|
Post by tortugapower on Nov 29, 2017 1:21:58 GMT -6
ddg top notch investigating there! I love to see hard data like this to understand a little more about what's going on "behind the curtain".
|
|
|
Post by aeson on Nov 29, 2017 3:35:58 GMT -6
I don't have a dataset for a full game (yet), but in an Italian test game that I've just started in version 1.34b (played through December 1903 so far), every power in the game has experienced 1% growth ticks in April and December each year, the USA has experienced an additional 1% growth tick every June, Germany experienced an additional 1% growth tick in March 1901 and March 1903, and the USA experienced an additional 1% growth tick in March 1901. Additionally, Austria-Hungary has experienced a 350k growth tick every February, and the USA and Germany have experienced 350k growth ticks every August. I haven't seen any other growth ticks thus far. I am not a great statistician and this is as yet not a particularly large data set, but this does not look random to me. Hypothesis 1: 1% growth events for Italy, Austria-Hungary, Great Britain, France, and Russia can be modeled as five independent random variables with a 17% chance of occurring each turn (expect about two 1% growth events in a 12-turn period). The probability of all five powers experiencing a 1% growth event on a single turn would then be roughly 0.014%, while the probability that all five powers would not experience a 1% growth event on any given turn would then be roughly 39%. In 48 turns, all five powers experienced a 1% growth event on the same turn 8 times, and all five powers experienced no percentile growth events 40 times. If 1% growth events for these five powers can be modeled as independent random variables with an occurrence probability of 17% each turn, this is extremely unlikely. Therefore, we can discard this hypothesis. Hypothesis 2: 1% growth events for Italy, Austria-Hungary, Great Britain, France, and Russia can be modeled as five independent random variables with some very high probability of occurring each April and December, and no probability of occurring any other time. We have had eight events (April 1900, December 1900, April 1901, December 1901, April 1902, December 1902, April 1903, and December 1903) where each of the five random variables has returned a positive event, and 0 events where any of the five random variables has returned a negative event. If we want a better than 10% probability of this sequence of events occurring so as not to rule out this hypothesis, then: - For all five random variables to return a positive result simultaneously, we need at least a 63% probability of a positive result on each trial. - For one of the five random variables to return a positive result in eight of eight trials, we need at least a 75% probability of a positive result on each trial. - For five of five random variables to return positive results in eight of eight trials, we need at least a 94% probability of a positive result on each trial. If we want a better than 50% probability of this sequence of events occurring so as not to rule out this hypothesis, then: - For all five random variables to return a positive result simultaneously, we need at least an 87% probability of a positive result on any one trial. - For one of the five random variables to return a positive result in eight of eight trials, we need at least a 91% probability of a positive result on any one trial - For all five random variables to return positive results in eight of eight trials, we need at least a 98% probability of a positive result on any one trial. Personally, I think these confidences are a bit low, but then, the data set is a bit small, too. Still, I'd be inclined to discard this hypothesis, especially if the USA and Germany are added in with their 1% growth events modeled as independent random variables with the same probabilities of occurring in April and December each year as the other powers' 1% growth events, simply because the required probability of a positive result on each test in order for the probability of the sequence of events shown in the data set occurring to become higher than 10% or 50% or whatever confidence level you decide is reasonable is rapidly approaching 1, and a 'random' variable with a 100% probability of returning a given result isn't really a random variable. Attachments:ItalyTestData.xlsx (27.83 KB)
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 29, 2017 6:56:29 GMT -6
ddg Do you think colonial possessions contribute to eco growth over time ala 1% ticks? In your set Russia, France, Italy and GB all have no eco growth traits but GB received more 1% growth and higher gross growth. In my own playthrus where I delete growth traits from all nations and repeatedly reset from 1950 to 1926, GB's resource will eventually surpass USA after several resets. My assumptions were either long wars or colonies. Atm it seems the latter.
|
|
|
Post by ddg on Nov 29, 2017 14:56:27 GMT -6
aeson You are correct. It's been over a year since I worked on this and I only looked at my summary page instead of the actual data. Going back through it, percentage growth events definitely seem to be related to the month. New chart: Month | USA | Germany | GB | France | Russia | Japan | Italy | January |
| | 1 | | | | | February |
| | | | | | | March
| 7 | 5 | | | | | | April | 24 | 25 | 23 | 23 | 23 | 25 | 25 | May |
| | 1 | | | | | June | 20 | | 2 | | | | | July |
| | 5 | | | | | August | 13 | | | | | 12 | | September |
| | | | | | | October |
| | 4 | | | | | November |
| | 2 | | | | | December | 22 | 23 | 20 | 23 | 24 | 24 | 23 |
This is the number of times a percentage growth event occurred during a given month stretching from January 1900 to December 1925. Now, back when I was originally working on this, I had to note all the growth events by hand, since some months combined percentage and fixed growth, so I may have made some errors. Some patterns are still clear. Great Britain is the major outlier here and I have no explanation for it. My notes also tell me I found three anomalies in British economic growth that I wasn't able to easily explain as a combination of percentage and fixed growth. Germany overtook GB in base resources fairly early in the game, but GB clawed its way back above Germany near the end. The new data here suggest that RGP1 combines 350k fixed growth and additional percentage growth in March and RGP2 combines fixed 350k growth and additional percentage growth in August (not shown here, but USA's August growths appear to occur later in the game). You can find the complete data set on my Google Drive here. I'd encourage anyone interested in this to check it out. Like I said, I may have made some errors collating. More eyes are always useful. skwabie I'm inclined to think colonies contribute separately (I'm only tracking national resources here), but I'm not absolutely certain. Have you tried changing colonial ownership and rerunning your tests? That should tell you something.
|
|
|
Post by aeson on Nov 29, 2017 18:30:58 GMT -6
Looking through your data set: - There is a 1,500,000 growth anomaly for all powers other than the USA in September 1903. This is the only anomalous growth event for any power other than Great Britain and the USA in the data set. (Note - I did not check for non-occurrence of growth events when one was expected, so that type of anomaly is not included in this summary.) - Great Britain experiences five other anomalous growth events: -- In December 1907, its base resources increase by 520,278. A 1% growth tick would have been 258,845, while a combined 1% and 350k growth tick would have been 608,845 or 612,345 depending on order. Interestingly, the German naval budget exceeded the British naval budget from March 1907 to May 1907. -- In April 1912, its base resources increase by 607,816. A 1% growth tick would have been 302,396, while a combined 1% and 350k growth tick would have been 652,396 or 655,896 depending on order. For much of the preceding year (possibly longer), the German and American naval budgets had remained at around 70% of the British naval budget. -- In November 1912, Great Britain gains 20,000 base resources. I'm inclined to chalk this up to typographical error as the preceding four turns indicate base resources are 31,977,521 while November indicates 31,997,521 .
-- In December 1912, Great Britain gains 622,748 base resources. A 1% growth tick would have been 319,975, while a combined 1% and 350k growth tick would have been 669,975 or 673,475 depending on order (if the November entry is erroneous, the growth tick experienced becomes 642,748, a 1% growth tick becomes 319,775, and a combined 1% and 350k growth tick becomes either 669,775 or 673,275 depending on order). As with the anomalous April 1912 growth tick, the German and American naval budgets have hovered around 70% of the British naval budget for many of the preceding 12 turns. -- In September 1913, Great Britain gains 834,465 base resources. This is exactly what one would expect of a 500k + 1% growth tick if the 1% growth tick occurs after the 500k growth tick. - The USA experiences 7 anomalous growth events, but I have not looked at them to see if they can be rationalized as combinations of 350k, 500k, and 1% growth events. I have not looked at the fourteen 1% growth events Britain experienced that occurred 'off-schedule,' i.e. not in April or December, but it would be interesting if they occur after prolonged periods where the British naval budget was similar to or lower than the German or American naval budgets. While this is obviously very preliminary, it makes me wonder if (some of) the budget modifier(s) related to Global Naval Power might in fact be base resource growth events. Also, attached is a spreadsheet containing ddg 's base resources data set and some characterization of the growth events into 1%, 350k, 500k, and anomalous events, in case anyone wants to take a look. Note that non-occurrence of a growth event at an expected time is not flagged as anomalous; if you want to check for that, you will need to do it yourself. Attachments:Book1.xlsx (203.3 KB)
|
|
|
Post by ddg on Nov 29, 2017 20:30:16 GMT -6
I believe all of the American anomalies were easy to rationalize as combinations of growth types or, alternatively, were the spoils of victory.
The 1,500,000 increase coincides with me getting the railroads event. Unlike every other time I've gotten it, that run I chose to strengthen the navy. This suggests to me that the AI powers either got the same event (and chose a different option, presumably building railroads) or automatically receive some effect based on the player's choice.
Your characterization as a typographical error is most likely correct as I hand-transcribed this data from Almanac each turn. I believe I caught a few other such errors prior to publication, but overall I'm fairly pleased with my accuracy.
Global Naval Power almost certainly affects the budget multiplier (which I denoted as "Navy share" in that spreadsheet). Great Britain is nearly always the leader or a close second in that category. It would be quite interesting if it also involved resource adjustments.
I look forward to continued analysis.
|
|
|
Post by garrisonchisholm on Nov 30, 2017 8:56:33 GMT -6
I love seeing this type of analysis done, bravo (I can't co-op Bcoop's catch phrase, so just a plain bravo). *thumb's up*
|
|
|
Post by bcoopactual on Nov 30, 2017 11:10:05 GMT -6
I love seeing this type of analysis done, bravo (I can't co-op Bcoop's catch phrase, so just a plain bravo). *thumb's up* Haha, it's not my catch phrase, it's just a navy one. Feel free. This is a game about navies after all.
|
|
|
Post by aeson on Dec 7, 2017 0:25:00 GMT -6
Global Naval Power almost certainly affects the budget multiplier (which I denoted as "Navy share" in that spreadsheet). Great Britain is nearly always the leader or a close second in that category. It would be quite interesting if it also involved resource adjustments. I decided to test this by giving Italy Global Naval Power status in a new test game (game resources, very large fleet sizes), and I've had an unexpectedly large (2.01%) growth event in December 1900 and an unexpected (1%) growth event in January 1901. I'm only to August 1901 so far, but considering that in your 25-year data set and the 5.5 year data set I have thus far from my other test game where Italy had no unexpected growth events (other than a 1.5M growth event which all the other powers had and which was probably the railroads event) and had two unexpected growth events in the first year and a half of a test game where it was given Global Naval Power status, I'd say that it seems likely that Global Naval Power can trigger base resource growth events if the budget ratios are unfavorable.
|
|
|
Post by garrisonchisholm on Dec 7, 2017 18:29:52 GMT -6
My! I am inclined to give that a go, as a-historical as it would be! Would be nice if New Rome could get a fair shake.
|
|
|
Post by Fredrik W on Dec 8, 2017 10:54:34 GMT -6
There is a function to simulate that Britain would reallocate resources to make sure she didn't fall behind in naval strength in this period. It might give non historical results if the game is continued much beyond 1925. In RTW2 the economy will be geared to give realistic economic development up to 1950.
|
|
|
Post by Airy W on Dec 8, 2017 12:19:36 GMT -6
There is a function to simulate that Britain would reallocate resources to make sure she didn't fall behind in naval strength in this period. It might give non historical results if the game is continued much beyond 1925. In RTW2 the economy will be geared to give realistic economic development up to 1950. I feel like the British commitment to make sure they dont fall behind should be represented with naval budget, not with national resources. Perhaps this could be represented by a separate "national commitment" variable that indicates how grandiose the ambitions a nation's politicians want to fund. For example, consider the US in 1925 vs. Italy and Japan in 1925. The US has vastly more resources then Italy or Japan, perhaps 15-20 times as much, but wanted to economize on naval spending. Italy and Japan on the other hand tried to squeeze every bit they could out of the naval treaty limits. As a result Italy and Japan did their very expensive refits of older ships to higher speeds while the US did not. However once WWII broke out, all those American resources were unleashed so the US built 18 Essex class carriers alone while the Japanese and Italian shipyard could not keep up. Or look at Germany during the two world wars. In the first world war, they had a huge navy so the blockade was considered something of a defeat for them. There was political pressure to have the battleships fight. In the second, they weren't trying to compete with the British navy so it was just assumed from day one they would be under blockade. The political pressure to have the battleships fight was much less. For the British, they would have a mechanic where if another nation has a higher commitment, British commitment will go up unless they are already at their maximum resources. And when there is a naval treaty in effect, liberal democracies have their commitment go way down but autocracies do not.
|
|
|
Post by Fredrik W on Dec 8, 2017 15:15:46 GMT -6
Theoretically, yes. However, it is not possible to simulate everything in a game. The reason base resources was used in this case is because in game, rises in naval budget will affect unrest levels. In Britain's case, I think reallocations from other sectors would have been possible without increasing popular discontent.
|
|