Game One Alternate Histories Spreadsheet
One of the recurring features of past seasons of AI Survivor have been our "alternate histories", running additional iterations on the same maps to see if the same events would play out again. Game One had a wild and unexpected outcome as pregame favorite Kublai Khan was eliminated and Hammurabi fought off a 4 vs 1 war to emerge victorious. Was that something which would unfold in each game? This was a topic that called for more investigation with alternate history scenarios. Following the conclusion of previous seasons of AI Survivor, I had gone back and investigated some of the completed games and found that they tended to play out in the same patterns over and over again. While there was definitely some variation from game to game, and occasionally an unlikely outcome took place, for the most part the games were fairly predictable based on the personality of the AI leaders and the terrain of each particular map. Would we see the same patterns play out again and again on this particular map?
The original inspiration to run these alternate histories came from Wyatan. He decided to rerun the Season Four games 20 times each and publish the results. The objective in his words was twofold:
- See how random the prediction game actually is. There's a natural tendency when your predictions come true to go "See! Told you!", and on the contrary to dismiss the result as a mere fluke when things don't go the way you expected them to (pleading guilty there, Your Honour). Hopefully, with 20 iterations, we'll get a sense of how flukey the actual result was, and of how actually predictable each game was.
- Get a more accurate idea of each leader's performance. Over 6 seasons, we'll have a 75 game sample. That might seem a lot, but it's actually a very small sample, with each leader appearing 5-10 times only. With this much larger sample, we'll be able able to better gauge each leader's performance, in the specific context of each game. So if an AI is given a dud start, or really tough neighbours, it won't perform well. Which will only be an indication about the balance of that map, and not really about that AI's general performance. But conversely, by running the game 20 times, we'll get dumb luck out of the equation.
Wyatan did a fantastic job of putting together data for the Season Four games and I decided to use the same general format. First I'll post the resulting data and then discuss some of the findings in more detail. Keep in mind that everything we discuss in these alternate histories is map-specific: it pertains to these leaders with these starting positions in this game. As Wyatan mentioned, an AI leader could be a powerful figure on this particular map while still being a weak leader in more general terms. Now on to the results:
Game One | Game Two | Game Three | Game Four | Game Five
Game Six | Game Seven | Game Eight | Game Nine | Game Ten
Game Eleven | Game Twelve | Game Thirteen | Game Fourteen | Game Fifteen
Game Sixteen | Game Seventeen | Game Eighteen | Game Nineteen | Game Twenty
(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.)
There was an incredible amount of interest in the alternate histories for this game after we had the surprising Hammurabi victory on Livestream. I had no fewer than three different people separately contact me afterwards asking if they could run alternate histories for the map... which I had to reject because I had *ALREADY* started running my own set of alternate histories the very next day! 
A lot of people wanted to know what would happen if this map was replayed and, well, it turns out that we did indeed see a wildly atypical result for this setup. Hammurabi did not win any additional victories in 20 map replays and in fact was the lowest-scoring leader of the bunch, with a 90% elimination rate while being overwhelmingly the most frequent First to Die. The top-scoring leaders were instead Peter, Mao, and Kublai who were all eliminated in the actual game without scoring a single point. So what in the world happened to throw everything so badly off track and what did a more normal game look like for this map?
The first major difference from the game that we watched on Livestream concerned how the Kublai/Mao relationship played out. The biggest change in the alternate histories was Mao being much, much stronger in virtually all of the map replays. He never had that second city of Shanghai culture-flip to Kublai in any of the alternate histories and basically never ignored researching cultural techs in the way that we watched. In fact, Mao actually founded his own religion about a third of the time which would then appear in that very city of Shanghai! He also sometimes placed his second city to the north by the copper, in which case it would be his third city going east by the gold resource. All of this resulted in Mao having much stronger early games that typically had him competing for the game lead, not being one of the weakest AIs getting dogpiled by his neighbors. Mao was only First to Die a single time in the 20 alternate histories which suggests that him being the first to exit was highly unlikely.
As for Kublai, he had to contend with a stronger Mao in most games, but this was counterbalanced by the Mongolian leader choosing a smarter tech path almost all of the time. I was watching his tech choices closely in these repeat games and he essentially never followed the same idiotic pattern of Turn 45 Mining and Turn 80 Pottery. This meant that Kublai connected his gold resource faster and built cottages on all those floodplains much faster, giving him a better tech rate. However, because Mao was also stronger in these games, these two factors tended to cancel one another out and Kublai ended up roughly comparable to the power level we watched on Livestream. He was not the overwhelming juggernaut that the community expected in the picking contest, although there were a couple of games where he had an unusually good landgrab plus favorable diplomacy and he just steamrolled everything. That was a rarity but it did happen several times.
With a stronger Mao and a roughly comparable Kublai to the south, the other massive change in the Livestream game was the status of Hammurabi. He was consistently the top AI from start to finish in the actual Game One and, uh, that turned out to be an exceptionally unlikely outlier result. Hammurabi had a beautiful starting area and he typically sat at the top of the scoreboard for the first 75 turns in the alternate histories as well. Then the diplomatic side of the gameplay would kick in and Babylon was absolutely screwed: Hammurabi was attacked 87 times (!) in 20 games, an astronomically high number. I've done a lot of these alternate histories and that might be the highest number of defensive wars I've ever seen. It requires a really rare situation to get attacked that many times: diplomatically unpopular enough to keep getting invaded while also strong enough to fight off repeated attacks so even more war declarations can follow.
Hammurabi was extremely lucky in the real Game One because the warring broke down into three separate 1 vs 1 conflicts (Hammurabi/Peter, Boudica/Mao, Kublai/Churchill) which lasted for long enough that Hammurabi could research his way into a tech lead. Then Peter followed that up by invading Boudica, something that did happen but was a rarity in the alternate histories, creating a window for Hammurabi to solo kill Mao while Kublai was still fighting Churchill. Let me be clear: that *NEVER* happened again in any of the alternate histories, nor did Hammurabi ever come even remotely close. Instead, the Babylonians were relentlessly dogpiled in game after game after game, with Hammurabi often facing 3 vs 1 or 4 vs 1 or even 5 vs 1 scenarios. The 4 vs 1 that he saw on Livestream was absolutely typical in the repeat map playthroughs, just imagine that happening 150 turns earlier and with no tech advantage for Hammurabi. It wasn't pretty and he was crushed over and over again.
The biggest beneficiary of Hammurabi's demise was Peter who essentially tied with Mao and Kublai for the top score in the alternate histories. Between having a sheltered corner start and possessing an appropriately low peaceweight score, Peter was rarely attacked and therefore scored a bunch of first and second place finishes, 14 Top Two finishes in 20 games. He ended up with fewer kills because he wasn't warring as often but he was the most consistent leader on the map in terms of advancing. Mao and Kublai were both more boom-or-bust leaders, with an equal number of victories and more kills earned but also significantly higher odds to be outright eliminated. They fought with one another a lot due to all their border tension and usually it was one or the other that would take a Top Two spot, not both of them together. In fact, they had two games where they shared the top spots and then 13 games where one of them was Top Two without the other one - that's quite a pattern! In other words, they would typically both get strong and then there would be a big clash where one eliminated the other one for supremacy.
For the last two remaining leaders, neither one of them tended to have a big impact on the gameplay. Boudica had the correct peaceweight for this map but her economy was simply too weak to remain competitive as she would always fall behind in tech by the midgame. This normally resulted in her getting eliminated as her survival rate was barely better than Churchill (30% to 25%) despite having a seemingly favorable diplomacy setup. It's hard to survive when you're still sporting rifles against tanks and mechs though. All of Boudica's Runner Up finishes were from backdoor results and she was never close to winning a victory herself. As for Churchill, he did win one time in the atypical Game #8 where he had an early solo kill on Kublai. More commonly though, Hammurabi would take an early exit and then Churchill was the next domino to fall due to his high peaceweight. He clung on for a couple of tenacious second place finishes but ultimately the diplomatic situation was too bad for him and his starting position wasn't rich enough to survive with the deck stacked against him.
Now for a look at the individual leaders:
Peter of Russia
Wars Declared: 55
Wars Declared Upon: 20
Survival Percentage: 75%
Finishes: 6 Firsts, 8 Seconds (46 points)
Kills: 15
Overall Score: 61 points
I want to stress again that Peter, Mao, and Kublai were all so close together in the alternate histories scoring that there was no statistical difference between them, certainly not across a mere 20 games played. From what I saw though, I'm confident enough to say that Peter had the best odds to take some kind of Top Two finish due to his more sheltered starting position, at the expense of winning fewer crushingly dominant victories. Peter had a surprisingly good economy given his cramped coastal area and he leveraged the Colossus and Great Lighthouse to good effect in most of these games. He was also the only leader to earn balanced victory conditions: 2 Spaceship wins, 2 Cultural wins, and 2 Domination wins as everyone else just won by Domination all the time. Peter had the best survival rate of this group and was attacked less than anyone else with only 20 defensive wars faced. Long story short, this is yet another year where Peter did better in the alternate histories than in the real game; eventually he's going to stop having such bad luck in the games we watch on Livestream.
Mao Zedong of China
Wars Declared: 48
Wars Declared Upon: 35
Survival Percentage: 60%
Finishes: 7 Firsts, 3 Seconds (41 points)
Kills: 19
Overall Score: 60 points
Mao's alternate history games didn't look much like the real Game One as he was consistently one of the strongest leaders on the map and competing for a first place finish. He technically had the most victories out of anyone in this set of map replays although with the score standing at Mao = 7, Peter = 6, Kublai = 6 I'm not going to draw any larger conclusions from such exceedingly close results. Mao's best results tended to come from when he fought Kublai and conquered the Mongol territories which happened reasonably often, I'd say about a third of the time. I had suggested before the game took place that I thought Mao and Kublai would fight a lot, and while that didn't happen on Livestream, it most certainly did in some of the alternate histories. His other path to success came from conquering Boudica first which was easier to do (since Boudica was a weaker AI) but yielded less of a reward; taking out Boudica typically wasn't enough to boost Mao into a winning position whereas eliminating Kublai most definitely was. Mao's central position meant that he found himself in a lot of wars and was at risk of getting stuck in multi-front conflicts, but when that didn't happen, he was positioned to be a real player in these games.
Kublai Khan of Mongolia
Wars Declared: 53
Wars Declared Upon: 29
Survival Percentage: 60%
Finishes: 6 Firsts, 3 Seconds (36 points)
Kills: 21
Overall Score: 57 points
The massive pregame favorite was nowhere near as dominant as expected while still being a powerful leader who had a good chance of winning on this map. I covered the main differences from Kublai's Livestream performance in the overview above: smarter tech choices in general, balanced out by having a tougher Mao to his west. Out of the six AI leaders, Kublai tended to have the most variance in terms of how much land he would obtain during the early portions of the game. Sometimes he would avoid early fighting and dodge barbarians to result in a huge empire, like in Game #14 where everything broke his way and he was double anyone else's size before any conflict began. Other times Mao would claim more of the floodplains region between them and barbarian cities would appear in awkward spots to limit the effectiveness of the Mongols. Again, I think Kublai's performance in the Livestream game was right about in the middle of the spectrum in terms of these results, not unusually lucky or unlucky (as opposed to Mao who was at the bottom of the luck scale and Hammurabi who was at its absolute peak). Kublai's best games occurred when he was able to solo kill Churchill and then snowball from there; he did less well when he fought Mao or when he went after Hammurabi first as he was often unable to make use of captured Babylonian cities. Kublai's performance really speaks to the randomness of these AI Survivor games: he wound up being a huge dud on Livestream but he also could have won an overwhelming Domination victory as happened about a third of the time in the alternate histories, and then the community would have looked 100% correct instead of face-planting in epic fashion. That's why we love these games though!
Churchill of England
Wars Declared: 28
Wars Declared Upon: 45
Survival Percentage: 25%
Finishes: 1 First, 2 Seconds (9 points)
Kills: 5
Overall Score: 14 points
Game One essentially had three co-favorites and then fell off a cliff before reaching the other three AI leaders. Churchill narrowly edged out Boudica for the best of this bunch although they were so close in score, propped up by the one victory won by Churchill, that there wasn't any clear difference between the two of them. Churchill was very much his normal self in AI Survivor terms, a leader without much economic heft who's always a different customer to eliminate by virtue of his redcoats and Protective trait. Churchill had a good capital and I think he would have been a real contender if he hadn't been undermined so badly by this game's diplomatic environment. His biggest problem was Hammurabi getting eliminated in almost every game, which both removed a potential ally and also strengthened the low peaceweight wolves that Churchill was competing against. The English leader was normally able to defend himself in a 1 vs 1 scenario, only to find himself often getting dogpiled by multiple leaders or facing a Peter/Mao/Kublai who now had the territory of two nations under their control. Churchill played reasonably well in these games but he just didn't have enough on hand to be competitive unless luck broke hugely in his favor.
Boudica of the Celts
Wars Declared: 47
Wars Declared Upon: 26
Survival Percentage: 30%
Finishes: 0 Firsts, 3 Seconds (6 points)
Kills: 6
Overall Score: 12 points
We've been watching AI Survivor for enough years by now to have figured out that Boudica simply isn't a very good leader for these settings. This was disguised during the early years of AI Survivor where Boudica benefited from some strong starting positions and the bonus Deity techs. Once those were removed though, Boudica has struggled badly due to her poor starting techs (Mysticism/Hunting), which always cause her to chase after early religions, along with militaristic over-aggression and leader traits which are awful for anything other than fighting. Boudica didn't have the best quality land around her capital in this game and she completely neglected to build an economy, always falling well behind the top leaders by the middle portions of the gameplay. She was never quite as badly off as in the Livestream game where that captured barbarian city dropped her down to 2 beakers/turn, but the Celtic economy still remained in low gear across all the map replays. Thus she was stuck in the classic Montezuma trap: a warmonger who won't stop invading other nations while falling generations behind in military tech. It's a recipe for failure and that was the case again here. As I said above, all three of Boudica's Runner Up finishes were distant backdoors as she never came even remotely close to winning a game herself.
Hammurabi of Babylon
Wars Declared: 11
Wars Declared Upon: 87
Survival Percentage: 10%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 2
Overall Score: 4 points
But Boudica was still better off than this poor guy, sheesh. The war tally really sums up Hammurabi's diplomatic situation: 11 offensive wars, 87 defensive wars. 
There was a point where I thought he legitimately might reach 100 times invaded across 20 games until he was attacked "only" twice in 3 of the last 5 games. It still added up to nearly 4.5 foreign invasions per game and no leader in Civ4 can be consistently successful when facing that many incoming hostiles. In retrospect, it was an absolute miracle that Hammurabi was able to win the game on Livestream as we saw him somehow survive NINE different attacks in that game before pulling off the Domination win. For everyone who watched that and thought, "wow there's no way Hammurabi could pull that off again", you were absolutely correct because he didn't in the alternate histories. This guy was a dead man walking and absolutely the correct pick to be First to Die.
What a wild game this one was and a perfect start to a new season of AI Survivor. Even though the game we watched on Livestream was bizarrely atypical and resulted in most everyone failing to score points in the picking contest, I'm glad that we ended up experiencing it because the whole thing was such an exciting nailbiter. Even with the benefit of hindsight in the alternate histories, this really was not an easy game to predict due to its open-ended nature. Three different leaders had roughly equal odds to claim the victory - something that's extremely rare for AI Survivor which usually has one or at most two favorites - and then both Runner Up and First to Die were total crapshoots if Peter/Hammurabi didn't take those categories. Good luck making sense of all those 10-15% results on the table, I watched the games and I can't sort them into a coherent narrative. This whole scenario was innately unpredictable and I suspect that if you ran another 20 alternate histories there would be a different finishing order for Peter/Mao/Kublai because they were all so close. I guess that I picked a good map for this first game.
Thanks as always for reading, I hope you enjoyed this look back at Game One!
