Game Four 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 Four had Mansa race out to an extremely fast cultural win while extremely strong Saladin got backstabbed by Qin to his doom. Was that something which would unfold in each game? 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. This particular set of alternate histories were run by Myth - many thanks for spending so much time on this task! 
Myth posted the resulting data from the alternate histories and then discusses some of the findings below 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.)
Myth: Back when the actual game happened, I thought Mansa would win about 40% and Saladin about 40%. I was sure there would be games where Washington collapsed and Saladin got to Mansa before he got rifles, or Zara would found a religion and attack Mansa. Instead, Saladin and Zara fought Mansa in a total of 5 games combined! Only one of those wound up being at all close, where Mansa held out against Zara for 40 turns and only lost a border city a few turns before his cultural victory ticked over (Game 4). Otherwise, these were very quick, very dominant victories, where Mansa just was not touched. I mean, look at those war counts - he didn't partake in a single war in SIX games, and only 23 total over the entire set.
It turns out the general community thought process was correct - Mansa would win too fast, with his neighbors liking him too much to attack him and those who didn't like him being too slow to make it through those neighbors. Washington and Zara acted as meat shields, taking aggression from Saladin and Qin while Mansa just sat there and teched away. He was always the tech leader, even with much fewer cities, except in one game where Saladin was very dominant but Pleased-locked and actually researched Plastics before Mansa won by culture. His land wasn't THAT great, but all he needed was the space he got. He was often winning with 8-10 cities while his rivals had 15+.
Mansa played his opening pretty terribly in the actual game, finding himself slow to expand and with an awful tech path. This was... honestly, a fairly common occurrence. He often went for one of the first two religions, missed, and then immediately researched Masonry and Monotheism to found the third religion. His expansion wasn't fantastic, often letting Washington found cities right on his border (sometimes south of him!). It just straight up did not matter because he was the only Financial leader and was diplomatically shielded. This isn't to say he was generally weak, though: he did wind up first in score by the mid game in most games, having a higher population and vastly superior tech than everyone else. He also only four times won from a place worse than second on the scoreboard. None of his wins were just backdoor cultural wins: Mansa was consistently very, very far ahead in tech.
The game also devolved into warring between everyone else, which slowed them down enough to allow Mansa to make it over the finish line. Most of the AIs were evenly matched, and this is exemplified by the two games with no one dying (Games 7 and 15). Those games just stalemated, and a stalemate is EXACTLY what Mansa wants. Honestly, despite the low point totals of most of these leaders, there were only two AIs who consistently played poorly and faltered, which we'll get into in a minute.
The biggest general trend on this map was religious conflict. These games constantly devolved into wars between religious faiths, with Saladin, Qin, Zara, and Mansa founding one of the first 3 in almost every single game. This led to conflict to explode in the west, with Washington, Zara, Qin, and Saladin fighting each other based on where the religions spread. These 4 were fairly evenly matched, but Washington drew the short straw and got dogpiled almost every game, leading to a high elimination rate for the American leader.
The 6 non-Mansa leaders are in three tiers; to start, Saladin and Pericles held 85% of the second places between them, and their strengths laid in the weakest two links on the map. Alex and Washington died in the majority of games, and these two were the biggest gainers of their demise. Saladin was as strong as he seemed in the actual game, only dying in the last two games and generally being very strong. Washington almost always got Mansa's or Zara's religion, and then was a ripe target for Saladin's aggression. Meanwhile, Pericles was more boom-or-bust, dying in a quarter of the replays, but also MUCH stronger than the version we saw in the actual game. He never once went for religion in any of the 20 alternates - this was hugely beneficial, and he only lost the Greek Civil War two or three times total. Similarly to Washington's for Saladin, Alex's weakness is the reason for Pericles's success.
The third tier were Qin and Zara. These two were also very strong in most games, but not QUITE as strong as Saladin and Pericles, which led to them having low point totals. Don't let that fool you, however: they often were able to expand & tech well, and their survival rates were extremely high as a result. Both of these leaders were defined by religion, as mentioned before, which caused Qin and Zara to distract the western leaders, and Mansa was able to be left completely alone. Something that came up a lot post-game was the possibility of a religious Zara vs Mansa conflict, but very rarely did they fight over their differences, as even when they were under opposing faiths, Zara wound up in war with Qin or Saladin more often than Mansa. They did fight, but only in 3 games across all 20 replays (two of which were started by Mansa!). The rest of the field NEEDED Zara to attack him in order to come out on top, and he just didn't.
Our runts were Alexander and Washington: these two were the only ones to die more than half of the time. Starting with the American, he was sandwiched between the two strongest leaders on the map, and that ended poorly for him. He usually died 1v1 to Saladin, but was often dogpiled by Saladin and Qin, and rarely Zara and/or Mansa. The only thing keeping him from being the runaway favorite for First to Die was these leaders were all usually distracted by other targets (other than Mansa). Alexander, on the other hand, fought Pericles 100% of the time, and often lost that 1v1 eventually. He also got attacked and was attacked by Qin quite a few times, and Alexander also did cross map attacks against Zara or Washington to bring their fury upon his head. Other than his own personality, I'm not sure what went wrong for him here: his land wasn't that bad, he just flailed about in most games. He had one single game he was able to do something, beating Pericles and then winning the 1v1 w/ a strong Saladin, finding himself as the strongest AI on the map (but still trailing Mansa by over 1000 points).
Now for a look at the individual leaders:
Mansa Musa of Mali
Wars Declared: 17
Wars Declared Upon: 6
Survival Percentage: 100%
Finishes: 20 Firsts, 0 Seconds (100 points)
Kills: 1
Overall Score: 101 points
For the first time in AI Survivor History, we have someone who won all 20 replays in the Alternate Histories. His odds weren't actually 100% (as shown below...), but things had to go perfect for the rest of the field for them to win, and that never arose in the official replays. Mansa would need to be slowed down, and that just didn't happen, even when getting into an early war with either Saladin or Zara.
Really, it all comes down to speed. Mansa was just way too fast, and no one else could keep up. If we had an Elizabeth, Huayna Capac, or Pacal on this map, these might have been races, but no one else in this game had the economic caliber to keep up with him. This allowed him to found at least 4 religions in every game and run away culturally. Likewise, if someone who could plot at Pleased was in Saladin's spot, things might have also wound up differently. Saladin was very strong in multiple games but just could not plot by the time he got to Mansa's borders. At the very least, some of these victories would have been stopped through a city capture of one of Mansa's Big Three. The story here is the war count: 23 total over 20 games is incredibly low (less than 6% of all wars across all 20 games, less than half of 1/7), and only 1 kill across 20 victories is just insane.
However, I would argue that this was NOT the most dominant alternate history of AI Survivor. Mansa might have won all 20 times, but he only got 1 kill. Pacal's performance in Season Six Playoff Two, where he won 19/20 and got 20 kills, as well as Huayna's performance in Season Four Game Seven (which I ran), where he won 16/20, got 11 kills, and was close to domination in almost every game, were more impressive to me. Heck, even Hatshepsut in Season Seven Game Seven got 6 kills in her 17 victories! Now, he never lost a war, but he also was almost never attacked early. If he had been, he would not have won this much. The diplomatic bodyguards were much more important for Mansa's success than his actual strengths as a leader (though of course those were on display as well).
Saladin of Arabia
Wars Declared: 39
Wars Declared Upon: 30
Survival Percentage: 90%
Finishes: 0 Firsts, 10 Seconds (20 points)
Kills: 13
Overall Score: 33 points
I remember there being a LOT of heated discussion about this game after the actual game happened, with some people believing Saladin was robbed and should have won the game. They were correct in part of that - he was most definitely robbed... but just of survival and a likely silver medal. Saladin's biggest flaw is really what came back to bite him in these replays: he cannot plot war at Pleased relations. Mansa needed to survive until Free Religion, where open borders, trading, and years of peace would tick Saladin up to +4, which is all he needed to hit that magical barrier. Unfortunately for Saladin, this happened too fast, with Mansa usually swapping to the civic around Turn 180 at the latest. There just was not enough time for him to snowball through a neighbor (usually Washington), consolidate, then attack Mansa. He also was in a centralized start, surrounded by religious enemies or cultural civs with considerable border tension: Qin and Zara often founded opposing religions, Washington was the closest (and easiest) target, and Pericles's borders often pushed into Saladin's. There just isn't much to be done when the competition is winning at or before Turn 280 in game after game, especially when considering that Saladin is not the greatest leader in terms of research.
Pericles of Greece
Wars Declared: 11
Wars Declared Upon: 31
Survival Percentage: 75%
Finishes: 0 Firsts, 7 Seconds (14 points)
Kills: 10
Overall Score: 24 points
The Pericles we saw on Livestream, who put on one of the weakest performances of the season, was the most atypical part of the official game: even in his first to die games, he never performed nearly as bad. As aforementioned, this lies in his decision to not found a religion: in the actual game, he founded one of the initial two, and that set him back by over a dozen turns & turned out to completely doom his game. When he wasn't chasing religion and instead researching worker techs, he got off the ground much faster, killing Alexander in most games and being a contender for a top 2. His worst games came at the hands of a dogpile by either Qin or Saladin (usually the latter) before he could finish off Alex. I was genuinely impressed here - I would say Pericles probably had the 4th or 5th best start, and he greatly outperformed that, being the second best non-Mansa leader on the map, reliving some of his glory days. At least he made it to the playoffs through the Wildcard!
Zara Yaqob of Ethiopia
Wars Declared: 23
Wars Declared Upon: 24
Survival Percentage: 75%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 8
Overall Score: 10 points
For both Zara and Qin, their low points belie their true strengths. Zara only had 5 more points than Washington, and Qin only 2, but those totals do not paint an accurate picture at all.
Zara played kingpin in these games quite often. Not because of what he did, but because of what he did NOT do: attack Mansa Musa. He was the only leader in a position to do so early enough for it to matter, and he refused to in game after game, despite being able to plot at Pleased and often practicing a different faith from Mansa. This is actually understandable, if we look deeper into his personality. He only penalizes other leaders by up to -2 (-3 if he controls the holy city) for differing religions and has base relations of +2, effectively canceling each other out. Even when under different faiths, it was quite easy for him to tick over to Pleased with Mansa, which then greatly reduced his odds of attacking the Malinese leader when compared to those he shared larger borders with. When practicing the same faith, Zara can give up to +7 diplo points, and that was more than enough to make him Friendly with Mansa when they shared a religion and give him absolute immunity to his southern neighbor.
In terms of his actual performance, Zara was alright. He often expanded quite well & researched decently, actually teching better than everyone except Mansa and Pericles, and turned that into some strong third places. There were actually a couple of games where he was second in score, but a third place Mansa won by culture and didn't allow him to take a podium spot. It's rough being in a world with Mansa Musa.
Qin Shi Huang of China
Wars Declared: 39
Wars Declared Upon: 36
Survival Percentage: 80%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 5
Overall Score: 7 points
Our actual silver medallist, Qin managed to perform admirably throughout these replays. His peace weight of 2 was sorely outnumbered in this game (where he touched borders with a 0, 4, 6, 6, and possibly an 8 in most games!), yet he managed to expand extremely well and fight good wars with his neighbors. He just had the same problem as Zara: he was often in a strong third place, but behind Saladin or Pericles and unable to make it onto the podium. His second place finish wasn't significantly better than his other performances, he just managed to get more of the spoils from his wars, and therefore actually make it into the second place spot in terms of score.
The biggest surprise from Qin was his tendency to found Taoism, which was a huge part of the dynamic of this game. He and Saladin fought a lot more than I initially thought they would, and this generally stalemated (to the benefit of the Malinese leader). His strongest games came from expanding well into attacking (or being attacked by) either Zara or Alex (but not both!). He generally won the 1v1s with Zara, which makes his placement below the Ethiopian leader a bit sad.
Washington of America
Wars Declared: 23
Wars Declared Upon: 45
Survival Percentage: 40%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 5
Overall Score: 5 points
Poor Washington, doomed to be a punching bag by his neighbors in this game. He was attacked by every single leader other than Pericles at least once across these 20 replays, and his death in Game 18 was Mansa's only kill. He performed admirably for what his role was: he often teched well, but his land wasn't nearly as good as anyone else's (when 4 leaders have gold at their capitals, it's a bit rough!). His biggest issue lay in being culturally crushed by Mansa and having too high of a peace weight to get along with Saladin, especially since he almost always followed the religion of either Zara or Mansa. He usually was able to hold 1v1 with a neighbor, but he pretty much never got that luxury, being forced into 2v1s with some combination of Qin, Saladin, and Zara, mostly to the benefit of the Arabian and Malinese leaders. There isn't much more to say: Washington was pretty screwed in this game, being extremely lucky to get 5 kills.
Alexander of Greece
Wars Declared: 41
Wars Declared Upon: 21
Survival Percentage: 20%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 2
Overall Score: 4 points
At least he wasn't Alex, though! Alexander's performance was just sad, and he really only has himself to blame. He had nice land, he just floundered it by being way too aggressive, killing his economy through wars in game after game. There was one game where he was on longbows when Mansa had tanks - a bit of a mismatch there! My opinion on Alex is that he does not belong in Pool Two, and this game did nothing to dispel that notion. Outside of his one second place, which was extremely unlikely, Alex did nothing but manage to killsteal in this one. It does not pay to be a hyper warmonger in fields with stronger leaders. Maybe things would have gone better for old Alex if Zara was in Pericles's spot! 
Note: First to Die only adds up to 90% because of two separate occurrences of no one dying.
So it turns out that the game we watched actually turned out to be atypical! Yes, Mansa won by culture, but let's ignore that. Saladin was eliminated 3 times in 21 replays and was twice First to Die, and both of these happened in the actual game! The backstabbing of Qin wasn't atypical - they fought a LOT - but the result was. We also didn't see either Washington or Alex get eliminated, and both were more than 50% of the time! Those two making it through to the wildcard should not have happened. Finally, Pericles was abnormally weak, almost never being the also-ran we saw on livestream, and Qin came second, which only happened one other time. For a game with so much predictability, the nuances of each replay were different enough to be significant.
There was a "correct" option for a prediction: Mansa first, Saladin second, Alexander (or Washington) First To Die, Cultural, Turn 280(ish), and 10 wars. This was... almost exactly what I predicted, in a huge sad moment for me (I only whiffed on FTD, picking Qin).
--------------------------------------------------------------------------------------------------------------------------------
I wound up playing some extra replays of the map, just to see how long it would take Mansa to lose. We wound up with this result:
Pericles outraced Mansa to a cultural win in Game #27 of the map at the "late" turn of 320. This was caused by Zara attacking Mansa THREE TIMES in the game, the first of which was before Turn 120. Mansa did win every conflict due to vastly superior military tech, but this wound up slowing him down drastically; he still had about 9 turns on his own cultural victory in this one. So Pericles gets to put a feather in his hat as the only leader not named Mansa Musa to win in 28 alternate histories, something I was NOT expecting before I started these.
