Future of CEAW

[Cybvep]

So, with the release of 2.1 coming shortly, are there any preliminary plans for the future? If new bugs are found in 2.1, will they get fixed? Will there be the 2.2 or 3.0 version(s)?

I'm not demanding anything, just asking .

[Peter Stauffenberg]

Not sure yet. We will fix bugs for sure. There are no current plans for changes we will begin to make shortly.

It could be that Ronnie will have a look at the AI and we will focus on that. First all of us in the development team want a rest so we can just enjoy playing and nothing else. Then we will see what happens.

[ncali]

Turning the flawed odds calculator into a real combat results averager (per thread in forum) would be top on my list! Don't know how difficult this would be though. I'm not volunteering - but I've been amazed at the programming skills of Stauffenberg and others!

[Peter Stauffenberg]

One thing I can't understand is why it's working well at most odds, but not when antitank value is very high.

I might at some time have a look at how the odds are calcuated and compare it to the real combat results. There must be something different somewhere. I can't see anything that sticks out so the only way to find out is to make an intensive debug and hope I can find the needle in the haystack. What I need to find is the reason the odds are so much off.

I've seen the same in the vanilla game so it's definitely not something we have introduced. I suspect the culprit is the antitank value. It usually battles against units with high antitank values that you see poor odds predictions. Is that your experience too?

[Cybvep]

I see several potential changes that could make it to 2.2/3.0.

1. Making Suez/Gibraltar very important, so that capturing either of them will hurt the British PPs by lowering the size/spawn frequency of the southern convoy. We could also think about a small morale loss.
2. Introducing more serious consequences for giving up Libya in 1940/1941 without a fight.
3. Changing the way Barbarossa plays out, especially during the first year of the offensive. Less running, more fighting, more encirclements. This includes things like the transfer of factories to Siberia and efficiency drops for losing important cities when playing the Soviets. The current situation is balanced for the current rules, but it is as far from historical reality as possible.
4. Changing the logistical system to make it a bit more advanced. The most pressing issue is the local concentration of armoured/mechanised units. It's far, far too easy to do that in CEAW when compared to RL. Supply issues were a major limiting factor of force concentration IRL.
5. Implementing late-war submarines. Currently subs get shredded during late-war, but IRL the Germans actually developed subs that could be considered the first "real" submarines, as they were able to stay underwater for a relatively long time.
6. Here is a crazy idea of mine - give us the ability to play the game from 1939 to Feb/Mar 1940 faster. In most games there is really not that much to do during this period and if you are playing at a steady but slow pace, e.g. 1 turn/day, it can take a long time to finally start Case Yellow. An editor of sorts would be cool, I think. You could decide how to spend your PPs and focus points, place units etc. quickly. Just a thought, I have no idea whether it would be doable.

[ncali]

One thing I can't understand is why it's working well at most odds, but not when antitank value is very high.

I might at some time have a look at how the odds are calcuated and compare it to the real combat results. There must be something different somewhere. I can't see anything that sticks out so the only way to find out is to make an intensive debug and hope I can find the needle in the haystack. What I need to find is the reason the odds are so much off.

I've seen the same in the vanilla game so it's definitely not something we have introduced. I suspect the culprit is the antitank value. It usually battles against units with high antitank values that you see poor odds predictions. Is that your experience too?

My experience is that it's more flawed than just combat involving tank units. I think it becomes more unreliable when higher efficiency land units are involved in the combat, particularly high-efficiency defenders. This is more common in the late game. That's why I think (and I'd have to review original rules explaining this) that it is an adjusted strength comparison (it does adjust unit strengths for terrain, entrenchment, etc.). In other words, it doesn't actually go through the combat sequence of attacker suppresses defender, defender fires first, than attacker fires.

[rkr1958]

One thing I can't understand is why it's working well at most odds, but not when antitank value is very high.

I might at some time have a look at how the odds are calcuated and compare it to the real combat results. There must be something different somewhere. I can't see anything that sticks out so the only way to find out is to make an intensive debug and hope I can find the needle in the haystack. What I need to find is the reason the odds are so much off.

I've seen the same in the vanilla game so it's definitely not something we have introduced. I suspect the culprit is the antitank value. It usually battles against units with high antitank values that you see poor odds predictions. Is that your experience too?

My experience is that it's more flawed than just combat involving tank units. I think it becomes more unreliable when higher efficiency land units are involved in the combat, particularly high-efficiency defenders. This is more common in the late game. That's why I think (and I'd have to review original rules explaining this) that it is an adjusted strength comparison (it does adjust unit strengths for terrain, entrenchment, etc.). In other words, it doesn't actually go through the combat sequence of attacker suppresses defender, defender fires first, than attacker fires.

What I would like to have is the ability click on a friendly unit and then say right click on any spotted enemy unit that this friendly could move adjacent to and have a combat simulator than would run a number of battles and give you average, min and max results from those simulated battles.

[Kragdob]

One thing I can't understand is why it's working well at most odds, but not when antitank value is very high.

I might at some time have a look at how the odds are calcuated and compare it to the real combat results. There must be something different somewhere. I can't see anything that sticks out so the only way to find out is to make an intensive debug and hope I can find the needle in the haystack. What I need to find is the reason the odds are so much off.

I've seen the same in the vanilla game so it's definitely not something we have introduced. I suspect the culprit is the antitank value. It usually battles against units with high antitank values that you see poor odds predictions. Is that your experience too?

My experience is that it's more flawed than just combat involving tank units. I think it becomes more unreliable when higher efficiency land units are involved in the combat, particularly high-efficiency defenders. This is more common in the late game. That's why I think (and I'd have to review original rules explaining this) that it is an adjusted strength comparison (it does adjust unit strengths for terrain, entrenchment, etc.). In other words, it doesn't actually go through the combat sequence of attacker suppresses defender, defender fires first, than attacker fires.

I agree completely.

Casualties are generally lower than indicated both on attacker and defender side. Maybe it is just not including effectiveness? But on the other side it can be observed in naval battles too. It pisses me off when I get 7:0 odds against transports and average result is 4:0. Does transport lower effectiveness of battleship so much?

[Plaid]

Also air attacks vs BB/DD give you usually great odds as 5:0, while in fact average result is like 2:1 or 1:1 even.

And in fact calculator is doing something wrong since the very beginning of game (maybe not so dramatic, but still).

I have just runned a hotseat test in 1939 scenario :



Odds are 5:2.

Actual results were (first number - german casualties, second one - polish).
2:3 2:2 1:2 2:3 2:3 2:2 2:2 2:4 3:2 1:3
None of ten samples showed promised 2:5 result.
Average result 1,9:2,6 which is more like 4:5.

So if anything in the game worth fixing, it is this calculator.
You have only 1 try when launching attack somewhere, and I think you deserve to have some more or less reliable information about how succesful attack can be (panzer general/korps - style for example).

People who play the game for a long time ofcourse have some experience about relative unit strength and dont have to rely on this calculator, but still it annoys, that thing show results which never happen.

[Diplomaticus]

A small but potentially significant thing: IMO troop transports are too resistant vs. air attacks. I saw two different cases of this over the last months:

1) mid to late game, my Axis was defending against repeated Allied amphibious moves vs. Sicily/Italy. In multiple actions, it was a very common thing (maybe 50% of attacks) for Luftwaffe fighters and Strategic Bombers to take 1 point of damage and to do only a very few points of damage to the transport (1-3). My airforce was very technically advanced (some of these results were with maxed-out tech) and operating at high efficiency in supply 5.

2) very early game, my Allies trying to fend off Operation Sealion. Again, British fighters and strat bombers took 1 point of damage in approx. half of their attacks on the German transports, and in turn they did very little damage to the transports--most often only 1 point.

I suggest that in the future we do a little historical research and find out a) just how vulnerable transports were vs. fighters or strats and b) how effective those transports or their minor escorts* were in inflicting casualties on the planes.

[*Maybe if troop transports are adjacent to major surface fleet units we consider the fleet to be escorting the transports and up their defense?]

[Cybvep]

I think that transports are too resistant to enemy attacks in general. Escorts are there for a reason - a transport should be decimated by a SAG or a CTF.

[rkr1958]

I think that transports are too resistant to enemy attacks in general. Escorts are there for a reason - a transport should be decimated by a SAG or a CTF.

You have to realize that a transport represents 50,000 men or 500 tanks. So if an air attack knocks of 3 steps then that's a loss of 15,000 men or 150 tanks. If a second air attack knocks off 2 more steps that's a loss of 25,000 men or 250 tanks. So, personally I think we've got this part about right.

[rkr1958]

Turning the flawed odds calculator into a real combat results averager (per thread in forum) would be top on my list! Don't know how difficult this would be though. I'm not volunteering - but I've been amazed at the programming skills of Stauffenberg and others!

I just thought of a possible near-term solution to this and one in which the entire community could participate. We could develop an off-line odds calculation (java or spreadsheet) based on a multiple linear regression fit to the key variables. If we came up with the equation (i.e., the variables in the equation that impact loss) then folks could record those variables and the resultant losses. With enough samples given that we have the right set of independent variables then we could do a regression fit and have equations that give us average loss and upper and lower error bound. The advantage of this type of calculator is that you could run the odds before you moved units. Anyone game?

[Plaid]

You have to realize that a transport represents 50,000 men or 500 tanks. So if an air attack knocks of 3 steps then that's a loss of 15,000 men or 150 tanks. If a second air attack knocks off 2 more steps that's a loss of 25,000 men or 250 tanks. So, personally I think we've got this part about right.

But we should also keep in mind that air unit have ~500 airplanes and turn is 20 days, so they do much more then 1 mission attacking this vessels.
There is quite little historical data about bombers sinking troop transports (allies launched their invasions with very good protection and total air superiority) , so this is some abstraction of game engine.

Personaly I have no problem at how things work now. In my recent game in 1944 I have gathered my german TAC units in Berlin area to hide them from allied fighters, since I had no chance to provide them fighter cover and use them in action. And when allies tryed diversional landing near Kiel I sunk entire transport with 2 attacks from that depleted bombers (5-7 steps). Next turn another transport appeared and I did it again. Fine for me.
Still its oftem MUCH more hostile transports then your bombers.

[TotalerKrieg]

I have a feeling that I am alone on this, but I actually think there shouldn't be an odds calculator at all. I also think you shouldn't be able to see the tech levels of the enemy when you put the mouse on them. That would be more realistic IMO for a war game. All offensives should carry some risk for the attacker, and if you already know what the outcome of an attack will be before you do it, then there is very little risk.

[Kragdob]

I have a feeling that I am alone on this, but I actually think there shouldn't be an odds calculator at all. I also think you shouldn't be able to see the tech levels of the enemy when you put the mouse on them. That would be more realistic IMO for a war game. All offensives should carry some risk for the attacker, and if you already know what the outcome of an attack will be before you do it, then there is very little risk.

Yeah, with real FOW you should see only unit type and nothing except. Stats should reveal the more turns adjacent to the unit/ more attacks made.

This would be nice and very realistic.

With stats visible it would be great to be able to click on my unit and than see (reliable) odds for any enemy unit I hover the mouse over.

I tried to make such simulator in excel but I have too little data. Can't this be done by combat code reverse engeneering?

[Kragdob]

You have to realize that a transport represents 50,000 men or 500 tanks. So if an air attack knocks of 3 steps then that's a loss of 15,000 men or 150 tanks. If a second air attack knocks off 2 more steps that's a loss of 25,000 men or 250 tanks. So, personally I think we've got this part about right.

But we should also keep in mind that air unit have ~500 airplanes and turn is 20 days, so they do much more then 1 mission attacking this vessels.
There is quite little historical data about bombers sinking troop transports (allies launched their invasions with very good protection and total air superiority) , so this is some abstraction of game engine.

Personaly I have no problem at how things work now. In my recent game in 1944 I have gathered my german TAC units in Berlin area to hide them from allied fighters, since I had no chance to provide them fighter cover and use them in action. And when allies tryed diversional landing near Kiel I sunk entire transport with 2 attacks from that depleted bombers (5-7 steps). Next turn another transport appeared and I did it again. Fine for me.
Still its oftem MUCH more hostile transports then your bombers.

Transport should be sunk (with normal luck) by a single vessel of any type. Definitely if no friendly escort adjacent.

[rkr1958]

You have to realize that a transport represents 50,000 men or 500 tanks. So if an air attack knocks of 3 steps then that's a loss of 15,000 men or 150 tanks. If a second air attack knocks off 2 more steps that's a loss of 25,000 men or 250 tanks. So, personally I think we've got this part about right.

But we should also keep in mind that air unit have ~500 airplanes and turn is 20 days, so they do much more then 1 mission attacking this vessels.
There is quite little historical data about bombers sinking troop transports (allies launched their invasions with very good protection and total air superiority) , so this is some abstraction of game engine.

Personaly I have no problem at how things work now. In my recent game in 1944 I have gathered my german TAC units in Berlin area to hide them from allied fighters, since I had no chance to provide them fighter cover and use them in action. And when allies tryed diversional landing near Kiel I sunk entire transport with 2 attacks from that depleted bombers (5-7 steps). Next turn another transport appeared and I did it again. Fine for me.
Still its oftem MUCH more hostile transports then your bombers.

Transport should be sunk (with normal luck) by a single vessel of any type. Definitely if no friendly escort adjacent.

I disagree. If you make transports that vulnerable then you would need to completely cover all adjacent hexes moving them. That would mean 6 escorts for one transport or 9 escorts for two. If, for example, you tried to escort a transport with 4 DD's leaving two gaps, then I certainly would trade 7 or 8 steps of damage to my sub to sink it.

Stepping back, I have to ask what is broken?

[rkr1958]

I tried to make such simulator in excel but I have too little data. Can't this be done by combat code reverse engeneering?

Maybe; but wouldn't it be fun as a community to generate the data and develop such a set of regression equations.

[Plaid]

If you ask me I would vote to remove entire roll/odds system and make combat prediction 100% accurate (you got what is written there, maybe +-1).
Its corps level game afterall, you generals should be able to estimate casualties , 2 steps considered 10000 men, its entire division.

Same goes for research.
In my last 2 games with totally same lab build up in one game I got some tech april 1941 and in another game - very same tech in may 1942. Cool spread, what can I say.

I think that this randomness, which player can influence in no way, is totally BAD thing in long and high level game player vs player.

(Leave alone the fact that roll randomness encourage cheating. Also if you get bad results and opponent get good ones, you might start to suspect him cheating, while in fact its all work of stupid RNG)

Randomness may sound cool (wow its like in real life!) but it break games and ruins experience.

Just my thoughts, its not like I say "please do it".