SGOTM 08 - XTeam

[Frederiksberg]

Updated Wonder Cost

 

[leif erikson]

The wonders clustered in a city will make the capture cost more efficient. The next problem might be the beakers needed for Astronomy and the hammer cost of transports and escorts.

btw - I got a message that says the updated wonder costs is an invalid attachment.

For rrau, please find attached a save to test with. I modified a random start.
Hope that Ike isn't troubling you too much rrau.

 

[leif erikson]

Been playing another game.
OMG!! Did they nerf bombardment units!

Bombarding a walled city, with 125% culture, with 8 Cats and 6 Trebs required 5 turns!

Wars take forever...

 

[Frederiksberg]

btw - I got a message that says the updated wonder costs is an invalid attachment.

That's odd - the link works for me. I have uploaded it again under a new name:

Updated Captured Wonder Cost

Been playing another game.
OMG!! Did they nerf bombardment units!

Bombarding a walled city, with 125% culture, with 8 Cats and 6 Trebs required 5 turns!

Wars take forever...

I looked in the Civilopedia and what you did was probably to attack a city with walls and castle. These buildings will decrease the effect of cats/trebs by 50%+25%=75%. This means that each cat bombards only 0.25*8=2% of the defense and each treb 0.25*16=4%. So per turn you should be able to reduce by 8*2+6*4=40%. In Vanilla and Warlords the percentages were relative to the maximum i.e. 3 turns should be enough. If the percentages are absolute it seems that 4 turns would do. Are you sure it was 5 turns?

Anyway this does indicate that we need more bombardment units in our stacks.

 

[leif erikson]

I looked in the Civilopedia and what you did was probably to attack a city with walls and castle. These buildings will decrease the effect of cats/trebs by 50%+25%=75%. This means that each cat bombards only 0.25*8=2% of the defense and each treb 0.25*16=4%. So per turn you should be able to reduce by 8*2+6*4=40%. In Vanilla and Warlords the percentages were relative to the maximum i.e. 3 turns should be enough. If the percentages are absolute it seems that 4 turns would do. Are you sure it was 5 turns?

Anyway this does indicate that we need more bombardment units in our stacks.

It also means starting war earlier than I did. I didn't expect to do so little damage and sure got a surprise!

The fifth turn was when the city actually got to zero.

The other thing I was thinking about is having an early war to gain a Great General and making him an instructor in a city that can pump out Cats and Trebs with Accuracy as a second promotion. The real meaning for me is that we will need more transport ships to go to other land masses which delays things unless we plan quite far ahead.

Points to early work boats for recon!
And adds to the cost of seizing cities...

 

[leif erikson]

A summary of build times, from the log:
3720 BC - Buddhism founded.
3240 BC - Hinduism founded.
2160 BC - Stonehenge
1280 BC - Oracle (a lot earlier this time!)
1200 BC - Judaism founded.
900 BC - Great Wall
450 BC - Colossus
400 BC - Temple of Solomon
350 BC - Great Lighthouse
125 BC - Temple of Artemis
100 BC - Confucianism founded.
25 BC - Parthenon
25 AD - Great Library
125 AD - Statue of Zeus
200 AD - Pyramids
325 AD - Christianity founded.
450 AD - Shwedagon Paya
500 AD - Taoism founded.
580 AD - Mausoleum of Maussollos
720 AD - Apostolic Palace
880 AD - Chichen Itza
1020 AD - Hanging Gardens
1030 AD - Hagia Sophia
1070 AD - Mahabodhi
1080 AD - Sistine Chapel
1200 AD - University of Sankore
1370 AD - Islam founded.
1410 AD - The Masjid al-Haram
1460 AD - Angkor Wat
1490 AD - Church of the Nativity
1505 AD - The Kashi Vishwanath
1540 AD - The Spiral Minaret

Turn 220 is 1550 AD

Again, Wonders are clustered in cities. I took a Chinese city that had Pyramids, Hagia Sophia, Chichen Itza, Sistine Chapel and Mausoleum of Marssollos in it.
Thebes contains Stonehenge, Spiral Miaret, University of Sankore, Statue of Zeus, Shwedagon Paya and Apostolic Palace.
Timbuktu has Kashi Vishwanath, Colossus and Oracle
Amsterdam has Temple of Artemis and Great Wall
In three cities there are 11 wonders.

EDIT - And in this game, Astronomy was required!

 

[killercane]

It also means starting war earlier than I did. I didn't expect to do so little damage and sure got a surprise!

The fifth turn was when the city actually got to zero.

The other thing I was thinking about is having an early war to gain a Great General and making him an instructor in a city that can pump out Cats and Trebs with Accuracy as a second promotion. The real meaning for me is that we will need more transport ships to go to other land masses which delays things unless we plan quite far ahead.

Points to early work boats for recon!
And adds to the cost of seizing cities...

Take your save, and worldbuilder in some airships and curassiers. It should be much easier to eliminate cities this way. No need of catapults which are nerfed too much.

 

[leif erikson]

Take your save, and worldbuilder in some airships and curassiers. It should be much easier to eliminate cities this way. No need of catapults which are nerfed too much.

Too bad we can't do that during the competition.

As I think about it, another thing I am having trouble doing is getting the Apostolic Palace and Great Library, Teching to Literature, with Asthetics in there, delays getting to Theology. I should try going to Theology first and then heading for Literature?

 

[killercane]

Too bad we can't do that during the competition.

As I think about it, another thing I am having trouble doing is getting the Apostolic Palace and Great Library, Teching to Literature, with Asthetics in there, delays getting to Theology. I should try going to Theology first and then heading for Literature?

Trade for Mono with the AI, research CoL, pop a Priest for Theology while researching through to Literature.

 

[leif erikson]

Trade for Mono with the AI, research CoL, pop a Priest for Theology while researching through to Literature.

I found another good solution. Researched Mono, traded for Priesthood, built The Oracle while researching towards Lit. Took Theology as a free tech and built both the Great Library and the Apostolic Palace. And founded Christianity as well!!

Of course, in building the AP, one must choose a religion and make some others mad, but what the heck!

 

[killercane]

I found another good solution. Researched Mono, traded for Priesthood, built The Oracle while researching towards Lit. Took Theology as a free tech and built both the Great Library and the Apostolic Palace. And founded Christianity as well!!

Of course, in building the AP, one must choose a religion and make some others mad, but what the heck!

The only problem with this is you take a great person highlightable technology with the Oracle, rather than one that actually requires researching. There is no rush to found Theology, unless you are going for an early AP win; the AI is pretty slow to get there. With the priest points from building a stonehenge and Oracle, you can easily found Theo. In this context, we can go for "sorta early" AP win with the caveat of building a lot but this doesnt necessarily require Theology.

Now, Theology does not help our research. There are only a few Research Technologies, including Currency, CS, Lit, Metalcasting, Education, Printing Press, and perhaps Calendar. What Theology does do is open up paper, in order to take an expensive Research Technology (Education) with Oracle. We also get maps this way from any civs around and about. The problem here is that paper is a manual research, no highlight of it unless we have math and alphabet.

 

[killercane]

To up our post count:

The merits of trying to take Education overall arent as significant as the standard CS slingshot. One can always bulb education with a scientist, in order to bring down its large time a bit. The problem only arises if we are lacking in a GP at the end of the game (assuming a Radio/MM game).

So I am still sold on a standard high research game, including CS slingshot, fast Oxford, and then on to Curassiers to eliminate a couple AIs, and then add airships in to eliminate the rest. Win by domination with Sushi or try UN if that is faster. The trick is keeping UN option open while eliminating AIs and not getting 75% of the vote.

I talked about Worldbuilder earlier. If you havent tried out the curassier/airship combo, definitely try it and compare it with standard siege+units. I think it is quite a bit faster, and efficient in less hammers spent in units. Airships are new with BTS, and you can use them to bomb units to a 25% HP reduction. So a musket in 40% defense city, would be 30% chance to win with standard curassier, but if you bomb it to 6.8/9, it becomes a favorable battle for your currassiers.

 

[Frederiksberg]

Some more strategy thoughts for discussion and common reference:

In the drawing below is shown 3 curves for the functions E(t), B(t) and W(t) where t is the turn number and:

E(t) is the number of Wonders that may be built at turn t including all that has been built already. So this number is a function of the known techs.

B(t) is the number of Wonders that are already built at turn t including World Wonders built by the AI.

W(t) is the number of Wonders controlled by us.

Obviously we have that E(t) >= B(t) >= W(t) and from this follows that the score W(t)/t <= B(t)/t <= E(t)/t. Since our score is upper bounded by B(t)/t and E(t)/t it is in our interest to maximize these upper bounds and to close the gaps:

Build Gap: BG(t)=E(t)-B(t) (The vertical distance between the blue and green line)
Capture Gap: CG(t)=B(t)-W(t) (The vertical distance between the green and red line)

The Build Gap can be closed by building more Wonders and the Capture Gap can be closed by capturing more Wonders. When the AI builds a Wonder the BG will go down and the CG up.

E(t) is maximized by fast research, tech trades.

B(t) is maximized by maximizing hammers and building Wonders. Gifting/trading Wonder techs to the AI can also help and so will gifting/trading resources that speed up Wonders.

W(t) is maximized by capturing Wonders (build military) and by maximizing B(t).

Imagine that we are able to predict E(t)/t with a good accuracy. Then we also know that if we can close the gaps exactly at the moment when E(t)/t has it's maximum and win the game at that same moment we can do no better (Given our ability to maximize E(t)/t - other teams could still be far better). This gives us an instrument with which to navigate and determine when to start closing the gaps and look for a victory. A practical approach would be to keep track of E(t)/t, B(t)/t and W(t)/t during the game and even more important make forecasts of them into the future.

It may sound easy but of course there are tons of practical problems. Most prominently there are the trade offs between producing hammers vs beakers, between building units and building Wonders, between using money for city/unit maintenance and for science,.... the list goes on . Anyway those are the problems we must solve in this game.

 

[leif erikson]

Nice work Fred!

I'll try to play a test game tomorrow and see if we can put this into practice. I assume that once a tech is discovered that enables a wonder, it goes into the E(t) category? That will be hard to do until we discover Alphabet and know all our adversaries. I don't think that will effect much though as it shouldn't matter that much in the early game.

The game will tell us when a wonder is finished, so B(t) should be relatively easy to track. And I think we'll know when we build one...

I think we will need to make some pretty specific tech tree decisions. I have been starting test games with Bronze Working to allow forest chops, then on to Fishing to chop work boats for food and recon, and then on up with Animal Husbandry, Wheel, sometimes Pottery and then Writing. May need Masonry to capture that Marble.

On the start, looks to me like we want to move southeast so that we can capture the Marble in the fat cross and still keep the other resources as well?

@Killer - Just curious, are you advocating that we avoid war until we have Curassier and Air Ships? That would allow a very different trip up the tech tree, I would think?

 

[ShannonCT]

I think we will need to make some pretty specific tech tree decisions. I have been starting test games with Bronze Working to allow forest chops, then on to Fishing to chop work boats for food and recon, and then on up with Animal Husbandry, Wheel, sometimes Pottery and then Writing. May need Masonry to capture that Marble.

I'm thinking AH first would be the winning move. Then BW and Fishing. Cows and Gold mine are the best two tiles.

On the start, looks to me like we want to move southeast so that we can capture the Marble in the fat cross and still keep the other resources as well?

We dont need marble in the BFC to be able to use it later. At 500 culture, we get third border expansion, and would thus capture marble in our borders if we settle in place. With Maths, we can build a fort and hook marble. Marble is not needed to build the cheap Oracle.

Best argument for settling 1 SE or 2 E is being able to get a workboat to the fish earlier. If we settle in place, a WB may have a long trip to get to fish. But settling in place gives more land tiles (more forest chops and more mines) in the capital.

@Killer - Just curious, are you advocating that we avoid war until we have Curassier and Air Ships? That would allow a very different trip up the tech tree, I would think?

An early rush of a nearby AI is always a winning move. We should be looking to take out an AI that doesn't like to trade. Let's see if we find copper or horses.

 

[killercane]

I think we will need to make some pretty specific tech tree decisions. I have been starting test games with Bronze Working to allow forest chops, then on to Fishing to chop work boats for food and recon, and then on up with Animal Husbandry, Wheel, sometimes Pottery and then Writing. May need Masonry to capture that Marble.

On the start, looks to me like we want to move southeast so that we can capture the Marble in the fat cross and still keep the other resources as well?

@Killer - Just curious, are you advocating that we avoid war until we have Curassier and Air Ships? That would allow a very different trip up the tech tree, I would think?

I played the test game. AH is best to open writing, but you want to build fishboat after worker1. So I start with fishing, AH, then writing while skipping BW. City 2 allows passage to the fish, and builds another workboat and some warriors. We really need some stronger secondary cities than this test game and a more reasonable neighbor! But anyways CoL discovered 1000 BC, Oracle 1 turn later, no fort on Marble yet. Should be very similar in the real game. The only question is on whether there are some oracle builders amongst AI that will foil our plan, but there is nothing to be done there, its just bad luck.

RE: war. My opinion is you take it when you can get it. If we are close, sure. If we find an AI within a few turns, and can get horses, we should be able to make chariots and add two cities from the AI, especially if they are alone. If we arent, the fastest way is via the curassiers, and then add in airships. You prebuild the curassiers, and upgrade the rest from HAs and chariots. If you can upgrade 10 you can pound on through without stopping.

 

[Frederiksberg]

I'll try to play a test game tomorrow and see if we can put this into practice. I assume that once a tech is discovered that enables a wonder, it goes into the E(t) category? That will be hard to do until we discover Alphabet and know all our adversaries. I don't think that will effect much though as it shouldn't matter that much in the early game.

Yes, your assumption is correct. It's not necesarry to record these functions in the early game where ending the game is not an option. I'm sure you can wait until Alpha. It's going to be interesting to see how these numbers develop.

In order to speed up test games it might be relevant to start focusing a little on the opening sequence i.e. the first 30-40 turns. Then we can start our tests from that point on instead of having to repeat those decisions over and over.

 

[leif erikson]

Played a test game to try to figure out a technique to put Fred's ideas into practice.

I've attached an XL file with the data in it. What I did was add 5 to each category as I played. I tried to get it to make a sensible graph, but it didn't like me this morning!

If we can make anything out of it, we may be able to automate it to make it easier while we play the game.

 

[leif erikson]

I just realized what I did wrong. XL treats each turn as a data point, so all the zero values between the events would make the graph read as zero. If we keep a running total in the spreadsheet, that should work.

As you can see from the data entered, it is pretty dull early on and starts to get busy after turn 100 plus. Playing on would, I believe, show it getting even busier.

 

[Frederiksberg]

I played the test game. AH is best to open writing, but you want to build fishboat after worker1. So I start with fishing, AH, then writing while skipping BW. City 2 allows passage to the fish, and builds another workboat and some warriors.

Can you tell in greater detail what your opening sequence was? Normally we do very detailed testing of the opening because this can be simulated very closely using the test game. The interesting stuff are build orders (i.e. worker, wb, ....), tech path (Fishing, AH, ....) and in some cases also discussions on which tiles to improve/work.

With all the forests and food BW may be strong early because we can get the first settler out fast using chops/whip. I think I read somewhere that the slave revolt can only happen in cities above size 4 (or was it 5) and only after Writing has been discovered. Is this correct?

I just realized what I did wrong. XL treats each turn as a data point, so all the zero values between the events would make the graph read as zero. If we keep a running total in the spreadsheet, that should work.

Yes, thats the way to do it. It would aslo be nice to have columns showing E(t)/t, B(t)/t and W(t)/t. I'm afraid I don't know how to do this easily in Excel. It will be really interesting to get a rough idea of when E(t)/t starts to decrease because this will give us an approximate victory date to aim for.