Wargames and Randomness

[GaryChildress]

How random and unpredicatable is war? My belief is that victory or defeat in war, especially on the macro scale is USUALLY attributable to certain predictable factors added together and not so much to pure randomness.

Take Midway: was Midway a fluke or were the Allies pretty much destined to win based upon things like carrier design, damage control effectiveness, the ability to warm up aircraft engines below deck and things of that nature? If Midway could be "reloaded" 10 times, how many of them would have been victories for the Allies regardless? Was Midway really as much a "lucky break" for the Allies as is commonly believed?

I've been thinking about randomness as it is commonly used in strategic scale computer games and the more I think about it, the less sense it makes to me to introduce radically divergent, random outcomes to combat. WWII was largely a numbers game. Germany and Japan lost because in the aggregate they failed the numbers game (both in quality and quantity), not because they were unlucky in combat. Success in computer simulated combat should be mostly attributable to who has the better quality or quantity and who uses their forces more wisely, not to blind luck of the die roll.

Maybe I'm wrong but it seems to me that computer games that rely on wildly random combat results are inherently flawed to some extent.

[gajdacs zsolt]

My belief is that there is no such thing as randomness. You just need to sufficiently understand the system to see how it works.

The 'problem' is not the machine, it's the man inside. You can have well built and thought out carriers if the crew uses the available tools improperly (sinking of Taiho for example).
Man happens to be a quite complex system. I believe that on a certain level it can be understood and predicted, but! There's a thing called free will. And that's where the 'real problem' starts. I don't think there will ever be machines that are able to tackle that one.

So in my mind randomness=free will

On the other hand I also believe that current games use random elements for things that the computers are not fast enough to model/calculate.

(Was unsure about posting this, I believe this to be a philosophical 'problem'...)

[herwin]

Very philosophical. Combat may not be random, but you can't really model it. "A model is an abstract representation of a system that provides useful estimates for parameter values of interest. Not all non-linear systems can be modelled; some have behaviour that cannot be predicted by any model." Combat is such a system. (So is climate...)

[7th Somersets]

Gary,

I suppose that you are right on a strategic level. To a man in the trenches in WW1 - whether a shell or bullet strikes home would appear 'random'. To him it probably is, but as you say - the real overall effect is probably more statistical so far as a battle is concerned...

There are a few times though when the actions of individuals may have a massively disproprotionate effect... and where luck plays its part. Look at the attempt to assasinate Hitler with the bomb... there are also many examples in combat where one or two people may have affected the whole battle - whether that would affect strategic results is another question...

[GaryChildress]

ORIGINAL: Zsolo007

My belief is that there is no such thing as randomness. You just need to sufficiently understand the system to see how it works.

The 'problem' is not the machine, it's the man inside. You can have well built and thought out carriers if the crew uses the available tools improperly (sinking of Taiho for example).
Man happens to be a quite complex system. I believe that on a certain level it can be understood and predicted, but! There's a thing called free will. And that's where the 'real problem' starts. I don't think there will ever be machines that are able to tackle that one.

So in my mind randomness=free will

On the other hand I also believe that current games use random elements for things that the computers are not fast enough to model/calculate.

(Was unsure about posting this, I believe this to be a philosophical 'problem'...)

Free will is a bit different from what I am arguing. Basically my belief is that, especially on the aggregate, combat results shouldn't introduce wildly divergent outcomes. In other words, lets say I was playing a grand strategic game at the scale to where the outcome of World War II represented a single roll of a die. How much variation should be introduced? Was an Allied victory in WWII EVER a question mark? Should a die roll (a purely random variable) even be invovled versus having a computer tabulate advantages and disadvantages and give a result based upon who clearly possessed the advantage?

Granted, there are a lot of factors involved and often random number generation is used in lieu of possessing the ability to calculate all the pertinent factors involved in a predictable result. The bad thing about this is that, on average, one out of every X times I roll the die, Japan and Germany could win WWII. If I'm rolling a 6 sided die then at least 1 in 6 times Germany and Japan could win the war. But is this a reasonably true result?

[DOCUP]

Gary

These are my thoughts on the matter. No ill feelings to anyone on what I type here.

We will never know if combat is random. We will never get a redo (ie never have the same exact people, weapons, weather, plans experience etc in a second battle). I myself have gotten mad at the computer over the randomness of some battles.

In my belief, there is more than just quality and quanity in battle. I would like to believe in quality, leadership, training, espirt de corp, experience, and something that is call Murphy's Law. History has shown many battles and wars won that contridict these thoughts. Ie Vietnam, Russian Afghan war, American Rev war under trained, under quality troops etc beat the well trained, equiped troops of there enemys.

As Herwin stated you cannot simulate combat, whether its in a video game, training mission. To many variables that can't be determined until the bullets start flying. It is then that leadership, training, experience, heroism (espirt de corp) and quality of equipment, tatics, murphy etc will decide the outcome of the battle and war.

A long post short. There are variable that can't be programmed because of them being unknown or unable to be put into the program.

Doc

[JeffroK]

IMH Undereducated O, the further up the food chain you get the less random events are.

Platoon level, lots of random events, little things are important.

As you work through Company, Battalion levels the little things are less important and results become consistat.

Get up past Division level and the randomness is minimal and a consistency of results occurs.

The same goes with time, review 1 action and planty of random results, look at 100 and they fall into "an average" result range

Look in the game, complaints about a single a-a or surface battle.  But in the long run the results even out and "random" isnt affecting it, its more the quality and quantity as DOCUP mentions.

Within a game, I'm all for small numbers of random results & events, just a quick look at history shows occasional results that you just couldn't explain.

[GaryChildress]

ORIGINAL: DOCUP

Gary

These are my thoughts on the matter. No ill feelings to anyone on what I type here.

We will never know if combat is random. We will never get a redo (ie never have the same exact people, weapons, weather, plans experience etc in a second battle). I myself have gotten mad at the computer over the randomness of some battles.

In my belief, there is more than just quality and quanity in battle. I would like to believe in quality, leadership, training, espirt de corp, experience, and something that is call Murphy's Law. History has shown many battles and wars won that contridict these thoughts. Ie Vietnam, Russian Afghan war, American Rev war under trained, under quality troops etc beat the well trained, equiped troops of there enemys.

As Herwin stated you cannot simulate combat, whether its in a video game, training mission. To many variables that can't be determined until the bullets start flying. It is then that leadership, training, experience, heroism (espirt de corp) and quality of equipment, tatics, murphy etc will decide the outcome of the battle and war.

A long post short. There are variable that can't be programmed because of them being unknown or unable to be put into the program.

Doc

Granted IRL no event can be recreated perfectly. However, computer games will ALWAYS try. The question is not whether an historical event CAN be recreated or not but one of HOW will it be recreated. Do you use randomness to simulate combat and how much randomness should be used? I remember all the complaints about Civilization I and how a Phalanx sometimes defeated a Tank. Yes there are seemingly random, unpredictable factors that play in combat but how much randomness is too much? Should the Japanese occasionally win at Midway? Should the Germans occasionally conquer the Soviet Union? If so, how often?

It seems to me, especially on the macro scale, where you factor in millions of random events, that things should average out to where the results largely coincide with predictions. It drives me crazy when a game predicts 2:1 results and NEVER produces them, instead producing 1:2 or 5:1 or 1:5 or whatever.

Commander Europe at War does this and that is what got me started originally on this question. I realize this is not the CEAW forum but I've heard similiar arguments here about how likely it is that Japan would have invaded Australia, etc. And since the people here seem to be, on average, more knowledgable about WW2 than the CEAW forum I thought I would bring it up here.

[PresterJohn001]

ORIGINAL: Gary Childress

How random and unpredicatable is war? My belief is that victory or defeat in war, especially on the macro scale is USUALLY attributable to certain predictable factors added together and not so much to pure randomness.

Take Midway: was Midway a fluke or were the Allies pretty much destined to win based upon things like carrier design, damage control effectiveness, the ability to warm up aircraft engines below deck and things of that nature? If Midway could be "reloaded" 10 times, how many of them would have been victories for the Allies regardless? Was Midway really as much a "lucky break" for the Allies as is commonly believed?

I've been thinking about randomness as it is commonly used in strategic scale computer games and the more I think about it, the less sense it makes to me to introduce radically divergent, random outcomes to combat. WWII was largely a numbers game. Germany and Japan lost because in the aggregate they failed the numbers game (both in quality and quantity), not because they were unlucky in combat. Success in computer simulated combat should be mostly attributable to who has the better quality or quantity and who uses their forces more wisely, not to blind luck of the die roll.

Maybe I'm wrong but it seems to me that computer games that rely on wildly random combat results are inherently flawed to some extent.

The game will end with the defeat of Japan. In that respect the ultimate outsome is not random.

The outcome of individual actions that go to make up this ultimate defeat are not random. They are influenced by random factors which represent all that cannot be controlled by the player, but are based upon the relative strengths of the sides. Given enough actions, which is almost certain in almost any wargame, the "luck" will even out. Factors that don't even out are the ability of the players and the inherent imbalances in the game (built in).

[LargeSlowTarget]

ORIGINAL: Gary Childress
Take Midway: was Midway a fluke or were the Allies pretty much destined to win based upon things like carrier design, damage control effectiveness, the ability to warm up aircraft engines below deck and things of that nature? If Midway could be "reloaded" 10 times, how many of them would have been victories for the Allies regardless? Was Midway really as much a "lucky break" for the Allies as is commonly believed?

The Japanese battle plan for Midway was seriously flawed and the USN had the benefit of good intel, so the outcome could be considered predictable.

But there were many elements of luck, fate, randomness, divine intervention or whatever you may call it. One example is McClusky's decision to continue the search for the carriers despite being low on fuel, which led to the discovery of DD Arashi which in turn pointed McClusky to the Japanese carriers. If McCLusky had decided to return to base earlier - or if the CO of Arashi had not abandoned the hunt for Nautilus when he did - or if Nautilus hadn't even been there where she was, or if a rain squall had obscured Arashi at the most inopportune moment etc., McClusky would have missed the Japanese CVs and the battle might have turned out differently. Things like this are unpredictable.

I have read somewhere that the USN did re-fight the battle of Midway in "sandbox" maneuvers several times and has never been able to achieve the historic results.

[KenchiSulla]

Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat.
Sun Tzu

Did the Japanese have a viable strategy?

In war and life there is no randomness, there are just a lot of things we havent foreseen (or taken/given the time to think the situation through - time pressure)

[PaxMondo]

Regarding the OP .... what is certain is the Gary (Grigsby) is a big beleiver in randomness and always has been.  I would characterize it by saying it is his hallmark as compared to other developers.  Not to suggest other developers don't use randomness, they do, simply not to the extent that Gary does.
 
As to reality ... there are any number of major historical events that sure seem to be pure chance.  My favorite is Mao's escape.  So, there seems to be at least some basis for high randomness at even high levels of conflict.

[LoBaron]

ORIGINAL: Gary Childress

How random and unpredicatable is war? My belief is that victory or defeat in war, especially on the macro scale is USUALLY attributable to certain predictable factors added together and not so much to pure randomness.

Take Midway: was Midway a fluke or were the Allies pretty much destined to win based upon things like carrier design, damage control effectiveness, the ability to warm up aircraft engines below deck and things of that nature? If Midway could be "reloaded" 10 times, how many of them would have been victories for the Allies regardless? Was Midway really as much a "lucky break" for the Allies as is commonly believed?

I've been thinking about randomness as it is commonly used in strategic scale computer games and the more I think about it, the less sense it makes to me to introduce radically divergent, random outcomes to combat. WWII was largely a numbers game. Germany and Japan lost because in the aggregate they failed the numbers game (both in quality and quantity), not because they were unlucky in combat. Success in computer simulated combat should be mostly attributable to who has the better quality or quantity and who uses their forces more wisely, not to blind luck of the die roll.

Maybe I'm wrong but it seems to me that computer games that rely on wildly random combat results are inherently flawed to some extent.

Your question is about a war, your example is a battle. The answer would be very different depending on which of the two you want to adress.

Depending on the size of the war (we are not talking about a minor local engagement but a monster war like WWII) the outcome is very predictable.
WWII could have been "replayed" several 100´s of times and lead to the same or a very similar outcome, because, simply, the industrial output, potential industrial output, and ressource
availability of the Allies enabled them to outproduce the Axis nations by far more than a margin. Over a large timescale this is enough to predict an Allied victory with a
high ammount of accuracy.
So, considering that fate in battle, technological advantages and heroism tend to cancel each other out over a certain ammount of time, the outcome of WWII was set in stone from the moment Germany
invaded Poland and Japan bombed PH.

Your example though, Midway, is a single battle in a war, and this is a situation where fate, desicions of individuals, and pure simple luck, can have as a big impact on the outcome
as proper allocation of forces or damage control. And even those are often the decision of individuals.
Nagumo made severe tactical mistakes, both, with the setup of his forces and his reaction to events unfolding. But he was already working with a flawed battle plan issued by Yamamoto
where the mistakes began much earlier (the most prominent was yielding to the will of the IJA by assigning a carrier division for operations in the Solomons and the PM invasion and so severely limiting
the forces available to KB for the attack on Midway).
When asking how a replay of Midway would end differently you´d have to define where the battle began. If it began with the assignment of forces for the different operations, I´d say
it could have ended differently, with the forces present on the day of battle the situation heavily favoured the US.

[RUDOLF]

[:D][:D][:D]

[decaro]

ORIGINAL: Gary Childress

How random and unpredicatable is war? My belief is that victory or defeat in war, especially on the macro scale is USUALLY attributable to certain predictable factors added together and not so much to pure randomness.

Take Midway: was Midway a fluke or were the Allies pretty much destined to win ...

When Midway was wargamed by the IJN on the Yamatto, a random dice roll accurately predicted the actual battle results.

[herwin]

ORIGINAL: PaxMondo

Regarding the OP .... what is certain is the Gary (Grigsby) is a big beleiver in randomness and always has been.  I would characterize it by saying it is his hallmark as compared to other developers.  Not to suggest other developers don't use randomness, they do, simply not to the extent that Gary does.

As to reality ... there are any number of major historical events that sure seem to be pure chance.  My favorite is Mao's escape.  So, there seems to be at least some basis for high randomness at even high levels of conflict.

There's randomness and there's stochastic chaos. Randomness is predictable once you know the distribution because the trajectory of the mean is well-behaved. Stochastic chaos is unpredictable as the underlying dynamics exhibit sensitive dependence on initial conditions. Random variation in the trajectory is inflated by the non-linearity, so you end up exploring unexpected regions of the state space. Combat is stochastic chaos, not randomness.

[crsutton]

The outcome of the war could not have been random. Given the economies of the Allies vs that of the Axis powers the end result was almost a certainty.

The outcome of any single event, yes. The bullet might hit or miss.

[herwin]

ORIGINAL: crsutton

The outcome of the war could not have been random. Given the economies of the Allies vs that of the Axis powers the end result was almost a certainty.

The outcome of any single event, yes. The bullet might hit or miss.

I was discussing combat...

[brian800000]

There needs to be randomness in the game because the player's control is so abstracted.

Take an engagement that matches a few destroyers against a fleet with cruisers and battleships. Obviously the destroyers are overmatched. But as Taffy 3 showed, if the destroyer commanders are extremely aggressive and have a few breaks go their way, and the other side is confused and not determined, they have a chance.

[herwin]

ORIGINAL: brian800000

There needs to be randomness in the game because the player's control is so abstracted.

Take an engagement that matches a few destroyers against a fleet with cruisers and battleships. Obviously the destroyers are overmatched. But as Taffy 3 showed, if the destroyer commanders are extremely aggressive and have a few breaks go their way, and the other side is confused and not determined, they have a chance.

If you simply rely on randomness, the one in a million chances won't happen--you'll get something in the distribution around the expected outcome. If you incorporate stochastic chaos, the results will split up at critical decision points and go to a distribution around one of a number of possible outcomes--you won't get intermediate cases, and each of the outcomes will make sense.