Hold at All Cost defense

[berto]

In my private testing, I have some interesting results relating to this issue. I am awaiting independent confirmation from other members of the AACW beta testing team. When and if I/we have something definitive and conclusive to share, we will do so.

Some tentative results (independently confirmed by Bigus):

Hold at All Cost defense tends to yield 3-day battles and longer with casualties of 13% or more for each side per day. Entire-battle casualty rates are, at minimum, ~40% and not infrequently even higher. (This is with open-field battles, defender entrenchment level 1.)

There might be other special circumstances that additionally jack up battle losses (a higher entrenchment level being one obvious, common circumstance). When added to the already high Hold at All Cost losses, the overall result would tend to be horrendously high losses.

And what about All Out Attack? So far, it doesn't appear to cause exceptionally high battle losses. Only Hold at All Cost defense seems to do this.

Are other players observing this?

(Caveat: There may be other circumstances that produce exceptionally, unreasonably high battle losses.)

Bloodbaths on a scale we sometimes see them in AACW are ahistorical. In the Real War:

• Most battles were 1-2 days, rarely 3 days (Gettysburg), almost never more than that (the Seven Days being the obvious exception).
• Most daily loss rates were 10-15%, with very few exceptions (Stones River, with its 25% being the noteworthy outlier).
• Where single-day loss rates were 15% or higher, those battles tended to be single-day affairs.
• Entire-battle loss rates were usually <20%, in a few instances >20%, never more than 30%.

In the Real War, single-day and entire-battle losses were limited by:

• Ammunition supplies
• Retreat
• Panic and rout
• Physical exhaustion
• Moral conscience (abhorrence of butchery)
• Risk avoidance (commanders not wanting to endanger their reputations or places in history; not wanting to chance being demoted or sacked; not wanting to gamble everything on a throw of the dice; in the Real War, generals almost never risked their entire command on a single battle, Hood at Franklin/Nashville and Lee at Antietam being notable exceptions)

For some of us players at least, horrifically bloody, ahistorically high battle losses are game breakers and spoil the AACW experience.

Tentative conclusion: Hold at All Cost is bad.

Tentative recommendation: Ways must be found to mitigate the effects of Hold at All Cost. (We are working on solutions.)

Thoughts?

[Coregonas]

Tentative conclusion: Hold at All Cost is bad.

+1


DAMAGE / COHESION RATE is needed to be changed. Review Bigus MOD?

I propose HALVING or reducing to a THIRD the DAMAGE done to the enemy, but NOT COHESION. This could be a hard work if all FILES are to be changed, but perhaps some math could help.

It is very important Forcing MORE DELAY in the battles... No more:
MESSAGE
1/20.- General x could avoid battle on Day 1 Hour 1.
2/20.- General x could avoid battle on Day 1 Hour 1.
3/20.- General x could avoid battle on Day 1 Hour 1.
4/20.- General x could avoid battle on Day 1 Hour 1.
...
16/20.- A battle taked place on... DAY 1 Hour 16!!!
17/20.- General x Routed from battle...

-----
Red Defense only is useful for losig all your troops, in fact it is the same as ORANGE, (as now it is defined)

I am only using NORMAL - DEFENSE (i.e the ORANGE) in case I have a CLEAR roughly 1,5-1 or best Defensive RATIO.

Between 1,5-1 and 2,5-1 - Blue - Defense can be good

When a 2,5-1 RATIO is expected, I usually use Green - Defense

Why?
A normal human Attacker goes against the enemy if clearly superior, he is NEVER going an attack on a 1-1 ratio, but, on a 3-1 ratio. So it is going to wipe all your troops out without pity.

Red "All OUT ATTACK" is exactly giving the same results as a NORMAL ORANGE ATTACK, except if the ATTACK is ONLY around a 1-1 ratio, in wich case is as disastrous as the Orange.

[Coregonas]

QUESTION...PROPOSAL
Total ELEMENTS involved in a BATTLE "ROUND" is needed to be REDUCED.

I believed in my first games than (when in superiority) only around X1 troops fought against the similar quantity Y1 enemy troops in each "hour" of battle, and, once the round was solved, another (different) X2 troops fought and so on.

The Y1 limit is around a division size.

Perhaps the game engine does somewhat that way

but RESULTS seems to me 4 divisions can shot in the same round to a single one.

[berto]

In an e-mail exchange, Bigus has diagnosed the problem as:

The problem with the Defend at all costs is that there is NO WILL to retreat AND no CHANCE to retreat. All these values are zero.

He has said he is working on a fix for this.

NOTE: Hold at All Cost is a statement of commander intent.

Going into battle, a commander may give the order to hold at all cost, but that doesn't mean such an order will, or can, always be obeyed. There may be practical reasons, both human and logistical, why the order won't or can't be carried out.

An order to "hold at all cost" seldom really meant "hold on and if necessary die to the very last man." It was usually just an expression, a strong statement of purpose, but not something meant to be taken quite literally.

Finally, a commander might change his mind if, facing the reality of it, the costs become unexpectedly, unimaginably high. Intent can change.

[berto]

I propose HALVING or reducing to a THIRD the DAMAGE done to the enemy, but NOT COHESION.

The trick is to reduce casualties in exceptional circumstances, not all circumstances. Most battles are fine (have reasonable casualty rates). It's the outliers we want to impact.

[Zebedee]

Is there a possibility of having such a defence as an exceptionally (and I mean exceptionally) rare occurance?

I totally agree with your proposals, just think that it should still be 'possible' if ludicrously unlikely. If that makes sense

[berto]

Is there a possibility of having such a defence as an exceptionally (and I mean exceptionally) rare occurance?

I totally agree with your proposals, just think that it should still be 'possible' if ludicrously unlikely. If that makes sense

I think that Bigus is working on a solution allowing for that.

[Coregonas]

Hold at All Cost defense tends to yield 3-day battles and longer with casualties of 13% or more for each side per day. Entire-battle casualty rates are, at minimum, ~40% and not infrequently even higher. (This is with open-field battles, defender entrenchment level 1.)

Can I know... What was the RATIO of FORCES involved in those tests?

In my own experience, a NORMAL well trenched troop (level 4) defending (ORANGE/ORANGE) against roughly 3-1 odds, usually finishes with a similar quantity of men lost, but the % of loses is some thing like 25-35% to the attacker and 60-90% to the defender...

I strangely concluded attacking is SAFER than defending if you are a prudent boy, while the reverse seems not true.

[berto]

I conducted 48 tests on 12 different Shiloh test variants, where in four of the variants, forces were roughtly 1:1 attacker to defender; were < 1:1 a-to-d, in four other variants; and were > 1:1 a-to-d in the four remaining variants. (Testing on 2:1, 3:1, 1:2, 1:3, etc. are planned for the future.)

In all 48 tests, the South was the attacker, and the North the defender. (Future tests are planned where the roles are reversed.)

(Setting up good, unimpeachable test cases is hard, time-consuming work. Actually conducting the tests runs fairly quickly.)

In half the test cases, it was All Out Attack; in the other half, it was Normal Attack. The attacker mode had little bearing on the outcomes. It was clearly the Hold at All Cost defense mode that yielded the excessive battle casualties (both sides, attacker and defender).

The results

Hold at All Cost defense tends to yield 3-day battles and longer with casualties of 13% or more for each side per day. Entire-battle casualty rates are, at minimum, ~40% and not infrequently even higher.

were observed in 8 out 8 test games where force ratios were 1:1, 6 out of 8 test games where ratios were < 1:1, and 2 out of 8 test games where ratios were > 1:1.

In other words, in 16 out of 24 test games, the quoted results were observed.

In other cases with Hold at All Cost defense, 6 out of the 24 total, the battles were two-day affairs with casualties still ahistorically high or right at the historical "limit". In only 2 out of 24 test games were the hypothetical Shiloh battles one-day affairs, but with casualties still somewhat high, 12-18%, single day.

Summarizing: Across all Hold at All Cost test cases, battle losses were quite high; and excessively, ahistorically high in by far most of the test cases.

[soundoff]

And what about All Out Attack? So far, it doesn't appear to cause exceptionally high battle losses. Only Hold at All Cost defense seems to do this.

Without knowing what your attack and defending armies consist of I really cant comment Berto. Unless the battles both for the when testing all out defence and when testing all out attack are evenly matched then an impartial result is surely impossible. But then I suspect that you are using evenly matched forces

[berto]

Without knowing what your attack and defending armies consist of I really cant comment Berto. Unless the battles both for the when testing all out defence and when testing all out attack are evenly matched then an impartial result is surely impossible. But then I suspect that you are using evenly matched forces

They were evenly matched, or nearly so. Johnston's army versus Grant's. (I took Buell's army out of the picture in order to achieve rough parity.) Similar mix of forces.

(And without exacerbating circumstances like high entrenchment levels.)

One can always quibble and nitpick about the details. But nitpicks pale in comparison to this: excessively, ahistorically high battle casualties should not occur in these Shiloh test variants, in (the earlier tested) Gettysburg variants, or any other battles (of significant size) in AACW.

There were hundreds and hundreds of significant engagements and dozens of large- to mid-size battles in the Real War, under all sorts of quibbly and nitpicky circumstances, and in all cases the Real War battle results were within the historical limits previously stated.

I assert that, quibbles and nitpicks aside, the same should generally, virtually always even, be true in AACW.

[Daxil]

I believe the hold at all cost option should not be changed when defending state/national capitals and depots, but yes, agree at some point they should break elsewhere and it should be far before 90% losses or whatever it is now. The reason I say state/national capitals is it would seem to me the troops would fight much more savagely if told to when defending national icons and very important depots. Look, if the fate of the war depends on it,they will die. Maybe it should be based on cohesion, but I see nothing wrong in those cases.

EDIT Just to add some more thoughts, not only cohesion, but the strategic rating of the leader, which would represent his ability to cnvince the soldiers they need to die.

This is not so much a Civil War question as a human nature question imo. There are plenty of cases in history where warriors have fought to near annihilation.. Thermopylae, Berlin, Stalingrad, Gallipoli for the offensive come to mind. Again national icons/very important objectives coupled with the will to die.

Maybe the button should be greyed out unless your attacking/defending one of these.

[bigus]

Confirmed for sure Berto!

I've run all Berto's test scenarios and a couple more of my own several times.
The conclusion...."Defend at all costs" is at least one reason for high casualties.

A fix?
I'm not sure yet but this has to be looked at.........
I have some working files that seem to be O.K ... as for the long term effects of them (ie:GCG)....I'm not sure.
This seems to be one more cause of unrealistic battle casualties that has to be looked at.

[Coregonas]

Some other thread concluded than TOO HIGH COHESION in the troops avoided routing, so wiped totally elements.

An infantry with 2 * experience (+10 cohesion each?)
with an elite (+5)
a good general (i e Jackson) +5
perhaps high NM or NMC adds another extra...

So finally lots of troops with 100 cohesion, have no time to drop to the proper level for rout.

Its the same thing... DAMAGE / COHESION Ratio is too high.

Im not sure now, but can... some MALUSES in an attack can drop a unit with just 1 FIREPOWER drop to 0? So no damage done to the enemy? (Ask this due to the "round to the nearest affair) i.e. a -51% maluses to firepower drops 1 to 0,49 and so 0???

[berto]

I believe we are surely justified to root out and eliminate the excessively high battle casualties--in certain situations. In general, though, battle losses seem to be fine. In my tests, I've been impressed by the numbers and battle durations in most instances (excepting Hold at All Cost defense, which I've identified as one situation causing excessive losses and overly long battles).

In our efforts to fix this problem, however, we must be careful not to create another problem--ahistorically, unrealistically low battle casualties (and/or too-brief engagements).

The crux of the problem, as I see it, is in the exceptions. The battle losses mechanism appears to be fundamentally sound, just broken in the high-side outliers. Identifying those exceptional cases is key.

[Coregonas]

Some idea to solve this.

The question is ... I believe we PLAYERS can not ORDER GENERALS to suicide themselves. The point of view of the General is for sure more prudent than we the players.

By now there are 4 "postures" i.e. colours
Green -> Run before start -> Used to avoid battle

Blue -> Run at the 2nd-3rd round? A TOO prudent choice, from the point of view of the PLAYERS

Orange -> The Default -> Gives HIGH DAMAGE results but Realistic (from the point of view of the players) There are big casualties also here, realistic in fact.

Red -> NEVER RUN -> TOO HIGH DAMAGE -> Unrealistic


What about CHANGING the RULES to:
Green -> As NOW
ORANGE -> As Now BLUE (or with an extra round...i.e per default, some more prudency)
RED -> As NOW ORANGE (perhaps delaying it a bit, forcing 1 or more extra rounds, but never disallowing a retret)
Blue -> Some new rule on the middle. i.e. Fight only 1 or 2 rounds

[Jabberwock]

What about CHANGING the RULES to:
Green -> As NOW
ORANGE -> As Now BLUE (or with an extra round...i.e per default, some more prudency)
RED -> As NOW ORANGE (perhaps delaying it a bit, forcing 1 or more extra rounds, but never disallowing a retret)
Blue -> Some new rule on the middle. i.e. Fight only 1 or 2 rounds

Attempted retreat at 0, 2, 4, and 6 rounds sounds initially like a very good idea.

[berto]

Attempted retreat at 0, 2, 4, and 6 rounds sounds initially like a very good idea.

Jabberwock, Bigus, Gray (others?): Do you intend/have an idea how to implement this? Might we try it out as beta testers?

[Jabberwock]

It's handled by the engine ... up to Pocus if he wants to create a beta using that rule, or export some more variables.

EDIT:
bigus - +1
Jab - +0

[bigus]

Jabberwock, Bigus, Gray (others?): Do you intend/have an idea how to implement this? Might we try it out as beta testers?

You might be able to do it in the ROE.Opt file.
For each round you don't want a retreat just set the numbers to Zero.
For each round you want to retreat set them to some variable.
You can do this for each stance and for each round.

This is one reason IMO 'Defend at all costs" is resulting in higher than normal casualties. If you look at the "Will to retreat" and the "Chance to retreat", they are Zero for each round! Right now I've changed it so you have a slight chance to retreat for rounds 4-6 for "defend at all costs". It's giving O.K results for me. I'm not sure if Berto has tested this out yet though.

[berto]

It's giving O.K results for me. I'm not sure if Berto has tested this out yet though.

No time yet. Is this based on the file you sent me last night?

[Jabberwock]

Sure enough, there it is! Thank you bigus!

I've change the retreat values in this file for hold at all cost on the 5th and 6th rounds to 80/60 and 90/70 for testing purposes. It may be too much of an adjustment. I've posted it here, so that anyone can try it, not just official betas. The more feedback we can get on this, the better.

[berto]

Sure enough, there it is! Thank you bigus!

I've change the retreat values in this file for hold at all cost on the 5th and 6th rounds to 80/60 and 90/70 for testing purposes. It may be too much of an adjustment. I've posted it here, so that anyone can try it, not just official betas. The more feedback we can get on this, the better.

I really think you guys are onto something here. I think that Bigus has definitely put his finger on the cause of the Hold at All Cost problem. (That said, I think there are still other situations that need fixing, situations that don't involve Hold at All Cost defense. Bigus has identified problems with ZOC, for instance.)

I just conducted a round of tests--12 test battles of my Shiloh variants. The good news is that the 3-day battles are gone. The bad news is that I think you maybe went too far in limiting durations, as now just my 1:1 variants give two-day battles, while the rest of the variants give one-day battles.

For the 2 two-day, 1:1 battles, entire-battle casualties were in the range of near 30% for both sides. For the 4 one-day battles with Hold at All Cost defense, the entire-battle and single-day casualties were 12-15%.

Might you back it off a little? Casualties are now more within reason, but I'd expect to see more two-day battles, fewer one-day battles. (In this latest round of tests, again, there were 2 two-day battles and 4 one-day battles involving Hold at All Cost defense.)

The percentage losses are good; it's just that the durations are maybe a bit too short. (The trick is: how do you lengthen durations while keeping casualties capped at no more than ~30%?)

I don't have time for any more tests today, but I'd like to pursue this further tomorrow, hopefully with a slightly tweaked ROE file from you. (I'd make tweaks myself, but I really don't know what I'm doing with the tweaks.)

[Jabberwock]

Here is a version with 5th round back at 0/0 and 6th at 80/60. Hopefully this brackets the desired effect. It's easy to adjust again in any direction.

[berto]

Here is a version with 5th round back at 0/0 and 6th at 80/60. Hopefully this brackets the desired effect. It's easy to adjust again in any direction.

Thanks! I will try out this new ROE file tomorrow.

This whole approach is well worth pursuing.

[Coregonas]

Noy sure if telling something stupid, dont know exactly what those values mean, but looking into these file ROE.opt, seems as if we can slow the FIREPOWER done in every round...

What about if... In the last "round" of all those options the FIREPOWER done is tremendously slowed in all these options (i.e to a 20% instead of a 100?)

Else... add, some more rounds seem posible as the explanation seems to allow ->

Adding some more rounds (7th,8th or more) with every one of those extra rounds reducing the firepower done in all the options.

[berto]

What about if... In the last "round" of all those options the FIREPOWER done is tremendously slowed in all these options (i.e to a 20% instead of a 100?)

If what you suggest is truly possible, it's an interesting proposal, for it would model the very real prospect of ammunition depletion after earlier hard fighting.

[berto]

Here is a version with 5th round back at 0/0 and 6th at 80/60. Hopefully this brackets the desired effect. It's easy to adjust again in any direction.

Your work on this, and Bigus' (also something else I can't detail publicly) motivate me to implement a "proper" battle test suite this weekend. If everything goes well, maybe I'll soon have something I can share with you all.