SGOTM 08 - XTeam

[Frederiksberg]

Just as well I've done enough Physics to understand what you're talking about

For a brief moment I even considered to explain how you should take the derivative of the function you want to optimize and determine when it goes below zero . I guess that old working habits die hard...

Your point about hammers being possibly a limiting factor is valid, but we should investigate carefully the captured wonders thought. Why use our own hammers when we can use someone else's? On the hammers thing of course, the early, cheaper wonders are obviously better since they score just as well for us. I would think therefore that achieving an early victory would generally result in a higher W(t)/t than a later one. Generally.

Captured Wonders are not free because it costs hammers to build the units that are going to do the capturing. Since most units will capture more than one Wonder it is, however, likely that we can strike a good deal here. Particularly if the captured Wonders are expensive and/or the units are built in a city with the Heroic Epic. Captured Wonders may also cost beakers if the captured city is not able to pay for the extra maintenance incurred.

Early victory is not necessarily scoring well because in the initial phase we need to invest most hammers in founding and improving cities. Thus it's likely that we have only a single Wonder after say 50 turns and thus a meager score of 5/50=0.1. I'm sure this is much below the winning score in this competition.

The spreadsheet is organized by listing the wonders in order of the techs as they appear in the tech tree.

Thanks for the spreadsheet. I wish it had a column with the hammer cost of the Wonders

I think we might also have to consider The Apostolic Palace and, if we get far enough, The United Nations must builds. I have had too many problems in games where some knucklehead who controlled the AP screwed up my plans by ending a war I needed to win. Or worse, gave back a city I took that had wonders in it that I needed, a real problem for us in this game should it happen.

I agree with building the Apostolic Palace. Even a late Religious victory is possible if we stay away from Mass Media (Can anyone confirm this?). And it would give us good control of when to win the game.

Not sure we will want to go up the religious leg of the tech tree with so much else to consider.

I think we should at least test the implications of founding some early religions since there is good synergy with producing many Prophets for Shrines and winning a religious victory.

 

[Frederiksberg]

I don't have much knowledge of the new features in BtS (Haven't managed to finish a game yet!) so I have some questions regarding these:

1. Random Events: Does the slave revolt reduce the usefulness of Slavery and is the chance of revolt related to how often you pop rush?
2. Forts: I read somewhere that a fort functions more or less as a city with no citizens, culture etc. Can we use a fort to grab the marble on the island if our culture reaches it?
3. Espionage: Is it worth using in a game where we assume that we will be clear tech leaders?
4. Maintenance cost: I understand that maintenance can be huge in overseas cities. How bad is it and when does it kick in?
5. Worker steals: Do they still work?
6. Early rush: Is it true that the AI has less defenders early on in BtS compared to Vanilla and Warlords?

 

[rrau]

4. maintenace is worse. I have trouble having effective starts over 3 cities.
2. Forts allow passage of ships - up to 2 in a row I think, but don't get the resource under them, so are good ways to cut a passage through a 2 or 3 tile deep land neck.
6. Unknown, I don't play BtS much and when I do, I tend to not get copper AND iron near my starts. Seems to have more frequent bad starts in BtS. Hope this start is nicer.

 

[leif erikson]

I think I can answer some of these.

Random Events: Does the slave revolt reduce the usefulness of Slavery and is the chance of revolt related to how often you pop rush?

I do not use pop-rushing enough and can say that the slave revolt has hit me just as hard when I have not pop-rushed as when I have. I do not think, but also do not know for sure, that pop-rushing effects how often the revolt occurs. I have had, in a GOTM, the slave revolts occur within 5 turns of each other. It hurts to lose that much pop early!

Forts: I read somewhere that a fort functions more or less as a city with no citizens, culture etc. Can we use a fort to grab the marble on the island if our culture reaches it?

Yes. This is being discussed in the BOTM10 now. Please see this post by Lexad. It shows quite readily how forts work. (Even has screenies!)

Espionage: Is it worth using in a game where we assume that we will be clear tech leaders?

It is worth using for two reason imho. The first is that you can't see the graphs of the AI players without having a certain level of espionage. Secondly, if one starts to get ahead, planting a Great Spy will enable us to see all sorts of things about them that we may need to know.

Maintenance cost: I understand that maintenance can be huge in overseas cities. How bad is it and when does it kick in?

Yes, the maintenance costs are higher because I think the designers want you to break off new colonies. I think it is manageable if we save the Forbidden Palace for overseas duty, or build or claim Versailles. In worst case, there is always State Property.

Worker steals: Do they still work?

I think the big difference here will be that Workers have guards. It may be harder to steal them, we may have to fight for them.

 

[killercane]

Early victory is not necessarily scoring well because in the initial phase we need to invest most hammers in founding and improving cities. Thus it's likely that we have only a single Wonder after say 50 turns and thus a meager score of 5/50=0.1. I'm sure this is much below the winning score in this competition.

I agree with building the Apostolic Palace. Even a late Religious victory is possible if we stay away from Mass Media (Can anyone confirm this?). And it would give us good control of when to win the game.

I think we should at least test the implications of founding some early religions since there is good synergy with producing many Prophets for Shrines and winning a religious victory.

I figure we can get an AP victory in around 125-150 turns (all of these are guesses but I hope educated guesses). How many wonders can we build in that time? Certainly not 20 even. Whereas, I think we can build/capture 55ish in 250 turns. So I am definitely in Fred's camp here.

Though I dont know why we would skip MM in favor of AP voting. The mechanics are essentially the same, and MM/Radio and that stuff opens up several new wonders. Remember you cant self vote anymore...

I also disagree with founding early religions. Assuming fast tech speed, the early ones will result in delaying our science.

 

[killercane]

I don't have much knowledge of the new features in BtS (Haven't managed to finish a game yet!) so I have some questions regarding these:

1. Random Events: Does the slave revolt reduce the usefulness of Slavery and is the chance of revolt related to how often you pop rush?
2. Forts: I read somewhere that a fort functions more or less as a city with no citizens, culture etc. Can we use a fort to grab the marble on the island if our culture reaches it?
3. Espionage: Is it worth using in a game where we assume that we will be clear tech leaders?
4. Maintenance cost: I understand that maintenance can be huge in overseas cities. How bad is it and when does it kick in?
5. Worker steals: Do they still work?
6. Early rush: Is it true that the AI has less defenders early on in BtS compared to Vanilla and Warlords?

1. Slave revolt is -2 pop to get it over with. It is a pain, but slavery will be necessary for the hammers. With Redentor, you can switch between slavery and caste quite a lot. So say we reach Communism, with 10 wonders left to build at the moment. You can switch between Representation/Uni Sufferage, and Slavery/Caste to get beakers and production on alternating turns. Then we can switch to Rep every other turn and finish out research on some wonder enabling techs like Democracy or whatever that we previously skipped.
2. This one has alternating viewpoints from what I know, it needs to be tested. (Or follow Leif's link).
3. We can use espionage to see what our tech buddies are researching and adjust accordingly, and steal some random tech or two in the Renaissance to supplement our tech rate.
4. Overseas maintenance kicks in after 2 cities on the same landmass. With Free Market, overseas cities are pretty strong for a while until you amass quite a few of them since you get "overseas" trade.
5. Yes worker steals are good as always. Workers are unguarded early and when the AI begins to build second and third cities. They like to road unprotected to these new destinations.
6. The AI is lightly defended at the beginning, similar to human play. On Monarch though, warrior rush is not as effective and often is the incorrect play with the free archer. Archer rush is expensive, and axe rush takes away from our tech rate unless the AI is very close.

 

[Cactus Pete]

CP checking in and immediately checking out . . . I'll be in for good on 9/20.

Glad to have killercane aboard. What/Who is Redentor?

 

[Frederiksberg]

Thanks for all the answers to my BtS questions

Though I dont know why we would skip MM in favor of AP voting. The mechanics are essentially the same, and MM/Radio and that stuff opens up several new wonders. Remember you cant self vote anymore...

One thing that could force us to skip MM is the tech cost of getting there (And the cost of the Wonders). Suppose that we have captured and built all "tech enabled" Wonders by then and our score is around 1.0. Then we would have to tech to MM and build the UN in 3 turns to make it a good deal. I'm sure something like this will happen at some point - only question is when.

Regarding self vote - is it possible to form a colony and get around the self vote limitation that way?

I also disagree with founding early religions. Assuming fast tech speed, the early ones will result in delaying our science.

Reduced tech speed is indeed a concern. It's probably best if the first GP is a Great Scientist so that we can get an Academy in our Capital. Then the 2nd should probably be a Prophet and it would be nice if we had a religion by then so that we can get a Shrine. Maybe Confu or Christianity would match such a time line.

4. Overseas maintenance kicks in after 2 cities on the same landmass. With Free Market, overseas cities are pretty strong for a while until you amass quite a few of them since you get "overseas" trade.

This indicates that the GLighthouse could be very important to get.

 

[leif erikson]

What/Who is Redentor?

It is a new wonder in . It can be built once you research Radio and allows you to change civics or religion without anarchy and allows you to change civics or religion every turn if you wish.

Regarding self vote - is it possible to form a colony and get around the self vote limitation that way?

How does self-vote work now? Will vassals help? Problem is that vassal wonders do not count.

I played a random game as Khmer last night to get a feel for things. The good news is that wonders got built, and were clustered in capitals for the most part. So much will depend on the map and where things get located. Need to play a bit more.

The normal game is 500 turns instead of 460.

 

[DJMGator13]

Gator checking in.

Is everyone aware of the favorite civic bug that was discussed in the BOTM09 pregame thread? Apparently once the AI is in their favorite civic you can no longer get them to change any civics. Here's a link. Not sure this will come into play but wanted to make sure everyone was aware of it.

(any XMLers here?)

Not really, that has been one of our problems. We have some very good readers and researchers but no one who delves into the code itself like some teams.

 

[DJMGator13]

Here's another link to the "updated" reference chart. On the title page you can select the game settings (monarch, normal, standard) and then go to the Wonders tab to see all the costs. The w/s has been recently updated for v3.17.

 

[Mad Professor]

I figure we can get an AP victory in around 125-150 turns (all of these are guesses but I hope educated guesses). How many wonders can we build in that time? Certainly not 20 even. Whereas, I think we can build/capture 55ish in 250 turns. So I am definitely in Fred's camp here.

So it boils down to the beaker cost of the techs required to build the wonder and the hammer cost of building it. If we have a tech lead, then wouldn't htis makee the wonder capture a whole lot less useful? It would only help us for those wonders we decided to skip ourselves. There would be a few of those no doubt - we can't build them all surely? I presume this is what you're pointing to here killer? Given this scenario, we should perhaps deliberately avoid quite a list of wonders so that we can capture them later? It would make more sense then to avoid the more hammer intensive ones, or the ones requiring beaker-expensive techs that are not on a beeline to something else important - like MM!

 

[ShannonCT]

For a brief moment I even considered to explain how you should take the derivative of the function you want to optimize and determine when it goes below zero . I guess that old working habits die hard...

After that, I'd like to see a delta-epsilon proof that the derivative actually exists at that point.

Not really, that has been one of our problems. We have some very good readers and researchers but no one who delves into the code itself like some teams.

The XML is pretty easy to figure out. The XML documents mostly contain values of variables (e.g. the cost of a unit, the AI tech discount at different levels, or a leader's propensity to go to war). The C++ documents are the real brains of the game. I learned JAVA (which is very similar to C++) programming in college and have passable ability at reading C++ code. If we have any questions about how something works, I can take a stab at finding the answer. I guess one problem we had in SGOTM07 was relying on inaccurate strategy guides.

 

[leif erikson]

After that, I'd like to see a delta-epsilon proof that the derivative actually exists at that point.

Nice to see you back!

Now what the hell are you talking about????


How are we going to set up tests for this game?

And, as Alan cautioned in the maintenance thread, please let us be very careful to play the practice save and not the actually save, as the game save has been posted and is available. My method is to keep them in separate folders.

 

[killercane]

So it boils down to the beaker cost of the techs required to build the wonder and the hammer cost of building it. If we have a tech lead, then wouldn't htis makee the wonder capture a whole lot less useful? It would only help us for those wonders we decided to skip ourselves. There would be a few of those no doubt - we can't build them all surely? I presume this is what you're pointing to here killer? Given this scenario, we should perhaps deliberately avoid quite a list of wonders so that we can capture them later? It would make more sense then to avoid the more hammer intensive ones, or the ones requiring beaker-expensive techs that are not on a beeline to something else important - like MM!

The beakers are easier to acquire than the needed hammers. The ideal situation would be to have in place everything to finish as many wonders as possible on the last turn. Assuming MM, this means, engineers to finish UN and Hollywood, and an artist to make Civ Jewelers. To fully rush both wonders, one would need high population or extra gold to finish a rush job. Further, Kremlin works for slavery cost as well as gold cost, and say a size 30 city can pop rush 15 citizens the last turn to finish something useless like Hermitage or whatever other wonder that was skipped. Im starting to think that to get the needed pop to get the most out of engineers we should just concentrate on domination, and have settlers in place to found on one of the last turns and enjoy the creative border pops.

If playing as SP, I would rush to curassiers and then airships (Physics). Earlier wars, esp. with the possibility of being on a mid- to high- water map, will slow us down techwise.

Lets compare the AP and MM approaches.

Build AP and win quickly:

1. Henge
2. Pyramids
3. Wall
4. Oracle
5. ToA
6. Paya
7. SoZeus
8. Parthenon
9. HGardens
10. Colossus
11. MoM
12. Her. Epic
13. Nat. Epic
14. TGLibrary
15. Sistines
16. Chicken Itza
17. Angor Wat
18. Hagia Sophia
19. AP
20. Notre Dame
21. 5 religious shrines

Total 26 by turn 150. 1 wonder every 6 turns.

MM approach- 55-58 out of 250 turns. Almost 1 wonder every 4 turns.

So that begs the question, are these numbers feasible? What is the best way to answer that question?

 

[leif erikson]

Played a test game to try and get a feel for dates. Pulled from the log:
3720 BC - Buddhism is founded
3360 BC - Hinduism is founded
2960 BC - Judaism is founded
2120 BC - Stonehenge
1440 BC - The Great Wall
1120 BC - Kashi Vishwanath
825 BC - Temple of Solomon
550 BC - Temple of Artemis
375 BC - The Great Lighthouse
275 BC - Confucianism is founded
125 BC - The Oracle
1 AD - The Pyramids
25 AD - The Great Library
225 AD - The Colossus
660 AD - The Statue of Zeus
720 AD - Christianity is founded
820 AD - The Hanging Gardens
960 AD - The Mausoleum of Maussollos
1060 AD - Taoism is founded
1100 AD - The Apostolic Palace
1120 AD - Chichen Itza
1190 AD - Angkor Wat
1210 AD - Hagia Sophia
1220 AD - The Parthenon
1240 AD - Shwedagon Paya
1380 AD - The Sistine Chapel
1410 AD - Notre Dame
1470 AD - The Taj Mahal

Astronomy was not needed to get to cities that had wonders in them. On one side of the globe, there was ocean that needed to be crossed but on the other side, we could get through. It would be easy for the map maker to make it required though.

Wonders were mostly located in capitals:
Washington - Temple of Solomon, The Kashi Vishwanath, Stonehenge, Angkor Wat, Chichen Itza
Karakorum - The Great Lighthouse, The Colossus, The Oracle, Notre Dame, The Great Wall
Beijing - The Pyramids, The Hagia Sophia, The Sistine Chapel, The Taj Mahal.
Guangzhou - The Hanging Gardens, The Temple of Artemis, The Mausoleum of Maussollos.

 

[killercane]

I'll add what dates IME these wonders fall at Monarch.

1. Henge- 2000-2500 BC
2. Pyramids 500BC-500AD
3. Wall- same as Henge
4. Oracle- 1700 BC-700 BC, that is an outrageously late Oracle build in Leif's game
5. ToA- 1000BC-500BC
6. Paya-after 1 AD
7. SoZeus-same
8. Parthenon-1 AD-750 AD
9. HGardens-500 BC-500 AD
10. Colossus-500AD
11. MoM-500-1000AD
12. Her. Epic
13. Nat. Epic
14. TGLibrary- 250AD-1000 AD
15. Sistines- Dont know
16. Chicken Itza- Dont know
17. Angor Wat
18. Hagia Sophia
19. AP-500 BC-1000 AD
20. Notre Dame
21. 5 religious shrines

 

[rrau]

Could someone post the start of a practice game? Usually, I'd make my own, but I'm fairly busy and don't have the time.

 

[leif erikson]

Could someone post the start of a practice game? Usually, I'd make my own, but I'm fairly busy and don't have the time.

I've been playing random saves to get a feel for timing of builds.

I'll try to put one together of the start for you this evening.

 

[Frederiksberg]

After that, I'd like to see a delta-epsilon proof that the derivative actually exists at that point.

Since W(t) is a sum of step functions the derivative will be a sum of impulse (delta) functions. At least from a practical (engineering) point of view .

Total 26 by turn 150. 1 wonder every 6 turns.

MM approach- 55-58 out of 250 turns. Almost 1 wonder every 4 turns.

These numbers indicate that we should expect good final final scores to be somewhere in the range 0.87 to 1.16.

Wonders were mostly located in capitals:
Washington - Temple of Solomon, The Kashi Vishwanath, Stonehenge, Angkor Wat, Chichen Itza
Karakorum - The Great Lighthouse, The Colossus, The Oracle, Notre Dame, The Great Wall
Beijing - The Pyramids, The Hagia Sophia, The Sistine Chapel, The Taj Mahal.
Guangzhou - The Hanging Gardens, The Temple of Artemis, The Mausoleum of Maussollos.

This is also a nice piece of information. We should expect multiple Wonders in AI capitals. This clearly increases the value of capturing some of the Wonders. I've been looking at the hammer cost of captured Wonders in order to get some guidelines regarding which Wonders we are better of building ourselves and which Wonders we are better off capturing. In the Captured Wonder Cost Excel sheet I try to get an estimate of this cost based on input such as the composition of our attack stacks, the survival rate of our units, the number of cities captured per stack and the number of Wonders per captured city on average. It seems to me that captured Wonders are likely to cost 150-250 hammers a piece but much depends on the assumptions made regarding the mentioned input parameters. Anyway this seems to indicate that only a limited number of World Wonders are cheaper to build than to capture. We should, of course, also think about the effect of each Wonder which may make it more attractive to build rather than capture.