Civ4 AI Survivor Season 8: Game Six Alternate Histories

(Note : "A" column tracks the number of war declarations initiated by the AI, "D" the number of times the AI is declared upon, "F" the points for finish ranking, and "K" the number of kills.)

Myth: The answer to the question of whether the livestream game was typical was a resounding "no." There were a few games that looked overall similar, particularly games 1, 5, and 12, where Victoria's financial trait allowed her to out-tech everyone else in the field or Gandhi was left alone long enough to get going, but these were not the norm. The common conception was that the southern leaders would be much better off, and this was mostly confirmed in these replays: our top 3 leaders did turn out to be Stalin, Joao, and Ragnar. Roosevelt turned out to be the only northern leader who could keep pace.

Starting with pre-game favorite Stalin, his performance during the actual game was baffling & extremely disappointing, pausing expansion in order to attack probably the single worst person he could have chosen in Napoleon. This was not necessarily the most unlikely occurrence and happened multiple times throughout the replays, with them fighting early another 3 or 4 times, but it was far from a common result. What was NOT uncommon was him going absolutely insane and declaring war way too early, and that was due to the copper located on a river (for instant connection) at his capital:

He declared war on someone else before Turn 60-70 in 90% of the games, stopping expansion in order to fight one of his neighbors. The target of this aggression was exceedingly important, and he had 5 options he chose from. If it was Napoleon (high unit rating, copper) or Gandhi (cultural defenses), he was basically done for from the moment he started plotting. The actual game was a great example; attacking Napoleon sealed both of their fates. If the declaration was on Joao or Victoria, things could go any way, either ending in one of their deaths (Game 5), his own (Game 1), or a stalemate. However, if it was on Roosevelt, things went spectacularly well: Games 2 and 19 were the most glaring examples of this, seeing an American exit before Turn 100. This singular copper tile was responsible for Stalin either being an unstoppable runaway or an also-ran, and the games where he made the "correct" decision allowed for him to be our top leader on this map. However, his starting position was so incredibly strong in all other regards that if we just removed this copper and replaced it with an iron, he probably wins more games. The fact that he declared so early destroyed his chances in game after game and allowed for others to shine.

Joao was the other big change from the actual game. He was consistently much stronger in these alternates, usually founding 10+ cities peacefully (including two games where he founded the ENTIRE tundra, including the territory south of Ragnar's capital, for a whopping TWENTY peacetime city foundings) while everyone else on the map was locked into conflict. His peace weight of 6 (along with warmonger respect, a hidden diplo value that basically made him +1 with our 3 low peace weights) allowed him to ally with either side of the alignment: he found himself succeeding in scenarios where Stalin, Ragnar, and Napoleon emerged as the leaders, as well as when Roosevelt and Victoria triumphed. However, he fought Gandhi in 18/20 of the games, only being able to get along when they shared a religion (and relations inevitably dropped when Gandhi went Free Religion). The only other consistent war for Joao was with Napoleon, usually fighting the French during the late game. He and Stalin fought occasionally, especially in the late game, but they were often good allies when favorite civics, mutual wars, and religion were factored in. He also benefited from the bloodthirstiness of the low peace weights: all of them being Pleased with each other meant their aggression could just as easily go against each other instead of him.

Our final southern leader, Ragnar, had one path to victory: through Victoria. Unfortunately for him, she was almost always ahead in tech, and he had to fight a battle against someone an era ahead in unit quality. She often had knights and longbows by Turn 120, and usually didn't avoid Rifling (due to redcoats incentivizing her to go for it). He was able to take her out usually only with the help of Stalin or through sheer weight of numbers, but she just as easily held her entire territory and all their war did was drag them both down.

The north was just as volatile as expected: there wasn't much room for these leaders, and trends we saw in the actual game played out quite often. Victoria's weird diagonal settling was repeated in almost every game, Napoleon got off to decent starts but his local terrain wasn't good enough to carry him out of that position, and Roosevelt had great land, but not a lot of it. Gandhi got off to slow starts, but had enough backlines to get to 8 cities consistently, which lended him a lot of strength (offset by Joao's own strength). These leaders had to have everything go right to emerge as the game leaders. Victoria had to survive her 1v1 with Ragnar (and yes, Victoria was very rarely left alone for more than 150 turns), Roosevelt had to get key barb cities in the south & conquer either Stalin or Napoleon, Napoleon had to get most of India or America, and Gandhi had to be completely left alone AND get enough religions to go culture. He was never a tech leader, with Victoria, Joao, Roosevelt, and Ragnar outdoing him in this regard, so he was unable to pick up the holy cities required to turn on the slider, and even when he did, he was on pace for a 350+ turn win due to not getting wonders like Sistine and the late game cultural wonders - except in Game 5, where he was almost completely left alone and, most importantly, Joao was copper decced by Stalin and died early.

These games were also defined by how much war happened. There were quite often three 1v1s going on at any given time, and there were many scenarios with multiple leaders getting 2v1'd at a time for some crazy diplo screens. Most commonly, Ragnar vs Victoria, Stalin vs Roosevelt, and Napoleon vs Gandhi played out again and again. Outside of extremely early copper declarations from Stalin on Roosevelt, these usually didn't do much, stalemates only being broken when someone else were to jump in. The most common outcomes involved Victoria, Roosevelt, Napoleon, or Gandhi getting attacked by multiple leaders and being dragged to their deaths. Pure solo conquests were rare, but when they happened, they allowed for dominant performances; Stalin & Ragnar were the most common leaders to pull this off. These wars slowed the tech pace to a crawl, as seen by the late finish dates: the average was 334, which is much later than some games with dominant Financial leaders. Victoria often found herself in a position where she was 5-10 techs ahead, but due to the small amount of land available to her was unable to defend against the hordes of units her foes threw at her. It wasn't uncommon for her to be defending with redcoats or infantry against lesser units, but to be overwhelmed by the larger production bases of her enemies. Getting rid of the only real economic leader (Ragnar is Financial, but let's be real, he's not an economic leader) left the door open to some much weaker techers to win very late victories.

Finally, religion played a huge part in the outcomes of these games. Gandhi always founded the first religion of Hinduism, and the second religion was important for both diplomatic reasons and, more mundanely, border expansions. Stalin, Joao, Roosevelt, and Ragnar are all leaders who often ignore mysticism, so if one of them were to chase the religion, even if they missed it, they were able to circumvent this huge weakness. Joao, Victoria, and Roosevelt most commonly founded this important second religion, but the games Stalin or Napoleon went for it were great examples of their strongest performances.

Now for a look at the individual leaders:

Stalin of Russia
Wars Declared: 64
Wars Declared Upon: 18
Survival Percentage: 55%
Finishes: 6 Firsts, 2 Seconds (34 points)
Kills: 27
Overall Score: 61 points

Stalin was the top leader on the map, as predicted by a large number of the contest entrants. His starting position was incredibly strong, with tons of room and great terrain. However, only having 8 top 2's (2 of which were carried off of having all of America before Turn 100!) shows glaring weaknesses of Stalin's AI: he is far too aggressive, and his tendency to ignore culture is a huge detriment on maps where he is in a central position. He stopped expansion way too early to plow units against his neighbors, threw away multiple second places because he just couldn't help himself from attacking Ragnar, Joao, Roosevelt, or Napoleon as they were running away with the game, and was culturally crushed by Victoria & Roosevelt, who are not really known for their cultural prowess. Although he is a pool 1 leader, the general consensus of many is that he lucked his way to the top, and his performance on this map, which truly should have been even more dominant than it was based on the start, shows ample evidence of that. Maybe I'm being a bit unfair to him, but I had some extra time after running these games and ran some tests by putting Huayna Capac and Justinian into Stalin's position, and they both won every single game I ran (5 each), with Justinian in particular just running away with things (winning domination before T300 - and as early as 245! - all 5 times). That's a small sample size, but shows a consistency present in other, stronger leaders that Stalin just does not have, particularly those who aren't suicidally aggressive and prioritize culture in some way. In the end, one of the strongest starts of the season was enough to give him a lot of kills and placements, but also exposed the weaknesses of a controversial AI.

Joao of Portugal
Wars Declared: 30
Wars Declared Upon: 25
Survival Percentage: 65%
Finishes: 2 Firsts, 8 Seconds (26 points)
Kills: 12
Overall Score: 38 points

Joao was probably the biggest oddity of the actual game. He, for some reason, spent a lot of turns without any production queues at all, which may possibly be due to a crashed economy (?). This DID, in fact, happen a couple more times, which resulted in two of his First to Dies: Games 5 and 6 (his other FTD saw his capital captured by barbs around Turn 35, somehow). In fact, this weird occurrence happened to Stalin, Roosevelt, and Napoleon a couple of times as well. I'm not 100% sure what was going on, their economies weren't crashed, but they did declare war soon after and that fixed it. Why they weren't building units if they were plotting war, however, is beyond me.

Joao mostly expanded, then declared war after building up for 150+ turns. In a world of war, his desire to remain peaceful for the early and mid game allowed him to get ahead of everyone else. His most common target once he made his move was Gandhi, and the two of them fought at some point in 90% of the replays. He was extremely strong in game after game, mostly collapsing in late game showdowns where someone else was an unstoppable runaway. Don't let his low number of firsts fool you: in 3 of Roosevelt's 4 wins, Joao had much more land & population, and only lost because Roosevelt hadn't been at war for 100+ turns by the end and managed to launch his spaceship first. In fact, 2 of those games had Joao sitting 2-3% below the domination threshold with a much larger army, and if he hadn't been Pleased-locked (as Joao is only one of two leaders on this map who could not plot at Pleased relations), he would've just ran over a few cities and won. This is a genuine weakness of some AIs, the inability to finish off a game, and it was on full display here (though you could argue it also saved his life in other alternate replays - it's a double edged sword!). Similarly, in Game 15 he was 2-3 turns away from winning domination off the back of America, but Roosevelt's spaceship arrived first. So Joao's actual strength is probably slightly underrated by his score (coupled by the fact Ragnar kept kill stealing from him!), and Roosevelt slightly overrated. If nothing else, 10 top 2's shows his consistent strength from a genuinely strong position.

Ragnar of the Vikings
Wars Declared: 60
Wars Declared Upon: 14
Survival Percentage: 50%
Finishes: 3 Firsts, 3 Seconds (21 points)
Kills: 16
Overall Score: 37 points

Our final southern leader, Ragnar managed 6 top 2's and a large number of kills, mostly from Victoria and Gandhi. He did scam quite a few kills that he didn't really deserve, mostly to the detriment of Joao, but them's the breaks. His highs were very high - he took out Victoria, then some combination of Gandhi or Roosevelt, and finished by backstabbing one of his low peace weight allies. His lows, however, were very low: while he was never once first to die, he was barely an entity in any game he didn't top 2, struggling to deal with Victoria and often getting taken out by a stronger leader. His worst games came when the high peace weights succeeded, as he and Napoleon in particular were inevitably carved up. In order for Ragnar to succeed, he needed Napoleon or Stalin to do well early on, and often required help to take out Victoria. If he was able to get the ball rolling, he was able to tech well due to the virtue of being Financial and keep taking out others; if he faltered, his infamous aggression came back to bite him over and over again, as shown by his large number of war declarations (60) that often resulted in his own death instead of his target's.

Roosevelt of the Americans
Wars Declared: 24
Wars Declared Upon: 42
Survival Percentage: 35%
Finishes: 4 Firsts, 2 Seconds (24 points)
Kills: 9
Overall Score: 33 points

Roosevelt was surprisingly competent in terms of research, probably due to how great his land was. All he needed was to conquer a neighbor, and he was actually extremely strong. He was quite boring after that first conquest, sitting there doing nothing most of the time, but this allowed him to build research in every city and make it over the finish line first, kind of scamming wins from Joao in particular. However, all that matters is that he made it there first, and he did take advantage of the weakness of Napoleon and Stalin in multiple games. His low survival percentage shows his typical fate, however: fodder to the flames of Stalin, Ragnar, or rarely Napoleon, only surviving to the end in a single game that he did not place in.

Napoleon of France
Wars Declared: 52
Wars Declared Upon: 32
Survival Percentage: 45%
Finishes: 2 Firsts, 3 Seconds (16 points)
Kills: 8
Overall Score: 24 points

Napoleon grades out higher than our actual game winner and runner up by virtue of his peace weight, being able to stick around in games that our top leaders dominated. He was rarely strong, though it did happen, usually off of an early dogpile of Roosevelt or Gandhi while dodging aggression from everyone else. His seconds were mostly distant ones, dodging aggression from a runaway that killed everyone else en route to domination, while the atypical Games 9 and 13 were the best examples of when things went right (notice that four of his five top 2's had Gandhi or Roosevelt as first to die; this was not a coincidence). He struggled to take out his neighbors in 1v1s, mostly due to Roosevelt's superior research and land quality and Gandhi's cultural borders. Napoleon just didn't have the land quality or quantity to get to catapults in a quick enough time frame to deal with these two, and being attacked by one of them after he declared on the other usually sealed his fate, even if it didn't result in an elimination.

Victoria of England
Wars Declared: 8
Wars Declared Upon: 47
Survival Percentage: 30%
Finishes: 2 Firsts, 1 Second (12 points)
Kills: 6
Overall Score: 18 points

Our actual game winner, Victoria unfortunately (and predictably) suffered from terrible diplomacy. I played the first game and thought that maybe I was wrong about her position on the map, but she dodged Ragnar's aggression until after redcoats in that game, something that almost never repeated itself. Her economy was much, much better than anyone else in the game, but she had to survive to the point of military supremacy while also dodging one of her neighbors getting too large. As mentioned previously, even if she was able to get a military tech edge, the superior production bases of her attackers were able to overcome redcoats with knights or infantry with rifles and cavalry. Her economic strength was on full display, she just did not have enough land to expand into in order to deal with the low peace weights consistently. Her performance was better than her low point total, but she wasn't able to translate that into consistent success.

Side note: she also was part of my personal favorite thing I've ever witnessed while running alternate histories, when her last city flipped to Roosevelt after he saved her life from Ragnar, eliminating her.

Gandhi of India
Wars Declared: 6
Wars Declared Upon: 66
Survival Percentage: 15%
Finishes: 1 First, 1 Second (7 points)
Kills: 1
Overall Score: 8 points

Gandhi was actually quite strong in the early- to mid-game, likely due to his sheltered position, and was capable of holding off Napoleon, which surprised me. However, this makes sense in hindsight: he had 40-60% cultural borders in most cities, and in these alternates wasn't in early culture mode (a very good thing for him). Napoleon was also quite weak in most of the replays, as his land quality was rough. Gandhi really only collapsed when attacked by multiple people, usually some combination of Napoleon, Joao, Ragnar, and Stalin. The offensive vs defensive war ratio (1 offensive war for every 11 defensive wars) tells the story: this was a man marked for death, and his economy wasn't good enough to be strong enough to survive, despite being able to backfill 8-9 cities and having gold at his capital. The rest of his land quality wasn't great, but he also didn't cottage much, farming everything next to the rivers near him for whatever reason. The two games he managed to be left alone for 150-200 turns translated into top 2's as the rest of the field warred themselves into the ground (note that Joao died first in both games, and he was just lucky to avoid Stalin's aggression in his second place game), but he was unable to do much the rest of the time.

Conclusions

The game we watched turned out to be atypical, mostly in the performance of Joao & the lack of early war between Victoria and Ragnar. That being said, there were patterns in the actual game that played out fairly consistently: Stalin declared war extremely early, the second religion's founder wound up being strong, and Napoleon was fairly weak. Victoria was able to repeat her success in a couple of games, which puts her win firmly into the "plausible, if unlikely" category. This was not a 99th percentile huge outlier game like Hammurabi in Game One this season, but Victoria and especially Gandhi were very lucky to make it out alive, regardless.

A typical game on this map is hard to define, with every single leader having some kind of success at some point. Stalin won the most, but his success came down to luck of the dice on who he decided to attack. The low peace weights were more consistent, winning more than half of the games, but first to die had FIVE different leaders plausible (all with 3-5 each, with no clarity on who was actually most likely), and the high peace weights won quite a few games themselves. Joao's ability to be on either side of the alignment really contributed to his success, and his survival rate shows that. I would probably say a typical game would feature strong appearances from Joao and one low peace weight leader (usually Stalin), with atypical games having Roosevelt/Victoria in positions of strength. The one place of true outlier territory from the actual game was Gandhi, with him surviving to the end only 4 times out of all 21 (including the actual game), and I would say that this was caused by Joao's abnormal weakness.