SGOTM 08 - XTeam

[leif erikson]

Greetings Fred, for stopping by.

I hope your input will help focus us as we move into the last half of this game.

And X-ville can build Wat significantly faster without sacrificing as much.

Agree.

One thing I'm worried about is all of the focus on building a lot of stuff in Ulundi. In order to build stuff in Ulundi quickly, we have to work mines. But every mine worked is 0.5-1 fewer specialists earning GPPs. Ulundi is a GP farm. IMHO, the best future for Ulundi is to work the 4 food and 1 copper mine, build the forge so that we can hire an engineer, and then maximize GPPs. We'll have ~20% chance of getting a GE each time, and if we get 7-8 more GPs there, there's a good chance that 1-2 will be a GE.

If the 1 pop in each city from HG is important to us, X-ville can build it. If X-ville builds HG, Hagia, Ankor Wat, AP, and Minaret, and hires an engineer and a preist, every GP it produces will have a high chance of being either a GE or Prophet, both of which give us another elusive wonder.

This makes a great deal of sense. Sounds like the best way to go.

Yes, and get a look at the map.

Yes, I think we are nearly there. Can't wait.

 

[killercane]

Something I missed in KC's plan. KC suggests building Wat in Ulundi. Wouldn't it be better in X-ville. X-ville already has Prophet wonders and Prophet GPPs. We'd have a better chance of getting another Prophet in X-ville. And X-ville can build Wat significantly faster without sacraficing as much.

One thing I'm worried about is all of the focus on building a lot of stuff in Ulundi. In order to build stuff in Ulundi quickly, we have to work mines. But every mine worked is 0.5-1 fewer specialists earning GPPs. Ulundi is a GP farm. IMHO, the best future for Ulundi is to work the 4 food and 1 copper mine, build the forge so that we can hire an engineer, and then maximize GPPs. We'll have ~20% chance of getting a GE each time, and if we get 7-8 more GPs there, there's a good chance that 1-2 will be a GE.

If the 1 pop in each city from HG is important to us, X-ville can build it. If X-ville builds HG, Hagia, Ankor Wat, AP, and Minaret, and hires an engineer and a preist, every GP it produces will have a high chance of being either a GE or Prophet, both of which give us another elusive wonder.

Yes but Ulundi is going to outpace Xville 2to1 or so on GPPs. We run the risks of getting too MANY scientists there if we are not careful. We have not run an analysis on how many is too many. There is simply nothing to replace priests in this game. Running 3 priests, engineer, and 3+ merchants in the endgame there is pretty strong. It is not like Ulundi is going to be out of commission for long. I agree we can do HG in Xville, but I like Wat in Ulundi for those 3 priests in the last 40-50 turns. 4 with temple. You can get around not having an engineer but not around not having a priest.

Oh and those priest hammers mean Ulundi can build some other priest/engineer wonder while running those specialists.

 

[ShannonCT]

Yes but Ulundi is going to outpace Xville 2to1 or so on GPPs. We run the risks of getting too MANY scientists there if we are not careful. We have not run an analysis on how many is too many. There is simply nothing to replace priests in this game. Running 3 priests, engineer, and 3+ merchants in the endgame there is pretty strong. It is not like Ulundi is going to be out of commission for long. I agree we can do HG in Xville, but I like Wat in Ulundi for those 3 priests in the last 40-50 turns. 4 with temple. You can get around not having an engineer but not around not having a priest.

OK, glad we see eye to eye about not putting Hanging Gardens in Ulundi.

As for Wat, it probably depends on the length of the game we're shooting for. If we are stopping at Communism, getting the Prophets quickly becomes more important, and we would need only 3 scientists for bulbing (1 for PP and 2 for SM). If we are playing through to MM and Med, we could use 7-10 GSs. The longer the game, the more likely the AI will build the shrines for us, and the longer X-ville will have to produce Prophets.

 

[leif erikson]

Been doing some thinking...

As for Wat, it probably depends on the length of the game we're shooting for. If we are stopping at Communism, getting the Prophets quickly becomes more important, and we would need only 3 scientists for bulbing (1 for PP and 2 for SM). If we are playing through to MM and Med, we could use 7-10 GSs. The longer the game, the more likely the AI will build the shrines for us, and the longer X-ville will have to produce Prophets.

It seems that we are to a point where we should consider which direction this game, shorter or longer? Our decisions are affected, even now, as shown in your post above.

If we beeline to Radio and MM, we add 7 wonders to the list of Pre MT wonders. This means skipping Engineering and Chemistry as Shannon pointed out above.

This would be:

T155 MT
T160 Gunpowder
T165 PP
T172 SM
T183 Physics
T188 Electricity (part bulb?)
T191 Lib-> Radio free
T198 Mass Media

Yes, this would be the beeline for the slightly longer game. It is still only about 65 turns away and we need a lot of hammers!

In my simulations I calculated the number of beakers needed for beelining a number of selected techs and also calculated the number of Wonders enabled (E(t)) by the techs in the beelines. Using a function that simulates the scientific output in terms of beakers per turn it is possible to translate the sum of beakers needed into a number of turns and by dividing this number into the number of enabled Wonders, E(t), it is possible to make an estimate of E(t)/t. I ran simulations with different sets of beeline techs and also with different research speeds and the conclusion was that E(t)/t reaches its maximum when you beeline for MilTrad and Divine Right. This holds with and without Astronomy. Another maximum was Mass Media i.e. the E(t)/t curve has a global Maximum after MilTrad/DR and then a local maximum later at Mass Media. By increasing the simulated beaker output the difference between the two maxima was reduced but it didn't disappear i.e. the conclusion was clear regarding the global maximum for E(t)/t. The numbers for E(t) and beeline beaker sums were the exact number from the game and since the placement of the maximum did not change when the estimated beaker production was pertubed I don't think there can be any doubt that this result is correct.

Are we researching at a greater rate than in Fred's simulation, does anyone know? A faster research rate would help to make a build gap larger because the number of hammers available to build becomes the bottleneck instead of the number of beakers?

Obviously, maximizing E(t)/t is not the whole story since this is only an upper bound on W(t)/t. There are two more important factors: The abilty to close the "capture gap" B(t)-W(t) (i.e. capture the Wonders controlled by other civs) and the ability to close the build gap E(t)-B(t) (i.e. build the remaining Wonders). If the build gap is large and the capture gap small logic says that most hammers should be invested in Wonders and less in units and vice versa.

This tells me we need to find hammers to use to close the build gap. Do we do this at the expense of beakers and research?

Returning to the issue of VC selection Shannon described the 3 most promising scenarios: 1) Religious VC after MilTrad and DR, 2) Diplomatic VC after MM and 3) Domination after MM. It's not easy to select any of these and I will only list some of the pro's and con's of each of them:

1) Pro: Few extra hammers are needed for VC (Missionaries), E(t)/t has it's maximum here.
Con: There is little time to close the build gap.

2) Pro: No extra hammers are needed for VC.
Con: May be unable to fully close the capture gap due to need for votes.

3) Pro: Capture gap is closed 100%
Con: Expensive in terms of extra hammers (units) needed for VC and possibly slower.

Seems we could achieve victory more quickly than we can close the gap.

There must be more pro's and con's. It would be interesting to see an analysis quantifying some of these like how many hammers do we have available from now until DR/MilTrad, and from now until MM. This should give us some indication of how many Wonders and units we can build within a given time frame. An analysis of attack speed also seems necessary in order to understand if option 3) is slow or not. Without an analysis selecting one out of the 3 is random and could potentially lead to achieving a lower W(t)/t than otherwise possible with the current favorable state of the game.

I have been trying to do this. Over the next 30 turns, with our current cities, and reserving Ivoryville and Horseville to build units, we have about 2300 hammers available to us, without Golden Ages. In addition, once we get Jute and our new city, Furville(?), online this number will increase by about 24 hammers per turn.

The build gap is quite large, even with discounts for resources (stone, marble, ivory and gold). Given the techs we know, my estimate is about 1,650 hammers. For the run to Military Tradition and Divine Right, (excluding Drama and Engineering) that adds another 3,300 hammers needed to complete the wonder builds. These estimates do not include forest chops.

Should we choose to go the Mass Media beeline, the wonders would increase the requirement for hammers by 5,625. This is a total of 10,575 hammers.

As I said earlier the game is moving to a phase where decisions on techs and builds will be more and more influenced by the objective of maximizing W(t)/t. What this means is that any decision to research a tech that is not on the beeline to DR/MilTrad and/or Mass Media should be carefully examined since this will cause E(t)/t to decrease. In my opinion it must be possible to show that the loss in E(t)/t can somehow be more than compensated by turns gained in closing the build gap or closing the capture gap. One example is Communism. It’s not on the MM beeline (and so are several other techs listed as “en route” to MM). In order to justify researching Communism it must be shown that this will enable a faster closure of the build gap thus regaining lost turns (this may be the case due to Kremlin and State Property but it’s not a given thing). Other off beeline techs should be subjected to the same scrutiny before they are researched. The same process can be applied to builds. Using hammers for Wonders will help close the build gap and using hammers for units will help close the capture gap. Using hammers for more infra structure is getting increasingly dubious as we approach the end of the game and we should be pretty confident that using hammers for Markets or Courthouses (or other buildings) will significantly contribute to the tech rate before doing so.

We have focused thus far on maximizing beaker output. Is it time to sacrifice some beakers for hammers, such as in Ulundi?

Currently in non-golden age, X-ville and Ulundi are producing about 75% of our hammers, with Pigville, Canalville and Marbleville making up the other 25%.

Horeville will produce 14 base hammer per turn, without the golden age. Perhaps we should devote Horseville to wonders as well?

I think the game is going very well and it’s certainly a candidate for laurels if we can bring it home carefully.

Yes, what do we need to do to make this happen?

The current value of own wonder points W(t) = 60, and W(t)/t = .451
The current value of built (by anyone) wonder points B(t) = 80, and B(t)/t = .602
The current value of enabled wonder points E(t) = 155, and E(t)/t = 1.165

So the build gap is quite large right now.

Right now, about 1,650 hammers.

It's quite clear that we will need Astro for whatever VC we choose, so 12 turns from now, E(t)/t will be 1.069 unless one of the AI researches Divine Right.

I agree, Astronomy is a must.

Then the question becomes: How close to E(t) can we get in ~30 turns?

Not close enough, without more hammers.

1) Only 3 shrines have been built so far. The Buddhist shrine may be built by the AI in the next 30 turns, but the Taoist, Christian, and Islamic shrines may not. And we probably can't reliably grow more than 1 Prophet in the next 30 turns. So that probably leaves 2 wonders on the table.

Joao has the Taoist city where the shrine must be built, Coimbra.

2) The units that we can build in the next 30 turns don't give us a large military superiority. We can probably conquer 2 AI within that time, but that might leave the Great Wall and the Buddhist shrine out of our hands, assuming we are able to locate the owner of the 3 other wonders quickly. Our military superiority (Cuirassiers, Airships, and Frigates) will continue to grow as we conquer useful cities and continue through the tech tree.

Agree with your assessment.

3) Besides the 4 missing shrines, to close the build gap, we would need to build 18 more wonders (plus 6 courthouses and 6 theaters). It's obvious that we can't do this in anything close to 30 turns with our current 8 (soon 10) cities. We clearly need to double or triple our holdings to have the productive capacity to build all the available wonders.

We might be able to capture cities with Courthouses from Joao?

I agree with your assessment of our need to expand. This will hurt our current beaker output. How much of a hit can we afford? Waiting to expand until Military Tradition will allow for fast research but slower city development. Is this trade-off worth it?

Our GP farm in Ulundi is really starting to come into its own now. In the next 30 turns, we'll probably get 4 more GSs from Ulundi, and those GSs can't help us get to MT much faster (they can be settled but not bulb anything on that tech path). The increasing rate at which we get GSs from Ulundi (at least for a while) suggests that we should at the very least play through Communism (bulbing PP and SM), and use the power of State Property and Kremlin to build wonders faster. That tech path only adds 1 more wonder, so E(t)/t will decrease over that period, but without being able to close the build gap, there's no choice.

What if we are able to bulb it. Or if we can bulb some of the beeline techs, we get it with Liberalism. Although Liberalism isn't on the beeline either, it can get us a more expensive tech on the beeline.

The only question in my mind now is whether it is better to use Liberalism for Communism and skip the wonders from Elec-Radio-MM-Bio-Med, or whether our increasing research and potential for generating GSs will allow us to increase E(t)/t and W(t)/t by playing a longer game. In Fred's initial simulation, he predicted a E(t)/t = .98 if stopping at Communism (I guess this didn't take into account getting Communism as the free tech) and E(t)/t = .94 and .93 if stopping at MM and Med respectively. Later, when I gave him some more realistic bpt numbers to plug into his model, he said the difference between stopping at MT and stopping at MM was only 0.1. So I assume that means that he predicts stopping at MM is better than stopping at Communism.

It seems like, given the size of the build gap, we need to look seriously at the Mass Media beeline with a diversion for Communism?

It almost seems like the faster we research, the greater our build gap grows? And, as other civs aren't building wonders, the emphasis for us then becomes building the wonders.

 

[rrau]

What about trading away to select civs (ones a couple down the line in the conquest queue and won't trade it away to other civs so the target civ will hopefully build the wonder just in time for us to take it?) techs that have wonders attached? They can hopefully build wonders while we focus on military?

 

[killercane]

So are you two agreeing with me?

 

[leif erikson]

What about trading away to select civs (ones a couple down the line in the conquest queue and won't trade it away to other civs so the target civ will hopefully build the wonder just in time for us to take it?) techs that have wonders attached? They can hopefully build wonders while we focus on military?

Once we find them and see if they have any hammers to build with? Perhaps this might work. In testing, the civs seem to build wonders in their capitals. If they cluster them like that, it would make life easier, but would also seem to force us into the longer game?

So are you two agreeing with me?

To be honest, I'm not sure what I support atm.

If we plan to beeline to Mass Media to give us more time to build wonders, then it seems as though building Wat in X-ville is the stronger play?

I went searching the known map for production sites. Building a city on the jungle tile one south of our Axe west of Ulundi appears to be a strong site?

The Portugese territory is not very good at all, which may be why they aren't building wonders? I would like to get our hands on Lisbon and the city south of it, Guimaraes (Deer, Sheep and Crab w/Forests and Hills) And we would need to take Coimbra, because Taoism was founded there. The rest could be razed to pay the maintenance of the few?

 

[Cactus Pete]

Once we find them and see if they have any hammers to build with? Perhaps this might work. In testing, the civs seem to build wonders in their capitals. If they cluster them like that, it would make life easier, but would also seem to force us into the longer game? We need to consider this with each of the remaining early wonders we've little immediate need for.
To be honest, I'm not sure what I support atm. If we plan to beeline to Mass Media to give us more time to build wonders, then it seems as though building Wat in X-ville is the stronger play? Can we put the Wat location decision off until we've seen the map and have a better idea how long the game is likely to take?

I went searching the known map for production sites. Building a city on the jungle tile one south of our Axe west of Ulundi appears to be a strong site? I suggested this earlier and it was argued that it would take too long to grow to usefull size.

The Portugese territory is not very good at all, which may be why they aren't building wonders? I would like to get our hands on Lisbon and the city south of it, Guimaraes (Deer, Sheep and Crab w/Forests and Hills) And we would need to take Coimbra, because Taoism was founded there. The rest could be razed to pay the maintenance of the few?

SCT points out that with workshops many of the cities can be productive. We've obviously got to capture some productive cities somewhere soon. Again, let's get a look at the map and try not to box ourselves in in the meantime

 

[killercane]

It is getting too late to build new cities. Captured cities come with pop; we need pop to work hills/workshops/production tiles.

 

[killercane]

Wat is not really stronger in the longer game. Wat is worth 2+9 GPPs in Xville if we run all 3 priests there, which it doesnt really have the food for if we are using it as our main wonder producer. It has +9 excess food and will have cottages instead of irrigation. Ulundi has +12 and some irrigation can make that greater (+16 or more). It is also worth double GPPs in Ulundi.

Wat can be built in 12-13 turns at Ulundi once we get philo and add in the hill worked by canalville.

 

[leif erikson]

It is getting too late to build new cities. Captured cities come with pop; we need pop to work hills/workshops/production tiles.

And so we get back to the essential question.

How much research rate do we trade off, in city maintenance, for the increased production capability to close the build gap? I wish we knew how many of Joao's cities have Courthouses.

Of course, we may soon meet a civ with Wonders and Courthouses and can build a Forbidden Palace after claiming his/her six cities. The one with both religious shrines seems like they would have the best tech rate and be more advanced, but hopefully not too advanced. Please wake me, I think I'm dreaming again.

 

[killercane]

And so we get back to the essential question.

How much research rate do we trade off, in city maintenance, for the increased production capability to close the build gap? I wish we knew how many of Joao's cities have Courthouses.

Of course, we may soon meet a civ with Wonders and Courthouses and can build a Forbidden Palace after claiming his/her six cities. The one with both religious shrines seems like they would have the best tech rate and be more advanced, but hopefully not too advanced. Please wake me, I think I'm dreaming again.

I think it is easy to figure out. We dont have to research full speed to the end, we just have to tweak our research rate. We can turn to 0 and research with a bunch of specialists in non wonder building cities. So the build gap catches up.

 

[Cactus Pete]

Wake up, leif . . . Just because a city has a courthouse doesn't mean it will survive capture. My experience is that less than 50% do, so we're likely to have to build several.

I think the Wat decision is a close call and not a game changer. Unless SCT feels very strongly and wants to make a final arguement, suggest we let the man with the mouse decide.

 

[ShannonCT]

How much research rate do we trade off, in city maintenance, for the increased production capability to close the build gap? I wish we knew how many of Joao's cities have Courthouses.

Of course, we may soon meet a civ with Wonders and Courthouses and can build a Forbidden Palace after claiming his/her six cities. The one with both religious shrines seems like they would have the best tech rate and be more advanced, but hopefully not too advanced. Please wake me, I think I'm dreaming again.

Joao has courthouses in 3 cities: Guimaraes, Lisbon, and Oporto. If you zoom in on those cities, you'll see buildings that look like, well, courthouses. I expect he'll have a couple more by the time we conquer him.

Code of Laws would be a good tech to trade to any civ that doesn't have it yet so they can build more courthouse and perhaps Chichen.

Wake up, leif . . . Just because a city has a courthouse doesn't mean it will survive capture. My experience is that less than 50% do, so we're likely to have to build several.

I think the Wat decision is a close call and not a game changer. Unless SCT feels very strongly and wants to make a final arguement, suggest we let the man with the mouse decide.

Courthouses have a 66% chance of surviving city capture. So we would need to capture 9 cities with courthouses to get an expected 6.

I agree with CP that we need to get a little more info about the map to make an informed decision. If KC can manage to contact some other AI with caravels and get world maps to reveal the other cities, we'll be able to plan our endgame very precisely.

Since we wont be getting Philo until after Astro, the Wat decision can be put off until next turn set. For now, Ulundi should build Forge, grow to the happy cap, and grow a GS/GE.

 

[ShannonCT]

I think it is easy to figure out. We dont have to research full speed to the end, we just have to tweak our research rate. We can turn to 0 and research with a bunch of specialists in non wonder building cities. So the build gap catches up.

Yes, any city not building a wonder at the end can run merchants. The gold can be used to cash rush wonders and the 3 beakers per merchant can help us get the final tech(s) we want.

 

[leif erikson]

Wake up, leif . . .

Do I have to?

I agree with CP that we need to get a little more info about the map to make an informed decision. If KC can manage to contact some other AI with caravels and get world maps to reveal the other cities, we'll be able to plan our endgame very precisely.

Since we wont be getting Philo until after Astro, the Wat decision can be put off until next turn set. For now, Ulundi should build Forge, grow to the happy cap, and grow a GS/GE.

Yes, I agree as well. I'm ready to move ahead.

@Killer - Our discussions, me perhaps, have been all over the map we can't see yet.
Are there any issues you need help in resolving before moving forward with your set?

 

[killercane]

Sure I will play this evening.

 

[leif erikson]

Sure I will play this evening.

Good Luck!

Find us some civs and trade for their maps.

 

[ShannonCT]

I just spent some time calculating what the hammer output would be for some of our present and future cities with fully-functioning workshops (SP + CS + Guilds + Chem). Numbers shown are base hammers per turn (and base hammers per turn during a GA):

Guimaraes 46 (59)
Lisbon 57 (72)
Oporto 44 (57)
Braga 42 (54)
Ulundi 60 (76)
Horseville 44 (62)
Canalville 32 (41)
X-ville 50 (64)
Ivoryville 36 (47)
Jute 52 (66)
Furville 43 (12)

Total 506 (643)

Without any hammer bonuses and ignoring other cities that we will capture, the 10,575 hammers that Leif tallied would take 21 turns to accumulate. Then consider that some of the wonders we build will get a 100% resource bonus, and some cities will get additional bonuses from forges, Bureaucracy, possibly OR. The most expensive, resourceless wonders can be built in the most productive cities (so far X-ville, Ulundi, Lisbon) while some of the slightly less productive cities build the remaining cheaper wonders, and the least productive cities build the buildings that enable wonders (courthouses, theaters, banks, hospitals). It's a lot like a Space Race victory, where we time all of the Big and Little parts to complete at the same time, assisted by cash-rushing. We just need to worry about getting enough Prophets to complete the missing shrines. GEs would be nice, but are certainly not essential.

The discussion about where to put Wat next turnset should revolve around where it will be the most likely to give us Prophets, not where hiring preists will give us a few extra hammers. Preists are a far inferior way of getting hammers compared with workshops.

 

[leif erikson]

I just spent some time calculating what the hammer output would be for some of our present and future cities with fully-functioning workshops (SP + CS + Guilds + Chem). Numbers shown are base hammers per turn (and base hammers per turn during a GA):

Guimaraes 46 (59)
Lisbon 57 (72)
Oporto 44 (57)
Braga 42 (54)
Ulundi 60 (76)
Horseville 44 (62)
Canalville 32 (41)
X-ville 48 (62)
Ivoryville 36 (47)
Jute 52 (66)
Furville 43 (12)

Total 504 (641)

Looks much better when we find some more cities to provide us hammers. I am anxious to see what else lies out there for us to discover...

Without any hammer bonuses and ignoring other cities that we will capture, the 10,575 hammers that Leif tallied would take 21 turns to accumulate. Then consider that some of the wonders we build will get a 100% resource bonus, and some cities will get additional bonuses from forges, Bureaucracy, possibly OR.

I may have done it wrong but when I made the calculation, I halved the number of hammers needed for any wonder that we had the resources for. So the 10,575 number should take into account resource bonuses.

It seems to me, for resource bonus, the game takes the number of base hammers per turn and multiplies it by the bonus. So, if you need marble, the base hammers are doubled and then the multipliers are added?