A Modest Proposal "to tweak casualty rates"

[unmerged(19545)]

I've given some though, Math Guy et al, to what you had to say about SA vs GD, and the a-historical casualty rates we're seeing now.

Another poster had it just right, (apologies, I don't recall who, but you know who you are) when they pointed out, that the key to making a useful simulation is rigorously ensuring that each component accurately simulates some mechanism of the greater system. If you can do that, you can nuance your values, and get an adequately (however you define adequate) close match to empiracly observed outcomes. If you just womp-up some collection of equations (Lanchester's for example) that you fondly imagine will give the same curve as you've observed, you find you will only approximate reality for some limited area of the natural curve. This is usually accepted as adequate, and not just in games either. The k constant used to describe compression of a spring under load works nicely ony in the middle of the curve: at either end it's way off. It is very easy to test your model by reductio ad absurdum , testing it with extreme cases and seeing if it turns to s**t.

Well, HoI combats fails this test right now, as proved yourself with your analyses of casualty rates. What to do?

Fix the components. As it stands, most things aren't doing what they are supposed to do. Dunnigan points out in "How to make War" that in WWII only 3% of casualties were caused by small arms, 67% from artillery, and the rest by environmental attrition (the clap, and drunk driving )

So, most infantry advances should contribute to GD, not SA: things like submachineguns are for suppressing fire, not slaughter, and making the other guys leave your artillery boys alone, while they are killing them. In fact, most of these things are more effective in defense. Artillery on the other hand, kills the enemy, whether in attack or defense, though much more effectively in the latter (guys moving in the open through killing zones) So first we adjust the techs appropriately.

The second component we need to adjust, is the attack/defense percentages, now modded by weather, rivers, day/night etc. There is no particular reason why the base value of this should be 100%, after all, barring a turkey shoot like on the Finnish ice, the overwhelming majority of a division's tubes aren'y pointed at anything. I suggest we make this base value 20%, adjust all other modifiers accordingly, and we will much more quickly approach Math Guy's goal of historical casualty rates.

And while I'm ranting, I also propose that the C&C techs also raise this. The better reconaissane techs would certainly allow you to point more and more of your tubes at enemy, as you got better and better intelligence on where they are.

CRAP!! I'm late for work. Well, that is the rough outline anyway.

BtW, the real ideal thing to do, would be to change some source code, so that attacker or defender % changed the chance of achieving a hit with each point of SA, then we could really nuance our outcomes with sublety.

-from Hastur by the shores of Hali

 

[Mithel]

Lowering the basic attack efficiency is an intriguing idea. Do we have access to the basics though? Certainly we can tweak night, rain, mountain, jungle, etc but can we tweak the "non-modified" basic values? (i.e. shift it away from 100%)

I personally support raising the GD values much like Mathguy recommends. The rule of thumb for ages has typically been that you must have a 3:1 advantage to press an attack. In HoI it seems a 2:1 attack is quite effective.

- Mithel

 

[unmerged(19545)]

Good question Mithel, yes, we can adjust that starting point of 100% downwards. Not directly, but through a simple workaround. By adding to the first infantry tech commands of the sort that influence jungle or desert etc attack and defense percentages. For example, this is cut from Infantry tech and refers to the introduction of desert warfare equipment:
command = { type = desert_attack which = infantry value = 25 }

but applying the negative modifier we want to attack percentages across the board. Naturally, we would have to go through and adjust every other modifier as well, to maintain the right proportions. And that would be a little bit of work.

But as Math Guy is at pains to point out, casualty rates are too high, even when there is ample GD, so adjusting SA alone won't do the trick.

Edit!! Is it possible to change attack and defense in clear terrain? This might not work after all. Anyone out there know?

And does anyone have any comments on my other suggestion, that most techs in the infantry would contribute more to GD than SA?

Or that C+C and recon techs should increase attack and defense, by better use of existing resources?

 

[unmerged(14603)]

Originally posted by Dog Cavalry
...
Edit!! Is it possible to change attack and defense in clear terrain? This might not work after all. Anyone out there know?

And does anyone have any comments on my other suggestion, that most techs in the infantry would contribute more to GD than SA?

Or that C+C and recon techs should increase attack and defense, by better use of existing resources?

There is no modifier for clear terrain.

I look at SA,HA,GD etc somewhat more abstractly than you appear to. Balancing the combat units in HOI is very complex - even small changes in one factor or another can cause significant changes in combat outcomes. Unfortunately it is very easy to unintentionally create a unit that is "the only unit worth having" or is "not worth building at all".

For example, in land combat if you had 1 inf vs 1 inf with SA=10 and GD=10 there will be on average 2 hits per hour causing about .3 in strength damage per hour. If you lower GD by 2 (to 8) then damage increases to .54 per hour (up 80%) but if you increase GD by 2 (to 12) there is no impact on damage.

Although it certainly makes sense to link certain techs to certain modifiers (for example I don't propose linking HA to submachineguns) I prefer to take into consideration the overall "power" of the unit when modifying attributes rather than worry about specific historical links.

 

[unmerged(19545)]

Yes, Mikel, it seems you are correct. I ran a few tests last night, replacing 'desert' with 'plains' in the desert equipment tech (level 3 infantry) and I got 'illegal command' type errors, while loading the scenario. That is disappointing... for a minute there, I thought we'd be able to make loss rates whatever we (Math Guy, and others who have the historical data) thought they should be.

I suppose its possible I simply made a mess of the commands. Would some kind soul who understands what I want to achieve here be willing to duplicate my failure?

Re: abstraction - the game itself is unavoidably abstracted at many levels, but I certainly don't want to abstract the history right out of it. I want smg's, rocket artillery, and everything else you might choose to develop, to have as close to historically accurate an effect as possible. For one thing, that allows grognards who are new to the game to make intelligent research decisions.

 

[unmerged(10646)]

How about using the currently broken ground defence efficiency stat?

As I'm sure you're aware, the way it is intended to work is that it will increase the likelyhood of a single point of ground defence of blocking a single attack point. This probability starts out at 80%, and in theory should get better as you research the improvement techs.

Unfortunately, the stat gives gives your value to whoever you may be fighting and vice-versa.

So, why not just give every country in the world +15% ground defence efficency at the beginning and then let them reduce it through tech (effectively increasing their ground *attack* efficiency)?

This will create a situation where losses will be small when there is ample GD, but everything over the GD limit will remain the same. Not ideal, I know, but definitely workable.

Tim

 

[unmerged(14603)]

Originally posted by Timdog
How about using the currently broken ground defence efficiency stat?

As I'm sure you're aware, the way it is intended to work is that it will increase the likelyhood of a single point of ground defence of blocking a single attack point. This probability starts out at 80%, and in theory should get better as you research the improvement techs.

Unfortunately, the stat gives gives your value to whoever you may be fighting and vice-versa.

So, why not just give every country in the world +15% ground defence efficency at the beginning and then let them reduce it through tech (effectively increasing their ground *attack* efficiency)?

This will create a situation where losses will be small when there is ample GD, but everything over the GD limit will remain the same. Not ideal, I know, but definitely workable.

Tim

Using the ground_def_eff modifier to reduce casualties has a lot of merit.

However, if GD is made to be more important we would need to carefully balance the effect - particularly in large stack versus small stack combat. Very often large stacks will overwhelm the GD of the smaller stacks resulting in a lot of unblocked shots. If at the same time we have created a situation where few of the smaller stack's shots get through then we could be significantly favoring large stacks.