Recon underestimation

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

Recon underestimation

View related threads: (in this forum | in all forums)

Logged in as: Guest

Users viewing this topic: none

Printable Version

All Forums >> [New Releases from Matrix Games] >> Gary Grigsby's War in the East 2 >> Recon underestimation

Page: [1]

Message

<< Older Topic Newer Topic >>

Recon underestimation - 8/6/2021 6:14:27 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

An observation: in my experience the estimated CV of enemy formations is almost always (if not literally always) an underestimation of the true CV when the formation is attacked.

This wouldn't seem to be realistic - shouldn't just as often recon be an overestimate?

Currently you can look at an enemy stack, add around +30% to its CV and be reasonably confident you're in the ballpark of its real CV.

< Message edited by squatter -- 8/6/2021 8:53:09 PM >

Post #: 1

RE: Recon underestimation - 8/7/2021 12:23:47 AM

jubjub

Posts: 493
Joined: 5/2/2021
Status: offline

Like everything in life, it depends. Are you attacking an SS motorized division, or a stack of Soviet infantry?

(in reply to squatter)

Post #: 2

RE: Recon underestimation - 8/7/2021 9:42:43 AM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

Perhaps I'm not being clear.

What I'm saying is that the game impliments fog of war by showing you the estimated CV values of enemy formations based on current recon level of the hex.

However from experience these estimates are consistently an underestimate typically in the range of 30%.

Two things follow:
1 if I am able to reliably simply add +30% to the displayed value to come up with something close to the real value of the enemy CV then fog of war is not working.

2 The way to fix this is surely to have the estimated enemy CV anywhere between -30% and +30%, rather than always -30%. This would both fit reality, and return meaningful FOW to the game.

That make sense?

(in reply to jubjub)

Post #: 3

RE: Recon underestimation - 8/7/2021 9:48:47 AM

xhoel

Posts: 3219
Joined: 6/24/2017
From: Germany
Status: offline

The display CVs are almost always incorrect so you have to go with your gut feeling and judge a bit based on the units you are facing, like jubjub says. If its a stack of 3 Soviet RDs in a level 2 hex that is showing a 5 defensive CV, that is 99% a big underestimation. If a badly supplied German Regiment defending in clear with a 10 CV that is most likely a great overestimation.

You get used to it after a while.

PS: Just ssaw your second post. It is not always an underestimation, that depends on which side you are playing, weather, what time of the war etc.

< Message edited by xhoel -- 8/7/2021 9:51:49 AM >

_____________________________

AAR WITW: Gotterdammerung 43-45
http://www.matrixgames.com/forums/tm.asp?m=4490035
AAR WITE: A Clash of Titans 41-45
http://www.matrixgames.com/forums/tm.asp?m=4488465
WitE 2 Tester and Test Coordinator

(in reply to jubjub)

Post #: 4

RE: Recon underestimation - 8/7/2021 2:30:27 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

quote:

ORIGINAL: xhoel

quote:

PS: Just ssaw your second post. It is not always an underestimation, that depends on which side you are playing, weather, what time of the war etc.

My experience is always Soviet. And tbh I am using experience from winter 41/42 as my primary evidence.

What I can say with confidence is: during winter 1941/42, as the Soviet player the difference between the estimated display CV, and the true CV will usually be that the real CV is around +30%. (this is a generalisation, sometimes the display CV and real CV are the same, but mostly the real CV is in the +20 +30% mark).

1 So is this not the case for the Axis (I can't speak to this)?

2 So is this different during the summer for the Soviets?

3 The point remains that certainly during first winter, the FOW doesn't really work as far as the Sov player, as using the rule of thumb above, you can be pretty confident of the real CV, without even having to think about fort/terrain/type of unit etc. Just look at the display CV and add +30% and you'll be in the right ballpark. Do you not see the issue?

This being the case, can you not see my point? the FOW doesn't work if the 'estimated' display CV is consistently the same

(in reply to xhoel)

Post #: 5

RE: Recon underestimation - 8/8/2021 11:52:46 AM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

Just had a quick spin on the Vistula scenario and attacked all the German units across the front.

In every case the estimated CV was always less (occasionally equal to) the real CV of the defending units. Typically the real CV was somewhere between 20-60% more.

Surely if your recon is only ever giving you underestimates of enemy strength (and usually within a predictable bandwidth), and never over estimates then something is amiss with the FOW? It doesn't represent reality (in which if anything I would have thought overestimates were more common than underestimates?)

(in reply to squatter)

Post #: 6

RE: Recon underestimation - 8/8/2021 1:22:06 PM

loki100

Posts: 10920
Joined: 10/20/2012
From: Utlima Thule
Status: offline

Some guesses.

a) unless you are re-enacting the Army of the Potomac, I guess that most poor intel will tend to under-estimate whats there? So that maybe one bias
b) the recon system works in steps, from not having a clue, to there is something there, to its a tank, to its a unit with an estimated size, to a unit with a known size, the situation you are describing I guess is when you have recon in the second best category?
c) cv error is not the only way the game system uses DL, if you attack you can pick up some adverse results from tactical surprise, its one reason why what looks like a sure win done without recon can produce a nasty defeat

I'm more worried that you can get pretty good recon using the mouse over for a possible attack.

But both for this, and the instance where you think that there is a predictable under-estimate, to some extent get swept up in the combat system itself. Add on all the shifts that go on with the actual end cv and I think this is relatively minor. As Xhoel says, its just another thing to factor into your framing of what may happen if you actually do attack?

_____________________________

(in reply to squatter)

Post #: 7

RE: Recon underestimation - 8/8/2021 2:25:09 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

I think I'm not making my point very well so I'll try to illustrate.

Here's an image of a section of the front, March 41. I'm soviets. I've selected this region randomly.

Let's look at the string of enemy stacks starting with defensive CV 18 going south with CV of 7, 11, 19, 12.

To my mind, a fog of war should make me look at those CVs and think "I can't be sure what the real value of those units are. Could be more than the numbers I see, could be less"

But the trouble is that because of the way recon/FOW works in this game, and because I've been paying attention to this, I have a magic formula that will tell me pretty much their real CV. And that real CV is going to be somewhere in the region of +10% to +50% to the number shown, mostly around +30%.

Attachment (1)

(in reply to loki100)

Post #: 8

RE: Recon underestimation - 8/8/2021 2:28:37 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

To demonstrate, I now make hasty attacks with cavalry on the units I listed to reveal their true CV.

Firstly the stack showing CV18. The hasty attack reveals the true CV is 20. So the real CV in this case is +12%

Attachment (1)

(in reply to squatter)

Post #: 9

RE: Recon underestimation - 8/8/2021 2:32:00 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

Next I attack the hex with estimated CV 7, revealing a true CV of 9. So the true CV is around +30% here.

Attachment (1)

(in reply to squatter)

Post #: 10

RE: Recon underestimation - 8/8/2021 2:37:14 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

Next moving down to the hex with CV 11.

I attack to discover the real CV is 16. Which in this case is around +45%. A little on the higher end of the range, but more careful research would have revealed the mouse-over CV of the hex is 5.7 + 6.9 = 12.6. That brings the difference from recon CV to true CV back to the sweet spot of approx +30%.

Attachment (1)

(in reply to squatter)

Post #: 11

RE: Recon underestimation - 8/8/2021 2:39:56 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

Now for the unit displaying CV 19. My hasty attack reveals a true CV of 215. That's a difference of around +15%.

Attachment (1)

(in reply to squatter)

Post #: 12

RE: Recon underestimation - 8/8/2021 2:41:55 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

And finally the unit showing estimated CV of 12. The hasty attack reveals true CV of 17. That's a difference around +40%

Attachment (1)

< Message edited by squatter -- 8/8/2021 2:42:17 PM >

(in reply to squatter)

Post #: 13

RE: Recon underestimation - 8/8/2021 2:55:07 PM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

So in this small data set, we have the difference between the estimated CVs and the real CVs as +12%, +30%, +30%, +15%, +40%.

And in my experience this pattern is repeated across the entire front, every turn.

Which means as a simple rule of thumb, if I look at an enemy hex, and add +30% to it, I can be pretty confident I've got the real CV of the unit (or pretty close), without carrying out any kind of recon (air or scouting attack). At worst, I might have overestimated by 20%, or underestimated by 10% (obviously talking generalities here).

So again, my point is that the FOW is undermined by the way in which estimated recon of enemy CV is always underestimated in this predictable pattern. ie estimated CV always seems to fall within a bandwidth of 0% to +40%, with a strong tendency to around +30%.

Currently, at a glance I can be pretty certain I know the true CV of enemy stacks.

My strong suggestion is that if the bandwidth for estimated CV was changed to go both ABOVE AND BELOW the real CV, then suddenly I can no longer be confident of the true enemy CV.

I mean IRL, if you were sat opposite a German division with only basic intel, you would be just as likely to overestimate its strength (ie presume it's a full division, whereas in reality it might well be understrength), rather than underestimate it (ie assume something like you are facing just two regiments of the division, rather than the whole formation.)? I mean the default position would be to assume full strength (over estimate) rather than assume a depleted formation (under estimate)? Whereas in the game, all estimations are under estimates.

(in reply to squatter)

Post #: 14

RE: Recon underestimation - 8/8/2021 3:02:02 PM

loki100

Posts: 10920
Joined: 10/20/2012
From: Utlima Thule
Status: offline

quote:

ORIGINAL: squatter

I think I'm not making my point very well so I'll try to illustrate.

Here's an image of a section of the front, March 41. I'm soviets. I've selected this region randomly.

Let's look at the string of enemy stacks starting with defensive CV 18 going south with CV of 7, 11, 19, 12.

To my mind, a fog of war should make me look at those CVs and think "I can't be sure what the real value of those units are. Could be more than the numbers I see, could be less"

...

I think you are making your point quite clearly, so yes in most circumstances a less than full DL will under-estimate what is there but even your egs suggest this is over a fair range. Equally, and I realise this is a small effect, your stack doesn't show any cv gain off SU allocation.

My understanding is that every army in the war did a degree of constant recon. One thing that was of constant interest was unit designations, that was used to estimate what was there locally as well as to track wider moves - I realise you know this but I'm making the point as to how this feeds into the game.

So a commander would have a pretty decent idea as to what was detected and what might be missing - that is the calculation you are making. What you don't know is exactly what is missing or - perhaps more importantly - where it is. The sector you are showing is one where the units are in some degree of contact (I'm going by the fort levels here), so its at this level I'd expect uncertainty to exist.

Now all this is somewhat below the level of the game but is reflected in the combat model. As your DL slips away, the chance of something nasty happening in combat rises.

We can then add in all the other variables in the combat engine that mostly reward well led, well trained troops. In other words, the ability to cope well with the consequences of poor recon.

I guess I'm not sure (a) why you are seeing this as such a huge issue or (b) why you think a lack of recon should trigger systemic over-estimation. You are not showing instances where an attack is made on a unit that wasn't detected till the encounter happened, never mind if that unit was still at the weakest level of detection. That is where I'd expect errors of over and under-estimation, not when the issue is that there are gaps in your knowledge

_____________________________

(in reply to squatter)

Post #: 15

RE: Recon underestimation - 8/8/2021 3:10:05 PM

Sammy5IsAlive

Posts: 514
Joined: 8/4/2014
Status: offline

I'm pretty sure that the CV values you get on the pop up when you are about to attack (i.e. before you release the mouse button) are fully accurate and free of FoW. So you can find out that way. That was the case at least in earlier versions, haven't checked recently to see if it is fixed.

Edit - just realised that Loki made this point already above - missed it first time round

< Message edited by Sammy5IsAlive -- 8/8/2021 3:17:56 PM >

(in reply to loki100)

Post #: 16

RE: Recon underestimation - 8/8/2021 5:09:33 PM

loki100

Posts: 10920
Joined: 10/20/2012
From: Utlima Thule
Status: offline

quote:

ORIGINAL: Sammy5IsAlive

I'm pretty sure that the CV values you get on the pop up when you are about to attack (i.e. before you release the mouse button) are fully accurate and free of FoW. So you can find out that way. That was the case at least in earlier versions, haven't checked recently to see if it is fixed.

Edit - just realised that Loki made this point already above - missed it first time round

even that is not completely reliable. Here's a T2 situation, I have units on 2 hexes and decent air recon. I don't believe the claimed cv, its a normal division in a clear hex (vs AI game), so the 12 is not plausible but its waht is being reported:

attack with 2 regiments as I am a bit worried:

so the '12' really was a '<1', this isn't a Soviet unit shedding cv as leadership rolls are failed (though that has happened in the actual attack)

< Message edited by loki100 -- 8/8/2021 5:10:08 PM >

_____________________________

(in reply to Sammy5IsAlive)

Post #: 17

RE: Recon underestimation - 8/9/2021 4:47:58 PM

Mehring

Posts: 2179
Joined: 1/25/2007
Status: offline

I think squatter makes a good point. Just as you generally seek to attack where you are strong and the adversary weak, the defender seeks to appear strong where they are weak and vice versa. Dummy positions and fake hardware were built extensively to draw fire as well as to give an appearance of strength. There's a good case for under as well as over estimating defensive values.

_____________________________

�Old age is the most unexpected of all things that can happen to a man.�
-Leon Trotsky

(in reply to loki100)

Post #: 18

RE: Recon underestimation - 8/9/2021 8:31:24 PM

carlkay58

Posts: 8650
Joined: 7/25/2010
Status: offline

I think that you are tending towards underestimation as the effect because you have not extensively attacked units that have a reported high CV. I, personally, have found that attacking a high CV stack can result in retreats and routs fairly easily - or at least unless it can't. :)

The biggest point is the one that Loki made. With experience you can recognize that the CV being reported as very high can be ruled as an overestimated CV, the low CVs are the ones that can get you because they actually can be that low.

(in reply to Mehring)

Post #: 19

RE: Recon underestimation - 8/10/2021 8:11:43 AM

Mehring

Posts: 2179
Joined: 1/25/2007
Status: offline

But squatter's point is that you don't see over represented CV's. Given the enormous scope of CV for the same base unit, for the variables of supply, combat readiness, leadership, attached units, etc etc etc, it should not be possible to learn the CV of an unknown unit, certainly not those of human opponent.

_____________________________

�Old age is the most unexpected of all things that can happen to a man.�
-Leon Trotsky

(in reply to carlkay58)

Post #: 20

RE: Recon underestimation - 8/10/2021 8:18:18 AM

loki100

Posts: 10920
Joined: 10/20/2012
From: Utlima Thule
Status: offline

quote:

ORIGINAL: Mehring

But squatter's point is that you don't see over represented CV's. Given the enormous scope of CV for the same base unit, for the variables of supply, combat readiness, leadership, attached units, etc etc etc, it should not be possible to learn the CV of an unknown unit, certainly not those of human opponent.

except that I have provided an example of a clearly over-inflated cv in the image above.

There is (I suspect) a fair bit of attribution bias here. If you see a weak unit, you'll attack and may make a mental adjustment that it could well be stronger and/or get a reserve re-action. if you see a strong unit, you may well look for another option - now this is fairly sensible but what you then miss is evidence for appears strong/is actually fairly weak.

Or, to be rather pedantic, never sample on the outcomes, in effect (we all) tend to (in this case) to sample on 'appears weak' and very rarely on 'appears strong'.

_____________________________

(in reply to Mehring)

Post #: 21

RE: Recon underestimation - 8/10/2021 8:56:32 AM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

quote:

ORIGINAL: Sammy5IsAlive

I'm pretty sure that the CV values you get on the pop up when you are about to attack (i.e. before you release the mouse button) are fully accurate and free of FoW. So you can find out that way. That was the case at least in earlier versions, haven't checked recently to see if it is fixed.

Edit - just realised that Loki made this point already above - missed it first time round

quote:

realised that Loki made this point already above - missed it first time ro

This isn't true - the display CV is an approximation of the pop-up CV. Obviously if the pop-up CV was accurate to true CV, FOW would be completely broken. It's not completely broken, just a little bit broken!

(in reply to Sammy5IsAlive)

Post #: 22

RE: Recon underestimation - 8/10/2021 9:25:56 AM

squatter

Posts: 1033
Joined: 6/24/2006
Status: offline

quote:

I guess I'm not sure (a) why you are seeing this as such a huge issue or (b) why you think a lack of recon should trigger systemic over-estimation. You are not showing instances where an attack is made on a unit that wasn't detected till the encounter happened, never mind if that unit was still at the weakest level of detection. That is where I'd expect errors of over and under-estimation, not when the issue is that there are gaps in your knowledge

In response to a)

On reflection I don't think it's a huge issue actually. But an issue nonetheless. I've come to realise that my experience only applies to when front lines have settled down, and only to the Sov winter offensive, when lines are fairly static.

As you point out, the equation changes and recon variable increase when mobile warfare develops, lines are fluid and you bump into units during your turn. You also demonstrate in your example how in mobile circumstances it's possible that your recon will overestimate enemy strength, and the gap between estimated CV and true CV can be much higher than the band I've observed. This is all good and as it should be.

But what it leaves us with - and we all have to ask ourselves if that in terms of FOW realism whether we feel this is correct - is that once front lines settle down, your recon will never overestimate enemy CV.

And that as per my examples (and the 100s of others I've been observing), you may as well put in the rule book the fact that you can relatively safely assume that any enemy unit which started the turn in contact with one of yours will have a true CV of somewhere between display CV, and display CV +50%, with a very strong tendency towards the mean.

If we feel that this relatively narrow band of doubt adequately reflects reality of FOW (under circumstances of 'settled lines') then all is good. I guess I don't have a huge problem with this, perhaps this kind of bandwidth is correct.

But in response to your b) I am not and have not been suggesting the game gives you SYSTEMATIC over estimation of enemy strength, rather that it give you OCCASSIONAL over estimation of enemy strength. Again, this is caveated by me now talking specifically about circumstances in which stable lines have formed (which let's not forget is the majority of game time, over the campaign).

I have to say for me it still feels like IRL commanders would be getting reports of enemy formations that overestimated their strength perhaps as often as the reverse, even when steady lines have formed. If I'm a division commander, it suits my purposes to err on the side of over-estimating the strength of the formations facing me to my boss so that he sends me more resources. And I also know as a division commander that the worst sin is complacency and assuming the enemy is weak when he is strong.

So in sum:

1 Are we happy with the +30% mean 'display CV vs true CV' during static warfare? This may as well as go in the rulebook to help new players if so.

2 Are we happy that once front lines form, your recon will never again overestimate enemy CV?

My feeling is that if you change 2 to include overestimations, then 1 is fine.

And as I say, this is all caveated to being observed solely during Sov winter offensive of 41 (and the beginning of the Vistula scenario from the Sov side). So my data set is imperfect.

(in reply to loki100)

Post #: 23