G-Major 14

[WastinTime]

This Gauntlet is being a success in many an aspect, even though we might not reach the 1000AD threshold (come on, people, we can do it!).

Does your 1000 AD game have to be on Stanard size? Although I believe it to be an impossible goal, you'd have a better shot on Small I think.

 

[jesusin]

Does your 1000 AD game have to be on Stanard size? Although I believe it to be an impossible goal, you'd have a better shot on Small I think.

For this Gauntlet it has to be std. For my own personal goal, any size would do. I play both std and small maps. Maps smaller than small feel a bit unreal. Maps bigger take too much time.

EDIT: Your first sentence made me think you had already done it in other size. The second one took away my excitement.

 

[FiveAces]

There is a whole thread on this, but you can't bomb your PA cities. You can, however, gift them GA's and hope, or add them as specialists. (vanilla civ--may have changed in Warlords or BTS)

Ah well that kills it then as I only have vanilla. Thanks for letting me know before I committed to a marathon game. Gonna try having Liz found Budda, Confu, then alpha while generating a GS for Philo/academy and hope I have copper. I've been having luck with this as Saladin, but I'm still getting to Lib too late - like 400AD or so. Hopefully the extra commerce from Liz makes the difference. Budda will require a bit of luck, but I've done it before with a 2g tile and non-myst AI.

 

[Alan_of_Dale]

Finally got a win. Big mistake in early GA bombing the capital since it went legendary about 1555 AD. As a result my time bombed out to 1665 AD which should take last place. I had 14 GA (no GS) for 7-6-1, but should have bombed 7-7-0. Would have saved me 3 turns. Used Saladin, quick, inland sea, medium sea level. Founded Buddhism and Confucianism, but missed music first due to fumbling my research as I researched mono and monarchy first. Should have gone straight to music after alphabet. Founded only three cities. 15 pop capital with nine cottages, two gold mines and pig, and had two cathedrals and Hermitage, and 5 artists for 624cpt. 12 pop GA city had 7 artist specialists and NE for 114cpt. Third city had 11 pop and 5 artists for 144cpt. Anyway, at least I have got some things down now. May give Saladin another try as I should be a lot better than this. Also, was experimenting with just three cities, but this probably slowed my research down ultimately, and could have had more religious buildings as well.

 

[Sun Tzu Wu]

Anyway, there does seem to be a certain "cheesy" aspect to this type of strategy. You have to leave yourself so under-protected at Deity level -- some of my cities were only defended with one warrior -- the AI should really NOT let you get away with that (and, my understanding is, it no longer does in BTS -- why aren't I playing that rather that obsessively trying to play this vanilla gauntlet??? I am making no sense here!) It seems like you can only win by exploiting a significant defect in the AI. Can anybody pull this off in BTS? (Uh, oh! I probably should keep my mouth shut.)

The only peace-loving reason for needing a Warrior in a city is to reduce unhappiness due to not having any military in the city with typically Population 3 or more.

It is perfectly reasonable to expect that one can avoid wars with peaceful civilizations, even though one may have only a single Warrior in each city.

If you do all the right things to make all civilizations pleased/friendly, I would consider it a diplomatic defect in a peaceful AI if _does_ attack you.

On the other hand, if you are playing with any non-peaceful AI leaders like Alexander, Brennus, Genghis Khan, Montezuma, Napolean or Ragnar, you must not expect "peaceful" diplomacy alone to avoid war with them. In such cases, you can bribe others to keep the non-peaceful civilizations busy.

In conclusion, a peaceful AI civilization should not attack a leader/civilization whose diplomatic relations are at least pleased, no matter how weak their military is. This is not an AI defect; this is an expected effect of solid, good diplomacy. (This is not to say there aren't AI defects, but I'm not sure that any of them would be serious enough to be considered exploits = a bug = something contrary/in contradiction to the game design.)

I'd prefer to _not_ see the phrase "AI defect" used too freely when the AI simply does something unexpected or fails to do something expected. In such cases, rather than say the AI has a defect, we should consider the situation from the "neutral" position of the game designer. Looking at the situation from a neutral (game design) viewpoint, it may become clear why the AI's decision/lack thereof is reasonable after all.

One more point: The AI in both Civ 4 and Warlords is quite primitive. Any talented and determined Human can beat it on almost any level. Only on Deity level does the game truly challenge the expert player. (I don't have BTS yet, so I will avoid comments on it + its not allowed in G Major-14 anyway.) The AI is improving, but there is no way it can yet compete with a talented and determined Human. In this regard, the AI is grossly deficient compared to any Human player. However, the AI's many deficiencies/limitations are almost never defects, because the AI plays as it designed to play (as well as it can, but no where near the level of a Human). At Noble level where neither the AI or Human have game level related advantages over each other, a Human player can win without consciously taking advantage of any AI deficiencies. At Deity level, the Human player must take advantage of many/most/all of the AI's deficiencies to even have a chance to win, since the AI Deity level advantages are overwhelming for the vast majority of (Human) players. AI deficiencies = AI limitations must be taken advantage of at high player levels. These are not AI defects = AI (design) bugs = exploits if acted upon. True AI defects are rare and ones that allow an unfair advantage might be considered exploits by the HOF and possibly banned in HOF play.

 

[RickWJ66]

Thanks for the feedback. I see it as a "defect" from the standpoint that there is no way a human player would allow another human player to win the game under such circumstances, with such a vastly inferior military, no matter how good a diplomat (as your discussion recognizes). My terminology may be too simplistic, however . . .

 

[Dracandross]

Thanks for the feedback. I see it as a "defect" from the standpoint that there is no way a human player would allow another human player to win the game under such circumstances, with such a vastly inferior military, no matter how good a diplomat

Well again I think that it is really good that relations matter even when considering can you win. As your sole aim is to win its agreeable that if someone is going to win you have to nail him.

But for AI it's very good that everything is not aimed for winning. Because if AI would then they should start warring alot before. Think about AI going on axerush on deity with 2 cities and some workers. Why they should aim for spacerace only but not on conquest/domination. If ai declares war he doesnt seek to win by conquest but to get more cities to win by space. That would be preventive attack as you could win later anyway, easier to dispatch you now.

Your example for cultural win is specially bad as if your culture is so dominant they just get caught under it and your culture is the one that they bring to space...

-Dracandross

 

[jesusin]

1325AD. Saladin, corn,gold, 3FP start. 6 cities, 4 religions, 8 cathedrals, 14GP. Disastrous game. It wasn't a sub1000AD map, but was enought to win the gauntlet, maybe.

Founded hindu, then Pottery, Writing. CoL before Alphabet was a terrible mistake, because I failed founding Confu by 1 turn. 1 religion spread, though. Tehn Alpha, GS was to be used on Academy (great capital) but Philo and founding Tao seemed more important. Then I refrained from second GS to give it a try. Fourth religion spread in 800BC, too good, I popped borders in CS and then slavery+OR for lots of missionaries till NE done in 335BC.

Payed Philo Monopoly to restore peace between them. Stupid mistake made me spent 1 turn researching Lite. I wanted to reload!!!! Then I lost liberalism by 1 turn, in 230AD, while doing 250bpt. I should have had another GS lighbulbed.

Trading for the rest of Liber and for banking at the worst possible time, I had to wait a lot to change civics. Trading for Music unexpectedly (no NotreDame yet) was terrible, because I had traded Marble and Copper away. I had not stone, needed for 1 of the religions.

5 turns in slavery+OR to whip cathedrals. Hermitage 530AD, stopped researching 560AD!!!! That's how much losing Liberalism affected my game.

GPFarm only 5 artists most of the game. Only 10 cottages built at 1AD, too few. More than 25 cottages at the end, when no hammers were needed.

1worker per city almost the whole game. It was enough for the cities, but no for the trees. Most of them were chopped on monasteries at the very end of the game, they should have been chopped on cathedrals.

Final mistake at the end, revolted from FM to Merc too late, had to wait 3 turns for my last GA to pop.

I have lost my opportunity. I don't think I'll have time for another try if the deadline is not extended. Good luck everybody.
Please, do me a favour. GET THAT sub1000AD WIN FOR ME!!!!!!!! I will keep on chasing for it after the Gauntlet is over, of course.

 

[Harbourboy]

I still don't think that 1000 AD victory is possible.

 

[Sun Tzu Wu]

Thanks for the feedback. I see it as a "defect" from the standpoint that there is no way a human player would allow another human player to win the game under such circumstances, with such a vastly inferior military, no matter how good a diplomat (as your discussion recognizes). My terminology may be too simplistic, however . . .

In G-Major-14, we are playing at deity level whereas all the AI players are at Noble level. If a human player took over an AI civilization (HOF violation of course), he would be playing at Noble level and should easily defeat the human player at Deity level.

The AI is so limited/deficient compared to a human player, comparing the AI to a human player is a futile exercise. However, the AI is a formidable adversary for a Human player at Deity level.

I think the Civilization 4 and Warlords AIs try to mirror the real world to some degree. If all civilizations of the past were warmongers and none were builders, the human race would still be in the stone age, if it existed at all. In my opinion, Civ4 and Warlords AI design has the proper balance of warmonger AIs and builder AIs. The diplomatic system seems to be well balanced and adaptive too ...

In a recent G-Major-14 game, I got way behind in research. Like most third world countries, I became the recipient of many valuable gifts from world powers. Two civs gave me two technologies each. Three other civs gave me one technology each. The last technology gift, Divine Right, came just before I started researching Nationalism, helping to reduce its cost. Unfortunately, DR was the most useful gift for a Cultural win. I've, never received so many technology gifts in a game before, but I guess I've never been so far behind in research before either.

------

I apologize to everyone that used the term "AI defect" to mean an "AI limitation". I was being too pedantic. For the brave, there's more of the same that touches on the concepts of verification and validation of software systems with a trivial AI as an example ...

------

The AI is a (complicated) software system. Every software system starts out as a design, the design is implemented and finally tested. There are two important aspects of testing a software system, verification and validation.

Verification consists solely of verifying that the final software is consistent with the design. Any inconsistency is by definition a defect (also known as a bug).

The simplest AI design, "Do-Nothing", does nothing - the initial units never move and no city is ever built - for every turn, the AI ends its turn. It should be very easy to verify the implementation of such a simple design. It would be rare for such a trivial design to have a defect (bug), since it is trivially simple. It is almost certain that the verification of such a trivial design and AI software system would pass.

However, does this trivial design do what the end user wants? This is where validation of a design comes in.

Validation consists of determining whether and how well the design fits the application in the real world. The AI design should help create a game that is both enjoyable and challenging. Clearly, the "Do-Nothing" AI design fails the validation test immediately. There is nothing enjoyable or challenging in defeating an AI that never moves its initial units. Areas where the desired application differs from the design are called design limitations.

The "Do-Nothing" AI design has no defects (bugs), since the AI software implements the design perfectly, but it totally fails validation. It has a design limitation the size of the Pacific Ocean, the AI's initial units never do anything.

------

Sorry, I still like the term "AI defect" to mean "AI bug" in the verification space. I'll drop using term "AI deficiency" in favour of "AI limitation" in the validation space. I was wrong to use "deficiency" rather than "limitation", since "deficiency" is really just another word that means "defect".

I apologize for using confusing terminology.

------

Sun Tsu Wu

 

[Lexad]

Stupid AI

My second attamept (all quick) that went further than losing a Parthenon race. The situation - 4 AI and me racing for Liber, I switch my ubercapital to producing research and carve a vital turn off Liber - meanwhile Asoka gets Nationalism - I pick PPress and trade it and Liber for Nat and some sweeties. As I have marble I attempt a rush for Taj - for culture, GPP and GA - bringing my workers from the other side of empire. I constantly check the trade screen to see other AI getting N-sm.The situation is that next turn I'll finish my prechops and build Taj - yay!
Then Fred pops a GE. I check - he has no N-sm yet.
Then the next turn some idiot trades it to him for some cash!!! WTH, man! Can't you read the messages? Naturally, Fred overtakes me... Hate AI.

 

[Sun Tzu Wu]

In G-Major-14, I've seen the same AI complete The Apollo Program and one or more space ship parts in the same turn. How is this possible?

As I understand it, The Apollo Program can be completed on turn <x>, and space ship parts can be _started_ no earlier than turn <x>. Since, it takes a minimum of 1 turn to build anything, space ship parts can be completed no earlier than turn <x>+1.

I'm playing Warlords 2.13.001. (I upgraded in order to play G-Major-14). Maybe this is a bug in Warlords 2.13?

Has anyone else noticed this?

Sun Tsu Wu

 

[KMadCandy]

Sun Tzu Wu, that is really weird. i know there's something different with the AI logic, in that the governor will have them sometimes start a wonder on the turn they discover the tech, in a way that it won't for the human. i forget the details.

there's also a delay in messages sometimes due to the turn order, like you won't see they completed a wonder (and get your gold refund, if applicable) on turn <x> until turn <x>+1. maaaaaaybe that applies to projects too, and you're getting that message a year later than it was built, but the spaceship part message on the same year it happens? but parts can't be hurried by a GE or rushed except through chopping, so it doesn't seem likely that one or more would be complete even the turn after Apollo.

apparently the patch for vanilla doesn't match the patch notes (cossacks didn't get reduced from 50% vs. mounted to 25%). i'd not be surprised if "little" things like you and i noticed are being missed since most people are playing BtS.

good luck to you and everybody else! grats to those with wins already. Harbourboy, yours especially made me grin! if they make you a Civ leader they'd have to make a new trait for ya, tenacious maybe .

 

[FiveAces]

Probably not, but you can get very close. Here's why:

Latest attempt as Saladin - capital on ph, 2gold, 2f, 3fp. 2nd city GP farm (9 artists maxed out) - gems, 3f (1 overlap w cap) 3hills. 3rd city (not legendary) - ph, 3 squares from cap, marble, 3fp overlaps w capital, rest is desert/plains - it worked the cap's fp cottages, produced workers, settlers and missionaries, and never grew past size 2 (should have grown it to 4). 4th city - 1f, copper, grasslands and forests - also 9 artists maxed. 2 more cities much later built temples for nothing, hindsight would have gotten them up earlier when they could have done some good.

Founded budda, popped ag, mining, pottery, founded confu (no more rel spread), alpha. Then first mistake - traded CoL to Liz who popped philo 5 turns later, 2 turns before my GS, now used for academy. CS before 900BC, 1st to music, edu and NE by 90AD, 2 cathedrals and Lib at 220AD, picked Nat.

Then I utterly destroyed my chance at winning the gauntlet. Because I had marble, I decided to work the mines in my GP farm (gems+1town) and build herm there. Meanwhile capital builds the taj since nobody else has nat or is even researching Lib. So I delay FS and pacifism for the 15 turns it will take for taj, only to have Liz research nat through DR, pop a GE (go figure) and rush 2 turns before I finish. Researched pp for 7 of those, but what a loss of culture. Finished 1465, didn't bother to optimize the late GA's cause I was so frustrated.

Had I popped Tao, I would have built either herm or pagoda in cap, the other in the GP farm, delayed the switch to FS/pacifism only 3-5 turns to finish while researching pp, stopped and traded for the rest of it later, and overall finished much earlier. Without Tao, could have immediately switched to pacifism, built herm in cap while researching pp, then gone to FS.

Anyway, enough moaning. My point is the current best-published game has 2 rel, 1st to music, lib by 220AD, no marble and a GP farm with cows and 3fp. Map I had permitted the same early research timeline (actually I was ~5 turns ahead at CS), but had marble, copper, and 2 city sites to run 9 Artists each. I might give this one another try (not HOF acceptable, of course) after the gauntlet just to see how early is possible.

 

[WastinTime]

Yes, please play that map again. or maybe we should all download it and have a little competition.

 

[Sun Tzu Wu]

... there's also a delay in messages sometimes due to the turn order, like you won't see they completed a wonder (and get your gold refund, if applicable) on turn <x> until turn <x>+1. maaaaaaybe that applies to projects too, and you're getting that message a year later than it was built, but the spaceship part message on the same year it happens? ...

Next time, I'll try to remember to look at the log for the turn (year) The Apollo Project was built and whether or not the space ship parts built by the same Civ are actually built the same turn or one turn later.

... but parts can't be hurried by a GE or rushed except through chopping, so it doesn't seem likely that one or more would be complete even the turn after Apollo.

Yes, I also considered how an AI could build a space ship part in a single turn too. Maybe, hammer overflow from building The Apollo Program or other large build? Perhaps the AI is allowed to use Pop rushing (Slavery) or Gold rushing (Universal Suffrage), or even a GE? In any case, humans aren't allowed to use these on space ship parts, right?

I still think there may be an AI related bug here.

Thanks for your response, KMadCandy!

Anyone else care to comment? There must have been others playing 2.13 and seeing the AIs start their space race.

Sun Tsu Wu

 

[Lexad]

Yes, I also considered how an AI could build a space ship part in a single turn too. Maybe, hammer overflow from building The Apollo Program or other large build? Perhaps the AI is allowed to use Pop rushing (Slavery) or Gold rushing (Universal Suffrage), or even a GE? In any case, humans aren't allowed to use these on space ship parts, right?

No bugs here. It's Deity - therefore hammer costs for AI are like 40% cheaper - it takes few turns to build, they have great big cities with lots of imrovements and building and no health/happiness problems. And Aluminium, I expect. And maybe Ironworks.

 

[Sun Tzu Wu]

Frederick declared war on me on turn 113, AD 995, in my latest quick game for this Gauntlet. My Diplomacy score with Frederick was 0, so I was quite surprised. However, we did share a border, which provides profit to attack and shorter distance for maintenance. Also, one of his high culture (80% culture defense) cities had revolted once a while back due to 15K+ culture built up in one of my Culture cities by creating five Great Works there. I was trying to flip his city and another of a second bordering Civ, which did flip. Frederick appeared to counter by sending in his own GA which probably gave him the 80% Cultural Defense in this city. Another important factor was my very small Power, six Warriors I built and two Longbows from one city flipping. Our religions were different too, but Frederick doesn't care too much about religious differences, only -2.

Diplomatic score of 0, perhaps not enough to avoid war:

Where did I go wrong? I'm always a little nervous about Diplomatic scores of 0-2, especially when I have a tiny Military (so I can focus on Culture).

--- Some highlights of the Game of pending Doom ---

Other that Frederick declaring war, losing Liberalism and not getting a religion until AD 410, and a few minor misses (such as missing Theology by 5 turns & The Sistine Chapel via AI GE rush) the game went quite well:

Built The Pyramids on turn 56 (BC 790) and switched to Representation.

Used GE (turn 69) from The Pyramids to rush National Epic on turn 69 (BC 335).

Asoka gets to Liberalism on turn 80 (AD 50), but I'm not far behind at turn 86 (AD 230).

Got GAs on turns 74, 84, 88, 92, 100, and 108.

Used GE (turn 80), generated via The Pyramids pollution, to rush The Hermitage far later than expected due to losing the Liberalism race and the Nationalism prize.

--- A Great Start and four Modest GA Farms ---

Inland Sea with Temperate Climate and Low Sea Level plus six AIs:

Had a great start with 7 flood plains, gold, and corn in the Capital's cross.
The second city was built on turn 26 (BC 2440) with 4 flood plains & a food resource.
The third city was built on turn 33 (BC 2020) with 3 flood plains & a food resource.
All three cities on the same river, so Wheel wasn't really needed much early on.

Modest GA Farms:
Capital - 6 Artists max + The Pyramids pollution + National Epic + Hermitage
Culture 2 - 4-5 Artists max
Culture 3 - 4-5 Artists max
Ancillary GA Farm with three food resources (4f, 6f, 6f) - just getting started with 4 Artists - would have max'ed out at 6.

--- Great Experience & Good Bye for now ---

At least it was a great learning experience ...

I don't think I'll have time for another try at this Gauntlet.

Sun Tsu Wu

 

[FiveAces]

Yes, please play that map again. or maybe we should all download it and have a little competition.

I'll post it tomorrow. I'm pretty sure someone can go sub-1100 on it.

 

[Sun Tzu Wu]

No bugs here. It's Deity - therefore hammer costs for AI are like 40% cheaper - it takes few turns to build, they have great big cities with lots of imrovements and building and no health/happiness problems. And Aluminium, I expect. And maybe Ironworks.

OK, but the AI was building a small space ship part in a single turn. The same turn they completed The Apollo Program as I clearly recall. The messages to that effect scrolled onto the screen right after each other at the beginning of a single turn. If an AI really can build The Apollo Program and a space ship part on the _same_ turn, that has to be a bug, since The Apollo Program must be completed before space ship parts can be _started_.

The Apollo Program can be completed and a space ship part can started on the same turn, but not also completed that turn. Nothing can be built in 0 turns!

Has no one else seen this apparent contradiction?

Thanks for your input Lexad! Maybe the AI can build a small space ship part in a single turn, but I'm still have some doubts about it. How many hammers does the AI need to build and how many hammers can an AI city output?

Building a space ship part in a single turn is possible, but zero turns ...

I should dig into my old save files to review what really happened.

Sun Tsu Wu