AiryW
Full Member
Posts: 183
|
Post by AiryW on Feb 12, 2019 7:18:06 GMT -6
I've been reviewing the internet for game theory and economics concepts and ideas. It is a very complex subject and the gist of the information says that most games can't really model economics that well and don't because of the complexity. The general solution to such economic models isn't even possible with current mathematics. It would require a complete solution to the knapsack problem which is a subset of the famous p=np problem. Without such a solution (which may not even be possible), there is exponential complexity to any complete model. However an incomplete model can be very, very powerful if you are careful about what you chose to include. The Federal Reserve Open Market Committee forward guidance estimate process is the world's most important economic model and they are based on math simpler then that which renders the globe image used in RTW. The trick is to just carefully chose what will be modeled and what will be treated as outside the scope of the problem, either predetermined or random (i.e. "endogenous or stochastic"). What not only trips up game designers but also professional economists is that they will make a model based upon an assumption then try to apply that model to circumstances where that assumption is not valid. A very common example of this assumptions-scope fallacy in games is the runaway snowball effect. This has the underlying assumption that having more resources will improve your outcomes (i.e. "expected values monotonically increase with expenditures"). Normally, that assumption is valid and obvious. However numerous counter-examples exist: the Kodak effect, Brooks law, the pirate problem. And the reason why the assumption normally holds is because of the deliberate avoidance of situations where it doesn't hold. If the designer ignores the incentives that produce this avoidance, the player is rewarded for strategies that violate the underlying assumption which produces runaway snowballing. One of the reasons I think RTW is so strong is that it limits it's scope to a range of outcomes within which it is valid to assume that having more resources will improve your outcomes. You can snowball but you can't runaway. I feel like the real takeaway is that you want to define your limits as the very first step. Decide what will be treated as external and random and what will be simulated. Then look to see if what will be included properly includes the tradeoffs you want. If you can't get that right, it doesn't matter what model or algorithms you use because when your assumptions are violated, any model will break down.
|
|
imryn
Full Member
Posts: 156
|
Post by imryn on Feb 12, 2019 7:54:56 GMT -6
I would be very wary of any economic modeling that tried to equate to any historical models. Any historical economic figures that we can reference are grossly distorted by WW2 and the game does not allow for conflicts of that size. WW2 forced the nations involved to exert every ounce of economic strength in a way that no conflict before or since has done, and this fundamentally changed the economic map of the world forever. Without that event the historical models cannot apply.
If it was possible to have a war where France, GB, Russia, and USA were allied and fighting against Germany, Japan, and Italy (also allied) then we could use historical models once such a war had been fought, but its not possible.
The best you can hope to do with historical figures is to take the numbers for 1900 and extend them in a straight line, with minor adjustments for in game events. The real world figures are contaminated by real world events.
|
|
|
Post by jwsmith26 on Feb 12, 2019 8:53:27 GMT -6
I would be very wary of any economic modeling that tried to equate to any historical models. Any historical economic figures that we can reference are grossly distorted by WW2 and the game does not allow for conflicts of that size. WW2 forced the nations involved to exert every ounce of economic strength in a way that no conflict before or since has done, and this fundamentally changed the economic map of the world forever. Without that event the historical models cannot apply. If it was possible to have a war where France, GB, Russia, and USA were allied and fighting against Germany, Japan, and Italy (also allied) then we could use historical models once such a war had been fought, but its not possible. The best you can hope to do with historical figures is to take the numbers for 1900 and extend them in a straight line, with minor adjustments for in game events. The real world figures are contaminated by real world events.That line needs to go on a T-shirt.
|
|
|
Post by generalvikus on Feb 12, 2019 12:38:24 GMT -6
I would be very wary of any economic modeling that tried to equate to any historical models. Any historical economic figures that we can reference are grossly distorted by WW2 and the game does not allow for conflicts of that size. WW2 forced the nations involved to exert every ounce of economic strength in a way that no conflict before or since has done, and this fundamentally changed the economic map of the world forever. Without that event the historical models cannot apply. If it was possible to have a war where France, GB, Russia, and USA were allied and fighting against Germany, Japan, and Italy (also allied) then we could use historical models once such a war had been fought, but its not possible. The best you can hope to do with historical figures is to take the numbers for 1900 and extend them in a straight line, with minor adjustments for in game events. The real world figures are contaminated by real world events. Firstly, I'd just like to point out that I do think it would be both easy and beneficial for some kind of coalition war mechanic to be implemented in the game. Having said that, if no such mechanic is implemented and the game continues to consist mostly or entirely of self-contained conflicts in which the player fights one enemy at a time, I do not see why this fact would make broad historical economic trends less relevant. If you're arguing that coalition wars by their nature tend to become more 'total' than non-coalition wars, then I think you're right - to an extent - but not to the extent that we should assume coalition wars are, all else being equal, different in kind and not degree from one - versus - one conflicts. If, for example, the Pacific War had been fought between only Japan and the United States, or if the war in Europe had been a purely Franco - German or German - Soviet affair, I do not see why that mere fact would mean that the belligerents on either side would be less willing or able to mobilise the same share of their resources for the conflict as they did in history. Of course, you can't assume that in a hundred alternate histories the economic realities would always turn out the same - and so some element of random chance, such as a depression event, is desirable, along with the usual fluctuations in budget with all their usual causes. However, I see no reason why, for the purposes of this game, broad economic trends cannot be relied upon. In any case, you have little to fear from me - unless Frederick has some particular formulae in mind, I don't think it is within my capacity (or, at present, my ambition) to devise a system complex enough to be particularly deterministic. If I do not simply draw a straight line from 1900 to 1950, then I will at best draw a curve.
|
|
imryn
Full Member
Posts: 156
|
Post by imryn on Feb 13, 2019 3:14:11 GMT -6
I would be very wary of any economic modeling that tried to equate to any historical models. Any historical economic figures that we can reference are grossly distorted by WW2 and the game does not allow for conflicts of that size. WW2 forced the nations involved to exert every ounce of economic strength in a way that no conflict before or since has done, and this fundamentally changed the economic map of the world forever. Without that event the historical models cannot apply. If it was possible to have a war where France, GB, Russia, and USA were allied and fighting against Germany, Japan, and Italy (also allied) then we could use historical models once such a war had been fought, but its not possible. The best you can hope to do with historical figures is to take the numbers for 1900 and extend them in a straight line, with minor adjustments for in game events. The real world figures are contaminated by real world events. Firstly, I'd just like to point out that I do think it would be both easy and beneficial for some kind of coalition war mechanic to be implemented in the game. Having said that, if no such mechanic is implemented and the game continues to consist mostly or entirely of self-contained conflicts in which the player fights one enemy at a time, I do not see why this fact would make broad historical economic trends less relevant. If you're arguing that coalition wars by their nature tend to become more 'total' than non-coalition wars, then I think you're right - to an extent - but not to the extent that we should assume coalition wars are, all else being equal, different in kind and not degree from one - versus - one conflicts. If, for example, the Pacific War had been fought between only Japan and the United States, or if the war in Europe had been a purely Franco - German or German - Soviet affair, I do not see why that mere fact would mean that the belligerents on either side would be less willing or able to mobilise the same share of their resources for the conflict as they did in history. Of course, you can't assume that in a hundred alternate histories the economic realities would always turn out the same - and so some element of random chance, such as a depression event, is desirable, along with the usual fluctuations in budget with all their usual causes. However, I see no reason why, for the purposes of this game, broad economic trends cannot be relied upon. In any case, you have little to fear from me - unless Frederick has some particular formulae in mind, I don't think it is within my capacity (or, at present, my ambition) to devise a system complex enough to be particularly deterministic. If I do not simply draw a straight line from 1900 to 1950, then I will at best draw a curve. I think it is already possible to have multiple participants in wars in RTW1. I have had wars where I am allied to someone and they fight as well (for certain values of "fight" - contributing a handfuls of vp's every couple of months hardly seems like they are very committed), and I have had wars where my opponent has dragged in an ally through alliances, so I assume it is possible to get into a war where both the player and the opponent have allies. Probably just very unlikely to get two a side and impossible to get 3 or more a side (3 on both sides - I've seen 3 on 1). "Broad economic trends" is a phrase that can cover a lot of ground. In the absolute broadest sense it would just be a slow steady rise, but as soon as you start to narrow your focus you come up against those real world events. That is because nothing in the real world changes without a reason; there is always an event to trigger a change so if a country's economy changes (up or down) it is because of something - a change of government, an invention, a war, an alliance. We can add fictional events to the game and give them economic consequences, but when we try to add events from the real world, (such as a regime change in Russia) we hit a problem. Those real world events didn't happen in isolation, they are the end result of a whole chain of other events (if no WW1 does the Tsar fall?) In my opinion, if you try to base an economic model on the real world you have to go all in and make it as accurate and detailed as possible for the simple reason that if it is reality based then your users will be able to pick holes in it forever, and they will, no matter how good a job you do. Better to come up with a generic economic model and when users ask why the USA doesn't have the budget to build 4 BB's and 27 CV's at the same time tell them its a game.
|
|
|
Post by elenian on Feb 13, 2019 16:45:22 GMT -6
I've been reviewing the internet for game theory and economics concepts and ideas. It is a very complex subject and the gist of the information says that most games can't really model economics that well and don't because of the complexity. The general solution to such economic models isn't even possible with current mathematics. It would require a complete solution to the knapsack problem which is a subset of the famous p=np problem. Without such a solution (which may not even be possible), there is exponential complexity to any complete model. Not to be too nitpicky, but, assuming P≠NP (which, I think, is very probably the case), NP-complete problems are at worst exponential. Some NP-complete problems even have currently known subexponential solutions! EDIT: depending on what exactly you mean by 'subexponential', I suppose. More relevantly, in general whether or not [possibly exponential] superpolynomial algorithms can be computationally tractable will depend on what kind of instances you are trying to solve. The maximally general 'n goes to infinity' question is really the only thing captured by the complexity class- for finite instances, sometimes you'd prefer a superpolynomial solution to a polynomial time one. The explanation is not especially interesting or subtle: suppose something runs in n 1000 (polynomial) and another thing runs in n logn (superpolynomial). Then the smallest instance size at which the polynomial algorithm becomes faster than the superpolynomial one has n=1.07x10 199 bits, so unless your instance is bigger than the observable universe, you're better off with the 'more complex' algorithm! Now, actual economies have a pretty large expected instance size , and so are likely to be pretty hard to model completely and quickly, even if they turn out to be in a 'low' complexity class. So I tend agree with your conclusion that an incomplete model is generally desirable, just for (I think) different reasons. Postcriptum: it's maybe worth pointing out explicitly that computational tractability and subexponential runtime are much stronger constraints than "general [solubility] with current mathematics", since there are strongly superexponential functions that are nonetheless decidable [eg, Ackermann functions], though computationally intractable for even small instances. And then there is quantum computation and randomized algorithms and all sorts of witchcraft that I don't understand. I'm certain you didn't mean to imply that exponential-complexity problems were unsolvable in principle - I mention this just in case anyone misunderstood your post, as I did originally.
|
|
AiryW
Full Member
Posts: 183
|
Post by AiryW on Feb 14, 2019 1:46:02 GMT -6
That's a lot to chew on, I was definitely being sloppy with my thinking and will need to continue to read up on this. Many thanks.
|
|
|
Post by generalvikus on Feb 14, 2019 19:53:05 GMT -6
"Broad economic trends" is a phrase that can cover a lot of ground. In the absolute broadest sense it would just be a slow steady rise, but as soon as you start to narrow your focus you come up against those real world events. That is because nothing in the real world changes without a reason; there is always an event to trigger a change so if a country's economy changes (up or down) it is because of something - a change of government, an invention, a war, an alliance. We can add fictional events to the game and give them economic consequences, but when we try to add events from the real world, (such as a regime change in Russia) we hit a problem. Those real world events didn't happen in isolation, they are the end result of a whole chain of other events (if no WW1 does the Tsar fall?) In my opinion, if you try to base an economic model on the real world you have to go all in and make it as accurate and detailed as possible for the simple reason that if it is reality based then your users will be able to pick holes in it forever, and they will, no matter how good a job you do. Better to come up with a generic economic model and when users ask why the USA doesn't have the budget to build 4 BB's and 27 CV's at the same time tell them its a game. I think that this is a very solid argument, and I think that my only response to it is this: despite all the variables and contingent factors, I think that an implementation of economics which approximates history will lead to a more plausible result than what we have now in RTW 1, which is more or less arbitrary. Certainly a game in which the USA does enjoy approximately the same build power that it enjoyed in real life has no claim to be a completely logically sound or objectively perfect representation of alternate history, but I do think that it has more claim to 'plausibility' than a system which is based upon arbitrary values, and I personally find that to be an attractive quality. By its nature this game must deal with alternate history, but nevertheless, I find it appealing to be faced with challenges that are historically plausible, and to devise historically plausible solutions to them. And, from a utilitarian standpoint, I think that people who are looking for challenges will find it much more compelling to play as, for example, a reasonably historically accurate Japan against a reasonably historically accurate United States than to play as Spain or Qing China against the world as some choose to do at present.
|
|
|
Post by barrybull on Mar 11, 2019 18:58:51 GMT -6
"Broad economic trends" is a phrase that can cover a lot of ground. In the absolute broadest sense it would just be a slow steady rise, but as soon as you start to narrow your focus you come up against those real world events. That is because nothing in the real world changes without a reason; there is always an event to trigger a change so if a country's economy changes (up or down) it is because of something - a change of government, an invention, a war, an alliance. We can add fictional events to the game and give them economic consequences, but when we try to add events from the real world, (such as a regime change in Russia) we hit a problem. Those real world events didn't happen in isolation, they are the end result of a whole chain of other events (if no WW1 does the Tsar fall?) In my opinion, if you try to base an economic model on the real world you have to go all in and make it as accurate and detailed as possible for the simple reason that if it is reality based then your users will be able to pick holes in it forever, and they will, no matter how good a job you do. Better to come up with a generic economic model and when users ask why the USA doesn't have the budget to build 4 BB's and 27 CV's at the same time tell them its a game. I think that this is a very solid argument, and I think that my only response to it is this: despite all the variables and contingent factors, I think that an implementation of economics which approximates history will lead to a more plausible result than what we have now in RTW 1, which is more or less arbitrary. Certainly a game in which the USA does enjoy approximately the same build power that it enjoyed in real life has no claim to be a completely logically sound or objectively perfect representation of alternate history, but I do think that it has more claim to 'plausibility' than a system which is based upon arbitrary values, and I personally find that to be an attractive quality. By its nature this game must deal with alternate history, but nevertheless, I find it appealing to be faced with challenges that are historically plausible, and to devise historically plausible solutions to them. And, from a utilitarian standpoint, I think that people who are looking for challenges will find it much more compelling to play as, for example, a reasonably historically accurate Japan against a reasonably historically accurate United States than to play as Spain or Qing China against the world as some choose to do at present. The problem is, a reasonably historically accurate AI United States is very difficult to beat or too easy to use by player. The crux remains that to properly weight game balance and the scope of influence to change history provided to players.
|
|
|
Post by corsair on Mar 11, 2019 19:44:06 GMT -6
The problem is, a reasonably historically accurate AI United States is very difficult to beat or too easy to use by player. The crux remains that to properly weight game balance and the scope of influence to change history provided to players.
Which is why there ought to be a "historical' and "sandbox" (or whatever designation one likes) modes so that those wanting the full historical challenge can do so, while those who want a balanced game can have that. Different challenges for the player in each case. A game such as RTW2 would seem a perfect fit for such historical and fictional/balanced modes.
|
|
|
Post by oldpop2000 on Mar 11, 2019 20:01:37 GMT -6
Just my thoughts on the economic power of the United States in the game. The first issue is that the game proceeds from 1900 to 1950. It assumes that the period between the Civil War, a half century the US economy grew profoundly, quicker than any other nation and by the turn of the century, it was a strong as the British, with the largest and most modern economy on earth. We became a world financial power.
Now, after 1900 there were some periods where we had some economic downturns but those were temporary. The Crash of 1907 where J.P. Morgan averted financial calamity by putting together a loan for the US government and another six years before the stabilizing event occurred when the Federal Reserve system was created.
The next crash was in 1929 and the following depression which struck the whole world. Globalization had begun in the early years of the 20th century and failure of our stock market caused problems all over the world.
This is not an economic history lesson, I am not qualified for that. But it seems to me that the key to changing the game’s economics is 1900. In other words, there should be, if that is what the forum wants, some way of having a choice as how, many of most powerful economic nations enter the 20th century economically. Examples:
Germany unification – no unification and no German economic power?
Great Britain’s economic status, could it be changed by the numerous colonial wars?
United States economic growth in the forty to fifty years after the Civil War. Does the North win the Civil War and how does this change our economic growth? What about the famous men who built America like Cornelius Vanderbilt, J.P. Morgan, John D. Rockefeller? Andrew Carnegie, Henry Ford? Would they have been able to perform the miracles of economic growth had the North lost?
Anyway, many of you are probably more knowledgeable about economic history, so put it to work.
The question is: Should the game give the player some setup boxes for economics at the beginning to allow the game to proceed down the “path not followed”. If we gather information on the GNP of each nation in 1900 and its average rate of growth from 1900 to 1950, then we can play with the numbers internally in the game and change the economics of the game. However, geography and natural resources internally do play a part in this economic growth. We can't change the geology of Japan, Italy and some of the nations. They either have resources or they don't. So that is, in my amateurish opinion, how we should proceed.
|
|
|
Post by generalvikus on Mar 11, 2019 23:43:02 GMT -6
Just my thoughts on the economic power of the United States in the game. The first issue is that the game proceeds from 1900 to 1950. It assumes that the period between the Civil War, a half century the US economy grew profoundly, quicker than any other nation and by the turn of the century, it was a strong as the British, with the largest and most modern economy on earth. We became a world financial power. Now, after 1900 there were some periods where we had some economic downturns but those were temporary. The Crash of 1907 where J.P. Morgan averted financial calamity by putting together a loan for the US government and another six years before the stabilizing event occurred when the Federal Reserve system was created. The next crash was in 1929 and the following depression which struck the whole world. Globalization had begun in the early years of the 20th century and failure of our stock market caused problems all over the world. This is not an economic history lesson, I am not qualified for that. But it seems to me that the key to changing the game’s economics is 1900. In other words, there should be, if that is what the forum wants, some way of having a choice as how, many of most powerful economic nations enter the 20th century economically. Examples: Germany unification – no unification and no German economic power? Great Britain’s economic status, could it be changed by the numerous colonial wars? United States economic growth in the forty to fifty years after the Civil War. Does the North win the Civil War and how does this change our economic growth? What about the famous men who built America like Cornelius Vanderbilt, J.P. Morgan, John D. Rockefeller? Andrew Carnegie, Henry Ford? Would they have been able to perform the miracles of economic growth had the North lost? Anyway, many of you are probably more knowledgeable about economic history, so put it to work. The question is: Should the game give the player some setup boxes for economics at the beginning to allow the game to proceed down the “path not followed”. If we gather information on the GNP of each nation in 1900 and its average rate of growth from 1900 to 1950, then we can play with the numbers internally in the game and change the economics of the game. However, geography and natural resources internally do play a part in this economic growth. We can't change the geology of Japan, Italy and some of the nations. They either have resources or they don't. So that is, in my amateurish opinion, how we should proceed. My feeling has always been that the strategic environment which was created by the convergence of economic and political circumstances that came about in history should be privileged over arbitrarily selected alternate scenarios like the non-unification of Germany or a Confederate victory in the Civil War. I agree with corsair's point that a 'historical' and a 'balanced' mode is appropriate, and of course that concept has already been implemented in the original RTW. And - again - as to the argument made by barrybull that putting the United States in particular in its historical position would be too easy to play and too difficult to play against; I say that since people have made mods to play as Spain, the Netherlands, and Qing China in the original game, there is clearly a strong appetite for challenging strategic situations. In my opinion, so long as I'm looking for a challenge, playing in the historical position of Japan in a World War II - like situation would be far more compelling than playing as Spain, the Netherlands, or Qing China in any situation. On the other hand, many of my most enjoyable moments in RTW came in wars where I had a clear, perhaps overwhelming advantage, because in every battle I was always trying to minimise my losses and maximise the damage done to the enemy; furthermore, the way battles are made in RTW means that, just as in history, there will often be instances in which the superior force will find itself in an unfortunate and unfavourable position, as in - say - the Battle of Samar. Finally, as corsair pointed out, I have never suggested that the 'balanced' game mode be removed; only that the 'historical resources' mode be made credibly historical.
|
|