CM model - its in our orders.

It doesnt model properly who sits on what radio. The information simply flows up the command chain and back down.

5 minutes ago, domfluff said:

Yes, it absolutely does.

The most obvious effect is in Cold War. The US pushes combined arms down to the company level - if the M150's spot the incoming armour, then the tanks will hear about it very quickly.

The Soviets instead mostly do not, and their combined arms layer is at the battalion level (e.g., that ATGM formation is probably attached at the battalion level, so has to go through more steps, and often worse steps, to spread to the equivalent tank platoon).

That has a lot of consequences, some of them subtle but it does mean that the Soviets will tend to suffer at the smaller unit level, comparatively speaking.

That's quite aside from differences in optics, number of crewmen (all else being equal, a three man tank crew will spot worse than a four man tank crew, since the autoloader can't tell you what it sees), and quality or quantity of radios.

That TO&E structure is entirely due to doctrine, and has a strong effect on the outcome of how information is shared, and ultimately how spotting will work.

It doesnt model things like an entire platoon being on the same net so as soon as information gets transmitted over the platoon net it should reach everyone but ingame it only reaches the platoon leader.

So while low level recon assets do help quite a bit of information sharing is simply not correct.