CXXC or OCP - An Investigation

[Pyrrhos]

CXXC or OCP – Test Description
Since it is an examination of the productivity of the core in order to determine if either placement confers any significant advantages over the other and in order to minimize random influences, the following rules have been adopted.
· The test will be played using an actual starting position on a randomly generated huge Pangea map (to allow enough area for expansion and as a buffer against AI civs that might skew the results)
· No alterations to the generated map are allowed except the relocation of the occasional starting location so that the interference of AI civs with the test will be minimal
· No scouting outside the territory is allowed in order to minimize the chance of encounters with AI civs
· Should contact be made by AI civs anyway, no trade is allowed
· The core is defined as the capital plus the first and second ring, in all 19 towns
· In order to further minimize the element of chance, 18 additional settlers and workers will be provided at the start
· The settlers will be placed as closely as possible to the respective pattern with the exception of deserts where no towns will be placed unless bordering on or at the very edge of a desert
· The build order will be as similar as possible: warrior – worker – (temple for OCP only) – granary – barracks – veteran spearmen
· No other improvements or wonders, even if available, may be built
· Research will be at max consistent with not making a loss although –1 to -3 or so gpt is acceptable
· Research order will be pottery > alphabet > writing > CoL > philosophy > republic > masonry > maths > currency > construction
· In addition to the workers provided, no town will be allowed to build more than one worker except to replace any expended as colonies
· The initially produced regular warriors will be replaced by vet spearmen and disbanded for shields
· No pop-rushing is allowed, cash-rushing is
· No town except the capital is allowed to reach pop 7. Before this happens, a settler must be built
· Sampling will be done at 3000BC and 1990BC
· Since it is assumed that the OCP test will finish constructing their barracks later on account of having to build temples, the test will be run until the date the last OCP finishes or cash-rushes its barracks. Then the CXXC test will be run until the same date

The following aspects will form the basis for comparison between the two patterns:
· Number of tiles controlled
· Number and type of strategic resources/luxes inside that territory
· Number of citizens and settlers
· Number of veteran spearmen
· Gpt gross, corruption and net
· Spt gross, wasted and net
· Science, ie who is ahead and by how much
· Point in time when changeover to republic occurs
· Score and culture

 

[Pyrrhos]

Just settling the capital and eighteen cities in 4000BC, it was obvious that CXXC confers a distinct advantage:

4000BC
OCP (or CXXXXC)
Income 35, expenses 28 (science 21, corruption 7) net gain 7 gpt
Shields 35, wasted 9, net 26 spt

CXXC 4000BC
Income 43, expenses 31 (science 23, corruption 8) net gain 12 gpt
Shields 38, wasted 8, net 30 spt

Since the number of cities were identical, the difference is due to distance corruption. Furthermore, the higher income meant that even though both researched Pottery @ 50% and it took four turns, CXXC could take that down to 20% with one turn remaining. Distance corruption was slowing down the OCP research, a difference that would become more pronounced.

3000BC
OCP
Score 458, Culture Value 92
Population 38
Treasury 62
Income 71, expenses 73 (science 42, corruption 21, maintenance 10) net loss –2 gpt
Shields 73, wasted 22, net 51 spt

CXXC
Score 407, Culture Value 20
Population 37
Treasury 97
Income 80, expenses 80 (science 61, corruption 18, maintenance 1) net gain 0 gpt
Shields 66, wasted 13, net 53 spt

As predicted, the much larger territory of OCP meant that it had access to more good tiles – BG, wheat, cow. However, suffering the despotism penalty meant that full advantage could not be taken and that distance corruption effectively nullified the advantage in shields. CXXC was ahead in science and economically, but again, not by much. However, the closer spacing meant that a larger percentage of available tiles had been improved including irrigation of grassland and mining of hills. This was to be an immediate advantage when Republic became available and it was evident that CXXC would get there first. The only bonus for OCP was that three luxes were hooked up as against only one for CXXC which would have to rely on the lux slider to compensate. (In fact, OCP would go through the entire simulation without once having to resort to the lux slider or clowns thanks to those three luxes and the temples.)

2650BC – CXXC has completed the change to republic.
2390BC – OCP has completed the change to republic.

1990BC
OCP
Score 672, Culture Value 958
Population 86
Treasury 91
Income 240, expenses 223 (science 50, corruption 49, maintenance 40, unit support 84) net gain 17 gpt
Shields 146, wasted 32, net 114 spt

CXXC
Score 459, Culture Value 66
Population 93
Treasury 284
Income 295, expenses 270 (science 75, entertainment 26, corruption 51, maintenance 30, unit support 88) net gain 25 gpt
Shields 123, wasted 19, net 124 spt

The prospects for the OCP civ began to look bleak. Even though it was far ahead of CXXC in score, cultural value and shields gross, it looked to be lagging behind instead of catching up. Furthermore, the process seemed to be accelerating.

1475BC
The test ends as the last of 19 barracks has been completed by OCP, cash-rushed the previous turn. The final tallies:

OCP
Score 815, Culture Value 2106
Science – Iron Working just completed, Construction in 16 turns @ 20%
Population - 31 towns with 129 + 3 settlers
Units – 42 workers & 64 veteran spearmen
Treasury 111
Income 354, expenses 334 (science 50, corruption 87, maintenance 57, unit support 140) net gain 20 gpt
Shields 226, wasted 58, net 168 spt

CXXC
Score 567, Culture Value 100
Science – Construction just completed, Map Making in 4 turns @ 40%
Population – 41 towns with 142 (to which should be added three plague victims) + 5 settlers
Units - 42 workers, 52 veteran spearmen
Treasury 583
Income 425, expenses 414 (science 127, entertainment 28, corruption 111, maintenance 40, unit support 108) net gain 9 gpt
Shields 194, wasted 53, net 143 spt

Further expansion by OCP is beginning to get stifled by unit support, even though two military units and one and a third worker per town cannot be termed excessive. CXXC is far ahead in science, has a much healthier economy and, currently, a greater rate of expansion Even though the OCP empire covers almost twice the number of tiles, the expansion rate of the CXXC civ is such that it will soon catch up, especially since the last spurt of expansion means that it too now has three luxes hooked up. The only strategic resource made available by science, iron, was present in neither territory which means that this criteria was not a factor.

Summary and evaluation
Distance corruption is the greatest enemy of OCP. Compared with the tight CXXC, OCP loses out in every respect except three – shield production, score and cultural value. The inclination to develop the best tiles, BG, meant that the OCP civ neglected to develop the population-producing food tiles and has paid for it. I feel that this should be a lesson to all civers!

The clear winner of this test is CXXC and it is evident why advanced players choose this city layout for their games and say that OCP is for low level games. Also, with luxes aplenty, the influence of temples would be limited and more of a drain on the economy than a benefit with CXXC and town no larger than pop 6-7 (on warlord, 4-5 on emperor).

 

[Aabraxan]

Pyrrhos, could you post some screens or may be even some saves so that we can look at this in a little more depth?

 

[Chamnix]

Instead of playing 1 particular map twice and claiming the results are significant, why not look at the GOTM results? You have several years worth of games played by many people with multiple victory conditions and a more or less unbiased ranking of every game. Download the games and take a look at everyone's first 2 rings and where they ended up ranking at the end.

Of course, after viewing the results one could claim that it is not that CxxC is better than CxxxxC but that the players who used CxxC are better than the players who used CxxxxC, but wouldn't that tell you something in itself?

 

[ecuwins]

Instead of playing 1 particular map twice and claiming the results are significant, why not look at the GOTM results? You have several years worth of games played by many people with multiple victory conditions and a more or less unbiased ranking of every game. Download the games and take a look at everyone's first 2 rings and where they ended up ranking at the end.

Of course, after viewing the results one could claim that it is not that CxxC is better than CxxxxC but that the players who used CxxC are better than the players who used CxxxxC, but wouldn't that tell you something in itself?

+1

 

[Pyrrhos]

Chamnix & ecuwins!

What GOTMs (and more advanced/experience players) tell me is that, judging by the results, CXXC has something to recommend it. There must be a certain something to CXXC that is not easy to see and which usually escapes the attention of civvers, at least until they advance far enough to find monarch or emperor too much of a challenge. Not even the advanced players or, to be precise, many of them, seem to know why CXXC is better - as is evidenced by several replies to the original CXXC v. CXXXXC thread - only that it is better.

Now to your question - why not download xyz number of those games and study the results?

In such games there are too many random factors present! Just think of this: I played these tests through three times each. Not once was the CXXC civ contacted by or even saw another civ. On every occasion though, the OCP was, and relatively early too. What would the results have been had the OCP civ been allowed to take advantage of this? Tech trade. Tech trade with an influx of money. The end result would have been very, very misleading - the OCP civ would have been far ahead in tech with a solid economy and this would have been due to trade, not an inherent advantage in the OCP configuration.

In order for any scientific test to be significant you have to eliminate outside influences! This is why I eschewed the use of GOTMs etc, set up a test and laid down stringent specifications to ensure and make certain that the results arrived at would be due to inherent strength and weaknesses of each style of city placement.

Another propos: Why did I start with so many towns? Again, to make certain that any differences, ie advantages and disadvantages, became pronounced enough to become statistically significant and couldn't be dismissed as "flukes".

 

[Pyrrhos]

Pyrrhos, could you post some screens or may be even some saves so that we can look at this in a little more depth?

How? With the easy download system gone, I suppose I could go thru all the faff of registering at photobucket and learn to use it in a week or so, but how do you upload saves?

Remember, as far as advanced PC-usage is concerned, I am a tech div...

 

[ecuwins]

Chamnix & ecuwins!

What GOTMs (and more advanced/experience players) tell me is that, judging by the results, CXXC has something to recommend it. There must be a certain something to CXXC that is not easy to see and which usually escapes the attention of civvers, at least until they advance far enough to find monarch or emperor too much of a challenge. Not even the advanced players or, to be precise, many of them, seem to know why CXXC is better - as is evidenced by several replies to the original CXXC v. CXXXXC thread - only that it is better.

Now to your question - why not download xyz number of those games and study the results?

In such games there are too many random factors present! Just think of this: I played these tests through three times each. Not once was the CXXC civ contacted by or even saw another civ. On every occasion though, the OCP was, and relatively early too. What would the results have been had the OCP civ been allowed to take advantage of this? Tech trade. Tech trade with an influx of money. The end result would have been very, very misleading - the OCP civ would have been far ahead in tech with a solid economy and this would have been due to trade, not an inherent advantage in the OCP configuration.

In order for any scientific test to be significant you have to eliminate outside influences! This is why I eschewed the use of GOTMs etc, set up a test and laid down stringent specifications to ensure and make certain that the results arrived at would be due to inherent strength and weaknesses of each style of city placement.

Another propos: Why did I start with so many towns? Again, to make certain that any differences, ie advantages and disadvantages, became pronounced enough to become statistically significant and couldn't be dismissed as "flukes".

The only problem here is that the "scientific test" is an ideal state. Versus the GOTM's representing "real world scenarios". The latter is how the game is played "for fun".

 

[Pyrrhos]

The only problem here is that the "scientific test" is an ideal state. Versus the GOTM's representing "real world scenarios". The latter is how the game is played "for fun".

Is your glass always half empty?

First, we would never have had Civ to haggle about - or even computers to carry out our arguments over it on and play the game on - were it not for the ideal state of the "scientific test".

Second, if you do not carry out "scientific tests", what remains is opinion and opinions - no matter how stupid, ill-informed, narrow-minded, malicious, self-aggrandising or downright wrong - are equally unarguable or subject to testing. I could claim that anyone who does not share my opinions is a pompous fool and since its "only" an opinion, cannot be disproven!

The problem here is not that "the "scientific test" is an ideal state". The problem is your attitude towards and opinion of science.

 

[ecuwins]

Is your glass always half empty?

First, we would never have had Civ to haggle about - or even computers to carry out our arguments over it on and play the game on - were it not for the ideal state of the "scientific test".

Second, if you do not carry out "scientific tests", what remains is opinion and opinions - no matter how stupid, ill-informed, narrow-minded, malicious, self-aggrandising or downright wrong - are equally unarguable or subject to testing. I could claim that anyone who does not share my opinions is a pompous fool and since its "only" an opinion, cannot be disproven!

The problem here is not that "the "scientific test" is an ideal state". The problem is your attitude towards and opinion of science.

You are taking it out of context. But nice try.

 

[Chamnix]

Now to your question - why not download xyz number of those games and study the results?

In such games there are too many random factors present! Just think of this: I played these tests through three times each. Not once was the CXXC civ contacted by or even saw another civ. On every occasion though, the OCP was, and relatively early too. What would the results have been had the OCP civ been allowed to take advantage of this? Tech trade. Tech trade with an influx of money. The end result would have been very, very misleading - the OCP civ would have been far ahead in tech with a solid economy and this would have been due to trade, not an inherent advantage in the OCP configuration.

I agree that in one test game, it would make the results meaningless. You are explaining exactly why you need a large number of games to be remotely meaningful. Maybe the OCP civ will get contacts earlier in general. Maybe this will lead to the OCP being far ahead in tech with a solid economy. If the OCP civ gets contacts significantly earlier most of the time, then this is an advantage of the OCP configuration and should be taken into account when determining which is better.

How could you measure this advantage in one game? You might end up with the OCP civ way ahead in tech, and the CxxC civ having advantages in several other areas, but which would be "better"? The only way to determine which leads to a better result on average is to look at a large number of results.

 

[Aigburth]

I don't think you can draw any meaningful conclusions as to how quickly contacts are made based on this test, in a real game you don't start with 19 settlers already in place to settle on the first turn. This scenario gives the OCP civ a distinct advantage in making contacts as it starts with units much further away from the capital than the CxxC civ. In a real game both strategies will only start with the capital and the early turns will be identical for both, they will only start to diverge once the first settler is 3 tiles away from the capital by which time you will already have a couple of exploring warriors way beyond the distance the first city will be settled and may already have made a couple of contacts.

That said, this test does a good job in highlighting the economic advantages of CxxC placement. In a normal game (i.e. where you only start with one settler) I believe the differences would be even more pronounced as each city will be settled earlier in CxxC and therefore everything those cities produce will be earlier etc...

 

[Pyrrhos]

I agree that in one test game, it would make the results meaningless. You are explaining exactly why you need a large number of games to be remotely meaningful. Maybe the OCP civ will get contacts earlier in general. Maybe this will lead to the OCP being far ahead in tech with a solid economy. If the OCP civ gets contacts significantly earlier most of the time, then this is an advantage of the OCP configuration and should be taken into account when determining which is better.

How could you measure this advantage in one game? You might end up with the OCP civ way ahead in tech, and the CxxC civ having advantages in several other areas, but which would be "better"? The only way to determine which leads to a better result on average is to look at a large number of results.

I have to disagree with your last sentence, but it is a qualified disagreement:

The optimum way to determine which city placement is generally more successful would be to have a very large number of players play the same game, using the same civilization, several times to the best of their ability, each using first the one, then the other method and count their best result for each method only. Then the same players would need to repeat this with different maps types, map sizes, difficulty levels and civs. Yet, in the end we could only say that "X placement style wins more games on ... (map type, map size, difficulty level with [trait/traits] civs)".

If you want to know the reasons why one method is better than the other, their advantages and disadvantages, strengths and weaknesses, you would have to set up rigidly controlled tests, similar to the one I performed, where you can isolate each factor in the sense that you can, with a degree of certainty, conclude that faxtor X favours type Y city placement and the manner in which it does so.

The objection against the contact hypothesis is that if you want to, you can set out specifically to search out other civs and it is a matter of chance as to who meets the most profitable contacts the earliest. City placement has no effect whatsoever on early unit movement as you only have a single pop 1 town when the option to send a warrior (or scout for exp civs) out to search for other civs.*

 

[MAS]

How? With the easy download system gone, I suppose I could go thru all the faff of registering at photobucket and learn to use it in a week or so, but how do you upload saves?

Remember, as far as advanced PC-usage is concerned, I am a tech div...

Hmm, While the forum wide easy upload system may not be functional, I think you can still upload post specific attachments when creating a new post.

Just go to the "Reply to Thread" page and scroll down a bit, there should be a button called "Manage Attachments."

 

[Pyrrhos]

Thx MAS!

Here goes:

 

[Chamnix]

The optimum way to determine which city placement is generally more successful would be to have a very large number of players play the same game, using the same civilization, several times to the best of their ability, each using first the one, then the other method and count their best result for each method only. Then the same players would need to repeat this with different maps types, map sizes, difficulty levels and civs. Yet, in the end we could only say that "X placement style wins more games on ... (map type, map size, difficulty level with [trait/traits] civs)".

Well, sure, and the players should have their memories erased after each game so that when they started the next one they have no foreknowledge of the map. I still maintain that GOTM results will provide a better basis for reaching a conclusion on which is better than 1 person playing 1 map.

If you want to know the reasons why one method is better than the other, their advantages and disadvantages, strengths and weaknesses, you would have to set up rigidly controlled tests, similar to the one I performed, where you can isolate each factor in the sense that you can, with a degree of certainty, conclude that faxtor X favours type Y city placement and the manner in which it does so.

And yet your test led you to an exactly wrong conclusion. Distance corruption is a problem with OCP, but rank corruption is a problem for CxxC. If you give each the same number of cities, then both will suffer the same rank corruption, and of course CxxC will have lower distance corruption than OCP. If you ran this test again using ICS (CxC), I'd bet that you would determine that ICS is even better than CxxC. In a real game over the long-term, corruption is a bigger issue for CxxC than for OCP - lower overall corruption is in fact one advantage that OCP has over CxxC.

To throw in some more facts, distance corruption is increased 50% in despotism compared to other forms of government. Distance corruption increases 25% for cities that are not connected to your capital. Neither of those 2 affect rank corruption (although there is a smaller decrease in rank corruption only for Republic and Democracy but not for other forms of government). Setting up a test like you did is very biased in favor of closer placement but still doesn't get you the real reasons why it is better.

Two of the real reasons are in fact right here:

arguments in favour include "using all tiles" and "making the best of high-yield tiles by sharing".

which you dismiss as "non-arguments" in the other thread.

There has been a decent amount of discussion on whether to compare the same territory (with more cities for CxxC) or the same number of cities (with more territory for OCP). The truth is somewhere between these two - OCP will end up with both more territory and fewer cities than CxxC.

 

[Aabraxan]

Pyrrhos, it looks like you got through attaching files just fine! I don't know about Photobucket, but you don't need to register for ImageShack to host pics there. I use it for my pictures on CFC, but I've never registered.

As for the test, there are some interesting arguments going on here. Foreknowledge of the map may play a role, but if Pyrrhos created the map and placed the settlers, he would have had foreknowledge of the map prior to playing either of them. Rather than trying to erase memory, why not just look at the map prior to playing, so that the player has foreknowledge for both runs? I do think that starting with the standard 1 settler & 1 worker might provide a more accurate picture of the dynamics, even with foreknowledge of the map.

 

[Pyrrhos]

Well, sure, and the players should have their memories erased after each game so that when they started the next one they have no foreknowledge of the map. I still maintain that GOTM results will provide a better basis for reaching a conclusion on which is better than 1 person playing 1 map.

Or, as Aabraxan points out, have foreknowledge. Or, as I suggested, play the same test three times and only count the best one.

And yet your test led you to an exactly wrong conclusion. Distance corruption is a problem with OCP, but rank corruption is a problem for CxxC. If you give each the same number of cities, then both will suffer the same rank corruption, and of course CxxC will have lower distance corruption than OCP. If you ran this test again using ICS (CxC), I'd bet that you would determine that ICS is even better than CxxC. In a real game over the long-term, corruption is a bigger issue for CxxC than for OCP - lower overall corruption is in fact one advantage that OCP has over CxxC.

If you look at it closely, you will realise that rank corruption is beginning to affect the CXXC civ. In spite of having 41 to the 31 towns of OCP, CXXC has a much better net income (before subtracting science) and still has the ability to research at 40% whereas OCP can only manage 20%.

To throw in some more facts, distance corruption is increased 50% in despotism compared to other forms of government. Distance corruption increases 25% for cities that are not connected to your capital. Neither of those 2 affect rank corruption (although there is a smaller decrease in rank corruption only for Republic and Democracy but not for other forms of government). Setting up a test like you did is very biased in favor of closer placement but still doesn't get you the real reasons why it is better.

Intriguing! Please erudite me!

Two of the real reasons are in fact right here:

Originally Posted by Pyrrhos
arguments in favour include "using all tiles" and "making the best of high-yield tiles by sharing".

which you dismiss as "non-arguments" in the other thread.

As they are. I refer you to my replies there and the maps of the actual test.

There has been a decent amount of discussion on whether to compare the same territory (with more cities for CxxC) or the same number of cities (with more territory for OCP). The truth is somewhere between these two - OCP will end up with both more territory and fewer cities than CxxC.

Not again!

Look, the most fair way would be for a great number of players to play a predermined game both as CXXC and OCP until no more peaceful expansion is possible, note down the difference between each player's best results for each measured in tiles, number of towns and total population. Furthermore, in such a test, settling a distant ring and then filling in the gaps would not be allowed as we're testing expansion, not gaming tricks.

 

[Chamnix]

If you look at it closely, you will realise that rank corruption is beginning to affect the CXXC civ. In spite of having 41 to the 31 towns of OCP, CXXC has a much better net income (before subtracting science) and still has the ability to research at 40% whereas OCP can only manage 20%.

I understand that CxxC is going to be better - my issue is with the claim that the distance corruption of OCP is the relevant factor. The main reason that CxxC is better is not lower corruption.

I'll try to post something in a couple hours that demonstrates clearly where I think the difference comes from ...

 

[Gosnork]

Just looking at the screenshots a couple of things pop out. The OCP civ starts with a much larger territory without having to use the turns required to get there. The earlier contacts for the OCP civ can easily be attributed to this. You also chose Babylon and the cheepest culture in the game, which does little for CxxC.

Despite this CxxC still comes out on top!