AI tournament, season 2

[Ray Patterson]

1535: Boudica capitulates to Justinian. This is very good news for Pacal's domi chances, as Justinian now cannot surrender until Boudica breaks free, and since she only has one island city that's not going to happen quickly

1555: Ragnar DOWs Gandhi

1565: Justinian goes WHE

The world AD 1610


Ragnar has quite possibly the worsed shaped empire I've ever seen. Note Cologne, Hippo and Gergovia, and how his "homeland" has been cut in two with the capture of just one city. Because obviously Gandhi is winning this war Ragnar started.

GNP lead: Pacal, 2nd: Gandhi
Prod lead: Pacal/Gandhi (virtual tie)
Crop lead: etc



Pacal is once again a sleeping giant. If he revassalizes Ragnar it's all over



Gandhi's culture is going nicely but he keeps being interrupted by war or threat of war

 

[Ray Patterson]

1615: Gandhi makes peace with Ragnar

Broadway: Pacal

1645: Justinian DOWs Gandhi. Is he going to do better than Ragnar? Stats say doubtful

1706: Delhi goes Legendary. Nice, but Gandhi isn't gearing for culture atm. He's gearing to crush Justinian





And building Apollo Project. Meanwhile Pacal is mainly just being awesome and extremely noncommittal.

 

[Ray Patterson]

Eiffel Tower: Pacal. Any chance Pacal could steal back the culture lead?

1712: Ragnar goes WHE

1714: Ragnar completes Apollo Program. Yeah, he's first

1722: Gandhi's AP

1724: Boudica breaks free off Justinian. Justinian and Boudica capitulate to Gandhi!

Rock n Roll: Pacal

Cristo Redentor: Pacal

1744: Pacal's AP

Three Gorges Dam: Pacal

1760: Pacal's SS Docking Bay.

1778: Gandhi finally goes back to 100% culture. Shangian still needs 57 turns

1784: Mutal goes L

1790: Gandhi builds UN. Pacal completes SS cockpit

Hollywood: Gandhi. Unfortunately he builds it in Delhi





Barring Pacal going for some last minute all out war against Gandhi, it's cultural victory for the Indian.

 

[Ray Patterson]

1826: Gandhi and Pacal are WHE. Each other or Ragnar?

1832: Boudica breaks free off Gandhi

1848: Bombay goes Legendary. 10 turns left for Shangian. And as if on cue:

Pacal DOWs Gandhi

1852: Hannibal breaks free due to cultural pressure



Pacal is rolling over Justinian's remaining cities rather fast, and from that direction he'd get to Gandhi too.

But not in time, because 1860 Gandhi wins a cultural victory.

 

[Ray Patterson]

This was a lot of fun and has brought me back into the game in a really cool way. Thanks for all the comments and questions, they really made it a lot more fun to not just start this but to actually finish it!

I think it would be nice if someone else would host another round sometime if you feel like it, with different parameters. One thing I'd consider is playing the same leaders against each other a few times on the same map, but with shuffled starting positions. But read through this thread and you'll see plenty of other suggestions as well.

Edit: lol, I just read back to see Gandhi was the defending champion, that time also cultural victory but as a vassal. And one other leader was a returning finalist: Justinian.

 

[RedKi-rr]

Ray, thank you! For making this tournament and for the nice reports!

Gandhi. He is so beautiful. However, if Justinian would win then it would three wins in row...I wanted that one of these two guys win)

 

[Timo33]

Thanks for all of this Ray !

Before this tournament started I never would have thought that Gandhi and Pacal would be the two top leaders :O

 

[enKage]

As it started I supposed Pacal and Justinian would be a top trier, but I would rather expect Asoka here instead of Gandhi. I am a little disappointed by Pericles, Kublai, Isa, Rome (in general) and Wang. Too pity also Vicky and Gilgamesh had no more chances.


This games showed again the only drawback of Civ4: lack of any kind of mechanism recognizing current victory condition and enabling flexible changes to choosen path of development.Hammi, Pericles and in final game Pacal proved it once more.

But it proved also, that peaceful leader is stronger in hands of AI. if there were more than 2-3 warmongers on a map, they coukd not dominate the map.

The strongest AI's in such a game is not always the same as in games with a human brain. In our games we often choose Gandhi as one of the weakest AIs, but we do it to have an early weakling who builds shrine for us. So in human games Gandhi simply very rarely has time to prove his power.
But even if I recall some games, there is often him, great, powerful, technologically advanced and amaasing huge army after Assembly Line - Great Pacal. If not in jungle or close enough to take his shrine, he is a real monster.

Zara ocuures to be not so powerful as it seems. In my opinion the reason we often raise him above other leaders is his hidden diplo bonus toward human players. Why to bother and attack early a leader who: is already pleased, does not have mids or great abusehouse or holy city, is not a mass unit spammer and on the top can be easily bribed against nearly anyone? So we often let him develop on his own and grow powerful when we take care of more serious threats or splash weaklings.

 

[elitetroops]

Thank you Ray for doing this! I haven't participated actively, but been following from the start. Very interesting read! I also hope to see some more of these in the future. Playing a few rounds to negate the effect of starting position would indeed be more fair, but I suppose it would take a long time. If you play until the same leader has won twice, you would probably end up with at least 3 times as many games.

Another interesting way to set this up would be to have all civs with the same trait face each other in the first round (inspired by all the "best xxx leader"-threads). It would probably not be the best way to find the best AI leader, but it would be fun and probably lead to some interesting situations, and it would give all leaders 2 shots to reach the finals. Would probably require large maps though.

Now all that remains is to take the civs who performed the worst in this tournament and pitch them against each other on settler difficulty.

 

[Aje]

Thank you Ray

I really thought one of the warmongers would take chargé of the game, but I guess that they balanced eachother out. Things would have looked à lot different with à succesful first war for one of them. That as normally what "we" the human player try to do.

Pacal the monster Better to take him out early, or even Better, to play him.

 

[pomthom]

Great thread Ray, it was fun.

Yeah figured Pacal was going to be too slow versus Gandhi's position

 

[Izuul]

Pretty interesting.

I would still welcome Pacal and Gandhi in my Deity games, hopefully on the same continent with me. Pacal is a great techer if left alone, and can actually assemble a formidable military once he gets a tech lead, but he's pretty easy to take out early. Ghandhi is even easier. The only thing that really makes Gandhi dangerous is that he's by far the best AI when it comes to pursuing a victory condition.

 

[nfw]

Interesting to see the builders do well. Like others mentioned I always take out Pacal and Gandhi first if they are near by since other wise they'd culture pressed the heck out of me, on top of that they build a lot of wonders/found religions.

I wonder what happens if you replay the game but switch the starting positions.

 

[Thrar]

Thanks, this was a very good read!

I was skeptical at first about what could possibly come out of this sort of format with that much randomness. But in the end, at least some of the leader personalities did manage to show very well what they were capable of. And those writeups were very entertaining, too!

 

[Culture Bomb]

It would be good to see this done with a mapscript like Hub or Wheel, where all the civs get a certain amount of guaranteed land and cannot be quickly boxed in, yet are not isolated.

I guess this would also reduce the likelihood of early wars as it would take longer before there were close borders between civs, so would probably favour peaceful civs over warmongers.

 

[Ray Patterson]

OMG I had never seen this hub format, I did try the mirror map on someone's suggestion, but that one is all land, would be very nasty on the peacemongers. Wheel is interesting, though I think it would really disfavor any kind of war strategy. I'll actually run a quick game, no mid reports, with 2 extreme warmongers, 2 peacemongers and 2 empire builders to see what happens.

 

[kossin]

Nice job!

If anyone's game for a third round, I think it would be more representative to use a points system and run more games for each AI before going to playoffs. Maybe more than one person could contribute as the number of games simulated would be quite high this way:
Make seeds from the current 2 seasons, divide AIs in 4 pools, play 2+ maps per pool so that each AI gets at least two showings.
First place gets 2 points, runner-up 1 point. (or 4, 2, 1 but evaluating runner ups can be difficult if a few AIs are close to victories)
Top 12 (or whatever) go to playoffs with best-of-three matches, then top 6 duke it out in a BO5.

That way, starting areas fairness and RNG stupidity (building Moai statues with 3 cities...) can be toned down a bit. Shuffle (or picking randomly) would be good, in the sense that it doesn't always favour the same kind of AIs.

 

[lymond]

but that one is all land, would be very nasty on the peacemongers.

I don't see that as a problem. An AI's inability to defend itself whether peacemonger or not should be indicative of their weakness. I think all land is more than fine and actually preferable.

I think Pangaea would be usually great if not of some extreme imbalance. Hub would be pretty good, but still would favor peacemongers I think, but with Deity it should be okay since they expand so fast.

 

[kossin]

Hmmm, using the cheatmode, this actually becomes very easy to simulate.

1. set cheatcode = chipotle in your .ini.
2. in game, reveal all tiles and AIs with Ctrl-Z.
Then, optionally,
3. open the python console by typing backtick (`) and presssing enter.
4. Type Game.AIPlay 100 (or some other number of turns) and press enter. This will destroy your civ.

 

[pomthom]

Big&Small with high sea level?