Civ4 AI Survivor Season 3: Game Eight Writeup

Game Eight was the final game of the opening round, featuring the last remaining AI leaders who hadn't yet appeared on the main stage in Season Three. We ended up with a surprisingly warlike collection of personalities for this game, a recipe for a chaotic and unpredictable result. This was due in large part to a trio of leaders: Boudica, Napoleon, and Shaka. All three of them heavily emphasize building units and researching military-orieneted technologies. There are some minor differences between them - Boudica usually founds a religion and won't declare war at "Pleased" relations, Napoleon endlessly demands tribute and civic changes, Shaka almost never builds wonders - but their broad gameplan usually plays out the same way. They will attack, attack, and attack some more until they either defeat their opponents or they themselves are defeated. This surely spelled trouble for Frederick and Hatshepsut, two peaceful leaders that favor Cultural victories and seemed certain to be the target of invasions. It didn't help either that Freddie and Hatty had poor records in past years of Civ4 AI Survivor. We also had Pacal in this game, one of the rare leaders that combines a low peace weight score with outstanding economic traits. If Pacal could claim a reasonable share of the land and stay out of too much warring, he would be the favorite to win by spaceship. And finally we had De Gaulle, the leader who was ranked dead last in the AI Survivor competition over the first two seasons. De Gaulle had been the first eliminated in both of his opening round games, somehow finishing in 50th place and 52nd place in those two seasons... in a competition with only 52 total leaders. Needless to say, the community wasn't expecting very much out of De Gaulle.

The map for Game Eight had the leaders distributed in a ring around the outer edge of the continent. It felt like an unusually small amount of land for a game with seven AI leaders, and this setup prompted some of the Livestream viewers to refer to this as the "Thunderdome" in the old Mad Max reference. Given the aggressive personalities in this game, it felt appropriate. This was a horrible setup for both Frederick and Hatshepsut, each of whom had violent neighbors with a sharply different peace weight score flanking them. It felt inevitable that one of the two of them would be the first to die, and the only hard part was figuring out which one would be the first to go. I guessed that it would be Frederick given his more cramped position and the nearby presence of both Napoleon and Shaka. That said, Hatshepsut would very likely found her own religion in this game, and that could get her into a world of trouble. And even though De Gaulle was one of the top picks to be eliminated first, this was a good map setup for him. De Gaulle would be predisposed to have good diplomatic relations with Boudica and he could potentially take advantage of an early exit from Frederick. Even if he was unlikely to be the winner, De Gaule should be able to avoid being the first one eliminated, and that alone would be a massive improvement over past seasons.

The race for the early religions played out in more or less expected fashion. Pacal went for an immediate religion while Boudica did not, the two of them being the only leaders that started with Mysticism tech, and the Mayans founded Christianity on Turn 11. Hatshepsut had opened with Mysticism into Polytheism, but she was beaten out to the religion by Pacal, and when Hatty belatedly switched over to Meditation tech, she lost that race as well to Boudica, with the Celts founding a random Taoism as their religion. Not one to be deterred by these failures, Hatshepsut would go on to research Masonry and Monotheism, eventually founding Hinduism on Turn 37. This meant that all three of the game's major religions ended up being located in the southern part of the continent. It also meant that both of Hatshepsut's neighbors had a different religion than the Egyptians, and that was a sure recipe for trouble. Furthermore, Hatty was forced to invest so many beakers into the religious techs that it slowed down her development. She had a copper resource at her capital that didn't get connected until almost Turn 50 because she was so slow at researching Bronze Working. This was not the start that Egypt needed.

In terms of expansion, the AI leaders did some strange things with their initial settlers. Pacal headed down into the southern tunda, where he planted a solid city with double food resources, but still ended up largely wasting the Holy City culture on a non-disputed part of the map. Boudica did something similar, grabbing a jungle-choked location on the eastern coast where her own Holy City culture didn't do much to grab contested territory. At least Hatshepsut's own second city was in a nice location, sent northwest towards Pacal. Shaka settled due south while Napoleon picked a puzzling location to the north of his capital up near the tundra. That was surely a waste on his part. Meanwhile, Frederick and De Gaulle settled towards one another and wound up with cities only three tiles apart, a settling pattern that seemed sure to lead to conflict. A little bit later, the AIs started to get their first self-built settlers out onto the map, and Hatshepsut headed towards Boudica. However, we suddenly spotted the city ruins graphic popping up on the map:

Are you kidding me?! She lost her new city to the barbarians. That's the only explanation for the appearance of the city ruins on the map, and sure enough when we watched the replay after the game was over, there had been an Egyptian city for half of an interturn that was immediately autorazed. It's truly amazing how the Deity AI can keep losing settlers and even cities to the barbarians, given the enormous bonuses that they get against them. Here in Season Three, an AI was losing a settler to a barbarian about every other game. Unreal. This would set Hatshepsut back even further, although her score remained among the leaders due to Creative culture and the expanded borders from an early Holy City.

Elsewhere, one of the biggest early developments was the construction of Stonehenge by De Gaulle on Turn 35. His trait combination of Charismatic/Industrious makes De Gaulle arguably the best leader in the game to build Stonehenge, getting the free culture for border pops in addition to the bonus happiness for being Charismatic with monuments. This would allow De Gaulle's cities to grow noticeably larger than most of the other AIs due to his bonus happiness. He had even put the wonder in his border city of Lyons, and Stonehenge's culture allowed De Gaulle to dominate the early border with Frederick's Germany. De Gaulle wasn't expanding very quickly but this was a big help to his civ. The other leaders continued expanding in their own right, with Pacal and Hatshepsut butting heads against one another along their respective border. Both of their cities in that part of the map had huge cultural footprints, especially Hatty's second city of Memphis, which soon had the Hindu Holy City culture, the Hindu shrine, the Great Wall, and the Oracle located inside its city boundaries. That would be a real pain for Pacal on his eastern border.

Here's where we stood after 51 turns:

Why 51 turns and not 50 turns? We had three different cities established on Turn 51 and it seemed more appropriate to include them in the overview screenshot. Pacal, Boudica, and Shaka were tied for the lead with five cities apiece. Pacal was making good inroads towards the center of the map, and Shaka still had room to expand both east towards the center and west in that backline peninsula where a barbarian city had sprang up. Boudica was taking advantage of the weakness of Hatshepsut in this game, with Camulodunum planted atop the ruins of the razed Egyptian city. Hatty obviously would have been in a much better position had she not lost that spot. Her expansion had therefore gone entirely towards Pacal, and that was cramping the Mayans a little bit on territory. In the north, Frederick was doing surprisingly well at expansion and had four cities, a settler out on the map, and two more under production. He looked like he would get seven or eight cities before the land started to run out, much better than I had expected. If his diplomatic situation wasn't so atrocious in this game he would have been one of the favorites at this point. Frederick was off to a good start in part because the two French leaders had been stagnating on expansion. Napoleon hadn't claimed very much land with his settlements, instead picking spots right next to his capital, and he had extra settlers sitting around not moving for long stretches of time. I have no idea what he was doing, but it certainly wasn't an effective landgrab. As for De Gaulle, he had invested a lot of production into Stonehenge and ended up getting locked out of the center of the map. He would be occupied with settling the territory to the north and east of Paris for the immediate future.

Shortly after taking this picture, we had our first war declaration of the game as Boudica attacked Hatshepsut:

That wasn't exactly shocking give the personalities involved, the religious tensions, and the massive difference in peace weight scores between the two female AI leaders. I went ahead and pasted in a screenshot of Hatshepsut's diplomatic status in the corner of the above image, and it helps to display why she was in so much trouble in this game. Seemingly everyone hated the Egyptians in this world; Hatshepsut was the "worst enemy" of Boudica, De Gaulle, Napoleon, Pacal, and Shaka. Yikes! This was a game of mostly low peace weight AI leaders, and everyone other than Frederick was predisposed to dislike Hatshepsut as a result. Regarding the actual war itself, we had noted when Hatty founded the city of Alexandria that it was a very defensive location, on top of a hill and with all flatground tiles surrounding the eastern approaches. There weren't even any forest or jungle tiles to provide defensive cover for invading armies; this almost looked like the kind of place where a human player would settle on a contested border in a Multiplayer game. As a result, Boudica's forces were slaughtered when they tried to capture the city. Memphis off in the west had similarly stout defensives, also located on a hill with massive culture to generate protection. Hatshepsut would rely on these locations as a bulwark against foreign interlopers for many long centuries on end.

In terms of religious spread, Frederick wound up adopting Hatshepsut's Hindu religion. This made the two of them mutually unpopular with pretty much everyone else, closing off one of the few ways possible for the two of them to make friends. De Gaulle picked up Boudica's religion of Taoism and the two of them would become fast friends. Napoleon adopted Christianity, the religion of Pacal, and Shaka started out by converting to Hinduism for a brief period before he also was converted to the Mayan religion of Christianity. In other words, we largely had a Taoist east, a Hindu center, and a Christian west on this map. It was almost like the actual religious pattern of Eurasia, minus the presence of Islam and Buddhism (and dozens of other religions). These religious differences combined with peace weight disparity were likely responsible for the game's next war:

Napoleon declared war on Frederick and immediately captured one of the German cities, which was autorazed since it hadn't reached size 2 as yet. That was a bit unfortunate for Napoleon, as he would have benefitted a lot from capturing a city intact. He only had five total cities and currently sat in last place on the scoreboard, with Frederick actually having a higher power rating. Napoleon was going to need some help if he expected to defeat Frederick in this war, and fortunately for him the diplomatic situation made it likely that Shaka or De Gaulle or Pacal would be willing to intervene on his side.

For now, the biggest question remained whether Frederick or Hatshepsut would be the first to suffer elimination. Given that Hatty was the "worst enemy" of nearly everyone else in the field, she certainly seemed to be in the driver's seat. This was further confirmed when Pacal joined Boudica and declared war on the Egyptians on Turn 79. Now Hatshepsut was facing opponents on both sides, and we all know how well the AI doesn't do when put in those situations. We were also in the process of watching Shaka walk a large stack around in the center of the map, and it was obvious that he was about to declare war on someone. At first it looked like Frederick, but then Shaka diverted further over to the east. Finally his stack was one tile outside the borders of both Hatshepsut and Boudica. He could theoretically be targeting either of them, since Shaka was still practicing Hatty's Hindu religion at the moment and Boudia followed rival Taoism. But no, it was indeed Hatshepsut who was the target:

Peace weight difference trumped religious brotherhood in this particular instance. Now Hatty was at war with three different opponents at the same time, and she was certain to be the first AI removed from the game. Or at least so we thought. Instead, a funny thing happened on the road to Hatty's certain early elimination. Her highly defensive border cities with massive cultural strength, combined with a lack of Construction tech to field catapults among her opponents, turned out to be sufficient to repel these attacks. Pacal was the first to cry uncle, suing for peace on Turn 90. He had tried repeatedly to attack at Memphis with absolutely no success. A city on a hill with 60% cultural defenses was essentially invincible against Mayan invaders who lacked siege units. Shaka would attack at the same location with no more success. Over the next two dozen turns, waves of Zulu units would crash against the hill fortress of Memphis and be dashed to pieces. Boudica had been attacking at Alexandria over and over again, never with the forces necessary to prevail. She had even tried to slip past the city once or twice and go after the Egyptian capital of Thebes, but that was no better, not with 60% cultural defenses in the heart of Hatty's territory. Over time, Hatshepsut started to get some highly promoted units as a result of winning so many battles on the defensive, and that only made things worse for the attackers. Finally they both sued for peace, with Shaka tapping out on Turn 109 and Boudica following suit on Turn 111. Incredibly, Hatshepsut had survived the 3 vs 1 scenario without losing a single city.

The same could not be said for poor Frederick up in the north. De Gaulle joined the war against Germany on Turn 92 and the combined armies of the two French leaders began picking him apart. De Gaulle had almost complete control of the border between himself and Frederick thanks to all that culture from Stonehenge, and this made it easy to start capturing cities from a German civ that was exhausted from long warring with Napoleon. Frankfurt was the first city to fall, followed by Hamburg. These losses then allowed Napoleon - who did have Construction tech and catapults by this time - to put the German capital of Berlin under siege. That city fell to western France on Turn 107:

Frederick was now neatly eviscerated into two pieces with no connection between the two halves. With the loss of his capital, he couldn't even trade resources between his remaining cities, and was left building nothing but archers in several of them. Despite all of the attacks launched against Hatshepsut throughout the game, it was now obvious that Frederick would be the first one eliminated. Shaka even decided to pile onto the poor Germans with another war declaration on Turn 112. The Zulus wouldn't manage to take any cities, but their presence wasn't helping. Instead it was De Gaulle who was the biggest winner here. He had spent the early game developing his cities while everyone else was fighting each other, landing both Stonehenge and the Pyramids to further his internal growth. When De Gaulle joined the war against Frederick, he was able to hit at the perfect moment and begin taking cities left and right. Out of the eight German cities that had existed prior to the start of the war, De Gaulle would end up taking five of them. He was now the overall game leader in score, much to everyone's surprise. Napoleon took the other three cities, although the first one was autorazed as we saw earlier, leaving him with only two pickups in this conflict. At least Napoleon was able to take credit for the final killing blow:

Napoleon had now picked up a kill in all three of the games that he had appeared in, despite suffering multiple eliminations and never advancing past the opening round. He appears to be as much of a wrecking ball in these games as Genghis Khen Temujin or Montezuma. My expectation for this game had been a quick elimination for Frederick that would allow Napoleon to snowball into one of the top spots. Instead, Frederick had indeed fallen but it was mostly at the hands of the other French leader. While Pacal remained the clear tech leader in the game, De Gaulle's advantages in land and population would allow him to start closing the gap over the following turns.

Speaking of Pacal, he had gone back to war with Hatshepsut again for a brief period before signing another treaty. This had been a bit wasteful for the Mayan leader, as even with catapults this time he was unable to break through the Egyptian defenses at Memphis. Pacal would have to tech up to better units and try again. The continued survival of Hatshepsut was throwing off a lot of expectations for this game, with most of the picking contest having either Pacal or Boudica to emerge as the dominant AI. Pacal's troubles were increased further on Turn 133 as Boudica launched a new attack against him. That was a little bit surprising given their distance from one another and lack of previous conflict, but apparently the fact that Pacal was practicing a different religion was enough for the Celts to declare war. The main effect of this war was to keep Pacal and Boudica out of the latest dogpile that was forming agaisnt Hatshepsut. First Shaka declared war on Turn 144, followed by Napoleon four turns later:

Napoleon was the first leader to capture a city from Hatshepsut, and this was an indication that the tables were finally beginning to turn against her. The city of Xukpi was a Mayan city originally which had culture-flipped to Hatshepsut due to the immense cultural pressure coming out of Memphis. Unfortunately for Napoleon, it was still swamped by the surrounding borders of Pacal and Hatshepsut, and he would eventually liberate it back to the Mayans. Napoleon would conquer a number of cities in this game that he couldn't retain due to stifling cultural pressure from other civs. In a similar outcome, Boudica actually managed to capture the city of Uxmal (visible on the left side of that screenshot) from Pacal on Turn 165. Then for unknown reasons she gave the city back to Pacal when the two of them signed peace shortly thereafter. Pacal had therefore lost two cities in his northeast and then gotten then both back again without lifting a finger militarily. What a lucky son of a gun.

Nevertheless, the main result of this little Pacal vs Boudica spat was to keep them out of the ongoing destruction of Egypt. Yes, amazingly the two leaders that directly bordered Hatshepsut were somehow unable to capture any cities from her, and gained nothing as her empire fell apart. Napoleon and Shaka were the ones doing the heavy lifting, with Shaka finally taking the border fortress of Memphis followed by Napoleon capturing the Egyptian capital of Thebes. Then De Gaulle decided to join the party too:

And that sounded Hatty's death knell for certain. De Gaulle was extremely far ahead on the power bar graphs at this point, largely because he was mostly staying out of wars and keeping his army fully stocked rather than throwing away units in endless combat. Once again De Gaulle had picked an excellent moment to invade an opponent damaged by an ongoing conflict. His knights rolled over Alexandria, the sight of so many Celtic failures, without breaking a sweat and continued onwards to the southern coast. In less than ten turns the Egyptians were gone:

We genuinely felt sorry for Hatshepsut. She had fought magnificently in defense, attacked by five different leaders over the course of this game, but couldn't hold back the enemies at the gates forever. Watching the last series of wars was a bit like seeing a pack of rabid dogs tear Hatty apart. Short and brutal. The leader draw for this particular game had been terribly unfair to both Frederick and Hatshepsut. Perhaps it was karma for the way that Hatty had trolled the United Nations in the last season of AI Survivor. In any case, Egypt had been partitioned in roughly equal shares between De Gaulle, Napoleon, and Shaka. However, De Gaulle's culture was dominant by far over the two western leaders, and he would end up crushing most of their new possessions, especially the ones taken by Napoleon. As for Pacal and Boudica, they were locked out of the gains entirely due to their foolish earlier fighting. The biggest story now was whether Pacal would be able to outrace De Gaulle to a peaceful victory condition of some kind, either space or culture.

Pacal had built the Mausoleum earlier and picked up the Liberalism free tech prize. He could have used it on Nationalism and lock in a guaranteed Taj Mahal but instead opted to waste it on Divine Right in a true facepalm moment. This would wind up biting Pacal in the rear end, as De Gaulle managed to complete Taj Mahal two turns before the Mayans by virtue of conducting an 8 population (!) whip of his capital city. This was a swing of 20 turns of Golden Age research and production, which helped De Gaulle to continue closing the technology gap with the Mayans. International peace didn't last for very long in this game either, as Boudica began a war with Napoleon over the colonial holdings in former Egypt, followed by Shaka launching an invasion of Pacal:

Shaka's timing here was excellent, as he managed to hit Pacal while the Zulus had a lead in military tech despite Pacal's overwhelming dominance in research. Shaka had beelined to the tech Military Science and unlocked grenadiers, which was by far his most advanced tech on the tree. He wasn't remotely close to Rifling, lacking even Paper tech at the time. Pacal had meanwhile researched all the way up to Democracy and Scientific Method techs while resolutely refusing to research Mitiliary Tradition or Rifling. Thus it was Zulu grenadiers and knights against the medieval armies of the Mayans. Pacal easily could have defeated the Zulus if he would just research up to rifles, but he repeatedly refused to do so, instead going after Communism, then Physics, then hitting the bottom part of the tech tree. He actually went all the way up to Combustion tech (!!!) before he started to research Rifiling. All the while his core was falling apart, as Shaka's armies started with the border city of Quirigua and then moved on to Uxmal and Xukpi. Shaka even managed to capture the city of Chichen Itza and began a siege of the Mayan capital of Mutal itself. What were you doing Pacal?!

Over in the east, Napoleon initially lost his remaining colony in former Egypt to the first surprise blow of Boudica. After that, however, their fortunes slowly began to reverse. Boudica had been trapped on five cities for a very long time in this game, and her repeated invasions had been foiled again and again. First she had been stymied by Hatshepsut, and then her campaign in Mayan territory had ended up achieving nothing. Napoleon was ahead in tech, fielding rifles and cavs against Boudica's medieval army, and he had more cities to boot. Slowly the Celts began to be pushed back, with their border cities falling into French hands. Then there was a key diplomatic shift: De Gaulle swapped from Taoism into Buddhism to take advantage of the Holy City he had captured from Hatshepsut. Without the shared faith bonus of Taoism, De Gaulle's relations with Boudica dropped from "Pleased" (where De Gaulle will never declare war) down to "Cautious". It wasn't long before eastern France decided to enter the fray:

De Gaulle had factories and infantry on hand from Assembly Line tech, plus he was in the process of researching Industrialism and unlocking his own tanks. Boudica didn't even having Rifling yet, which should give you an indication of how one-sided the following conflict turned out. Tanks demolish rifles without breaking a sweat; what they do to longbows and muskets is better left unsaid. The combined French forces of De Gaulle and Napoleon now routed the remaining Celtic territories, taking one city after another in rapid fashion. Napoleon was the one to take the Celtic capital of Bibracte, although again his new possessions would be crushed by De Gaulle's culture and wind up essentially useless following the war's end. As we had seen with Hatshepsut earlier, it took only ten turns from De Gaulle's entry into the conflict until it concluded with another elimination:

With Boudica's removal from the game, we now started looking in earnest at endgame scenarios. De Gaulle seemed certain to win at this point, unquestioningly the Runaway AI in this particular world. He was double anyone else's score and had seized control of the tech lead from Pacal due to the ongoing pain that Shaka was inflicting on the Mayans. De Gaulle could have easily rolled on to a Domination victory from this point, however he was "Pleased" or better with all of the remaining AIs and wouldn't declare war on them. That ruled out the military victory option, as De Gaulle was still about 10 percent short of the land requirement for Domination even with most of Boudica's territory under his control. He wasn't popular enough with the other AI leaders to win the Diplomatic victory in the United Nations, and despite building a lot of wonders, De Gaulle didn't have three cities even remotely close to the Cultural victory condition. The AI usually requires a lot of religions to pull off a Cultural win (via building a lot of cathedrals) and De Gaulle hadn't done much with religion in this game. It looked like it was going to be another Spaceship win in this game barring something truly unforseen happening.

Since De Gaulle was more or less locked into first place, the most interesting question was which leader would manage to take second and punch a ticket into the playoffs. Napoleon and Shaka were very close in score, and the two of them were jockeying for that runner up position over the following turns. Pacal had been damaged so badly that the main question at this point was whether he would survive. He nearly lost his capital of Mutal to Shaka before finally stabilizing on a combination of infantry and machine guns. Pacal had pretty much jumped straight from longbows to infantry as a result of delaying Rifling tech for so long, but he did manage to get just enough infantry in place to save his capital, and likely his game. This allowed Pacal to get a peace treaty with Shaka on Turn 262:

What a disaster this had been for Pacal. He should have been the unquestioned choice for second place, and perhaps even would have won via space or culture. Instead, Shaka had dragged him down to the bottom of the pile despite the huge edge of Mayan technology. Next time research Rifling tech, man! And Pacal's woes weren't over yet as Napoleon declared war on him on Turn 277. Pacal was still so far ahead in tech compared to Napoleon that he was fighting with tanks against Napoleon's rifles and cavs, and that allowed him to hold off this latest assault without losing any cities, even though Napoleon had an edge in production capacity. The ability of tanks to blitz and attack multiple times per turn makes such a huge difference against outdated units. Napoleon did sign peace after about 20 turns of indeterminate warring. But then Shaka came back again for more, invading Pacal yet again on Turn 301. These wars had real meaning because Shaka was less than 200 points ahead of Napoleon in score. If either of them could take a city away from Pacal, it might be enough to swing the final results. Then Napoleon declared war on Shaka on Turn 313, and we had a truly interesting situation on our hands. Even one city capture from Napoleon might be enough to leapfrog him past Shaka into second place.

It was not to be, however. De Gaulle had been building all the wonders and researching up to the end of the tech tree while the other AI leaders in this game squabbled amonst themselves. His score lead had ballooned while theirs had stagnated, until he reached a point where his score was as high or higher than everyone else combined. He launched his spaceship on Turn 305 and put an end to the game when it arrived ten turns later on Turn 315:

Game, set, and match to De Gaulle. Shaka was able to hold onto his second place position by a little over 200 points against Napoleon. I think that Napoleon would have taken second place if he'd been able to retain his conquests in former Egypt and former Celtica. Instead, all of those cities had been liberated or culture-flipped over to De Gaulle. It was a tough break for Napoleon, who had been militarily successful but unable to win the peace after his wars came to an end. As for Pacal, he did manage to survive and will now join Napoleon in the gigantic free-for-all of the Wildcard game. But De Gaulle and Shaka took the top two spots - seriously, what the heck. I don't think anyone was picking those two before the game began!

This game showed the importance of having decisive wars as opposed to merely fighting wars in general. Warfare tends to be inefficient in Civ4. Long stalemated wars are a surefire recipe for disaster, both in Civ4 AI Survivor and also in our Civ4 Multiplayer games between humans. The goal is always to have quick and decisive wars that gain territory with a minimum of time and units invested. De Gaulle pulled that off perfectly in this game, fighting only three wars in total but all of them short and extremely impactful. De Gaulle claimed the lion's share of the spoils against Frederick to establish a strong position, and then invested a grand total of only 19 turns in crushing Hatshepsut and Boudica to turn his strong position into a runaway position. It was one of the most economical uses of force that we've ever seen from an AI leader in these competitions. While I realize that this was mostly the product of chance and not any grand strategy on the part of De Gaulle's AI programming, this was still a masterful performance. It could have been a human player up there in the northeast picking and choosing the most surgical moments to strike.

Next up will be the Wilcard game, always one of the most entertaining aspects of AI Survivor for the way that it breaks so many of our normal rules. We'll see what this year serves up in a huge field of 11 total leaders. Thanks again for reading, and stay tuned for the rest of the competition.