Game Six 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 Six shocked everyone as Churchill, the overwhelming favorite as First to Die, instead went on to win a dominating victory over the rest of the field. 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.)
Game Six turned out to be one of the most difficult games to write about because it was hard to find clear patterns from repeated playthroughs of the map. While there was indeed one leader who was confirmed as the favorite for this scenario, the results were otherwise pretty muddled and there wasn't much distinction between most of the leaders. That favorite was Churchill who backed up his victory in the actual Game Six by winning eight additional times in the alternate histories. No one else was even remotely close to that many first place finishes and there was no doubt that Churchill was indeed the best leader for this particular setup. Everyone agreed that Chuchill had a beautiful starting position which was particularly well suited for England's Fishing/Mining starting techs, however there was essentially no support for Churchill in the picking contest due to his seemingly impossible diplomatic situation. Churchill was the only high peace weight leader in field full of low peace weight leaders; combine that together with Churchill's lousy Protective/Charismatic traits and poor track record of past performances and surely he was doomed, right?
We were all in for a surprise at Churchill's success both in the real Game Six and in these alternate histories. I found that there were indeed some games where Churchill was simply dogpiled into oblivion (he was indeed attacked more than anyone else) and suffered an early elimination. That tended to be fairly unusual though and Churchill was generally able to stay out of two-front wars in these games. He was a massive beneficiary of having Genghis Khan Temujin in the game on the other side of the map, as the eastern Mongol leader constantly attacked Cyrus and Tokugawa which kept pulling the Japanese away from Churchill. Boudica and De Gaulle were far enough away from Churchill that they normally left him alone in the early game, leaving Kublai Khan as the only leader who frequently sparred with Churchill. The English leader's far superior land meant that he almost always won that duel and could then snowball into a Domination or Spaceship victory as needed. The larger seven player field almost certainly helped Churchill in this regard as the bigger map kept some of the most aggressive AI leaders away from his start and allowed the English to develop in peace. Churchill's economy was consistently the best in the game and that was purely due to Churchill's amazing land with floodplains and grassland gems resources aplenty. A leader with actually good traits would have been terrifying to behold from this spot.
After Churchill, the rest of the leaders were basically a crapshoot in terms of what happened next. Boudica, Tokugawa, Kublai Khan, and Cyrus each won exactly three times apiece which suggests that they all had similar odds to emerge victorious. I'm curious how the numbers would shake out if we could run this map 100 turns and get a truly representative sample size instead of the 20 times that I watched here; just not enough time to devote to an exercise like that for better or worse. Boudica scored the best out of this group thanks to landing the most kills of anyone in the alternate histories (edging out Churchill by a nose) and also taking home the most second place finishes. Her runner up finishes may have just been dumb luck though as no one had any particular edge in that scoring category and we might as well have asked Random.org to pick who was going to finish second in any one game. Boudica killed De Gaulle a lot in the alternate histories and was often strong even in games where she wound up losing in the lategame. Her two weaknesses were poor teching (much worse than Churchill) and founding a religion that often didn't spread very well. A lack of rivers on the northern side of the map likely contributed to both of these issues. I think that Boudica was a slight cut above the other non-Churchill leaders but not by a wide margin.
The other five leaders were honestly pretty difficult to distiguish from one another. Leaving aside De Gaulle for the moment, the other four of them scored between 29 and 21 points across the alternate histories which is close enough to put them all in a statistical dead heat. Tokugawa, Kublai Khan, and Cyrus each won three times and had nearly identical victory points with the different amount of kills that they scored largely accounting for the separation in their totals. They were each close enough that slightly better luck in terms of claiming finishing blows could have shuffled the order of their finish around. Genghis Khan arrived at his 21 points in a very different fashion, scoring no victories at all and instead putting together a whole bunch of kills. This was completely expected for his insane AI personality, of course, as the eastern Mongols fought constantly with everyone and self-combusted in spectacular fashion in many games. Leaders like Temujin and Montezuma and Ragnar will almost never win their games outright but they have good odds to score a decent number of kills because they put themselves into so many total wars.
De Gaulle graded out as the weakest leader since he also didn't achieve any victories and had few kills to his name. De Gaulle seemed to struggle with his starting position more than expected and I wonder if that little lake to his immediate northwest did something to the AI's performance. But even De Gaulle wasn't that far behind the rest of this group and it would have taken only a single victory to put him right there with the rest of this uninspiring group. My main takeaway from running this game so many times was that none of these leaders were particularly impressive outside of Churchill (due to his amazing starting position) and maybe Boudica in some games. The others were mediocre leaders with low peace weight scores and militaristic tendencies who could all win games if the diplomacy happened to break their way. Kublai Khan has a great track record in AI Survivor and things may have been different for him with a better starting position. With the seafood resources and jungle terrain that he had here, however, Kublai wasn't any better than the rest of the bunch. These were mostly slow war-filled games characterized by poor teching and late finishing dates. I didn't find these alternate histories to be terribly compelling to watch which is why it took me about two months to play through all of the games. (By contrast, I ran all of the alternate histories for Game Seven in less than a week because I was highly interested in assessing Mansa Musa's performance.) This one was honestly kind of a slog.
Now for a look at the individual leaders:
Churchill of England
Wars Declared: 32
Wars Declared Upon: 59
Survival Percentage: 55%
Finishes: 8 Firsts, 2 Seconds (44 points)
Overall Score: 63 points
Churchill was unquestionably the top leader for this scenario thanks to having such a strong starting position. He won close to half of the time despite being attacked at the highest rate of any of the leaders on the map. Churchill won the first four replays of this map and I thought that he might be an unstoppable juggernaut before things evened out over the remaining alternate histories. He won most often by Domination thanks to getting attacked so often and then responding by overrunning his technologically backwards neighbors. Churchill was almost always the most advanced leader in the bunch solely due to his amazing local terrain. England was blessed with bountiful floodplains along with gold and gems resources which helped ensured that Churchill had an excellent GNP. As long as he could avoid getting dogpiled in the early game, Churchill would usually win eventually due to his superior economy. This is why Churchill had so many first place finishes but only two second place finishes and a mediocre survival rate of 55%: he either survived and won or wound up getting eliminated. Churchill's fate was somewhat surprisingly tied to what Tokugawa was doing and specifically whether Japan found itself embroiled in warfare with Genghis Khan. If Tokugawa ended up stuck in eastern wars then Churchill usually won whereas when Toku headed west then Churchill was in deep trouble. More often than not it worked out for the English leader. The fact that he "only" won eight times is a sign that Churchill lacks the strength of the top AI leaders who likely could have turned this opening position into a 100+ point performance.
Boudica of the Celts
Wars Declared: 49
Wars Declared Upon: 31
Survival Percentage: 55%
Finishes: 3 Firsts, 5 Seconds (25 points)
Overall Score: 45 points
There were a bunch of rather mediocre low peace weight militaristic leaders who were drawn into this game and Boudica wound up being the strongest of the bunch. She won three victories, scored the most kills in the alternate histories, and also achieved the most runner up finishes, albeit only by a nose and within the margin of small sample size error. Boudica had a clear path to success in these games: absorb De Gaulle's territory in early warring and then spread her religion enough to make at least one other leader into a strong ally. That was the route to victory in her strongest performance in Game #7 and Boudica also had some success with conquests of Cyrus on her other border. Boudica tended to do worse when she was pulled to the south into wars with Tokugawa since any land that she conquered down there tended to be culturally crushed by the borders of other leaders. It's also worth noting the victory dates of Boudica's three wins: Turn 372, Turn 382, and a earlier Domination result on Turn 326. Boudica suffered from a poor economy and she benefitted from slow, drawn-out games with heavy warring on all sides. The Celts tended to fall apart in games with a faster teching pace and there were several times when Boudica amassed a lot of territory only to collapse late against a rival who had superior technology. She didn't have great luck at spreading her religion either which I suspect was due to a lack of rivers in her core lands. This was a solid but not spectacular performance overall which follows the pattern of prior AI Survivor seasons with Boudica being a decent leader who falls somewhere in the middle of the rankings.
Tokugawa of Japan
Wars Declared: 39
Wars Declared Upon: 57
Survival Percentage: 30%
Finishes: 3 Firsts, 2 Seconds (19 points)
Overall Score: 29 points
Tokugawa was perhaps the unluckiest leader amongst the collection of low peace weight AI personalities drawn into this game. Tokugawa is another standard militaristic AI leader characterized by his isolationist bent as he famously refuses to sign Open Borders with his neighbors. Perhaps this was responsible for him being so unpopular in this game as Tokugawa was attacked over and over again in the alternate histories. Japan was invaded 57 times which was only a tiny bit less often than England (59 times) despite Churchill's far higher peace weight. The main culprit here was Genghis Khan who was relentless in attacking Tokugawa although Cyrus and Boudica also frequently found themselves in conflict with him as well. The net result was a situation where Tokugawa essentially had Churchill's bad diplomacy while lacking England's beautiful starting position. This translated into an elimination rate of 70% for Tokugawa which was far higher than anyone was expecting. Tokugawa was only First to Die twice which indicated that he was often softened up by the Mongols and then knocked out by an expanding Churchill at some point in the midgame. On the other hand, Tokugawa had 3 first place finishes and 2 second place finishes in the six games where he wasn't killed outright. Tokugawa has a Scientific flavor to his research which causes him to be decent at teching despite having no economic traits. He actually won Game #17 by surviving a 1 vs 3 war and then nuking Boudica into oblivion to break a stalemate war. I wouldn't exactly say that this was an impressive performance but Tokugawa was relatively competent when he could manage to avoid getting dogpiled.
Kublai Khan of the Mongols
Wars Declared: 40
Wars Declared Upon: 27
Survival Percentage: 45%
Finishes: 3 Firsts, 3 Seconds (21 points)
Overall Score: 26 points
Kublai Khan had by far the best prior track record of any of the leaders drawn into this game but was hobbled by a starting position which was poorly suited to his traits. Kublai found himself with five seafood resources at his capital which synergized poorly with Mongolia's Hunting/Wheel starting techs. He tended to expand into a jungle region which left him with low quality early game cities that lacked both food and production. This wasn't a recipe for success and it wasted most of the early advantage that Kublai typically gets from his Creative + Aggressive traits. Kublai most frequently found himself in conflict with Churchill as it was a near certainty that the two of them would wind up warring eventually. This was also bad news for Kublai since Churchill was the strongest leader on the map and had far superior terrain. Churchill's Protective trait also caused a lot of problems for Kublai and there were a bunch of games where Kublai spent the whole medieval period fruitlessly ramming Keshiks into Protective longbows behind city walls. To his credit though, Kublai still managed to achieve three first and three second place finishes despite his subpar situation. Kublai sometimes was able to take advantage of a Churchill dogpile as in Game #14 and Game #17. More rarely, he shared the same religion with Churchill and the two of them held hands for first and second place as in Game #6 and Game #11. There was also Game #10 where Kublai was simply never attacked and he stayed at peace teching while the rest of the field self-destructed. I think this was a good set of results for Kublai given the underwhelming position that he found himself occupying.
Cyrus of Persia
Wars Declared: 22
Wars Declared Upon: 38
Survival Percentage: 25%
Finishes: 3 Firsts, 1 Second (17 points)
Overall Score: 23 points
Cyrus was kind of the whipping boy amongst the eastern AI leaders as he was attacked again and again in the early game. He was much more likely to be First to Die than anyone else and suffered that fate eight different times. He also had the worst survival rate of anyone in the field while getting knocked out of fully 75% of the alternate histories. Genghis Khan was again the main culprit here and even scored a Turn 80 elimination of Cyrus in Game #11 where the poor Persians were never able to get a metal resource connected. More often, some combination of Temujin and Boudica and Tokugawa worked together to pound Cyrus into the ground. I thought that Cyrus would wind up with the lowest score thanks to all of those eliminations only to see him win three times in the last three games where he survived to the finish. It's unclear how much of this was the result of small sample bias but it does seem that Cyrus had decent odds to win when he wasn't dogpiled during the opening turns. There was a strong correlation between Cyrus being poweful and Genghis Khan being eliminated, with the eastern Mongols causing the most problems for the Persians. Their removal opened up a lot of doors for Cyrus to walk through. He could do well at teching in the few games where he wasn't being invaded constantly and Cyrus won all three of his victories by Spaceship. We saw a strong Cyrus in the actual Game Six which was a rarity in these alternate histories. He could do well when given space to expand but rarely received that kind of opportunity.
Genghis Khan Temujin of the Mongols
Wars Declared: 59
Wars Declared Upon: 22
Survival Percentage: 55%
Finishes: 0 Firsts, 3 Seconds (6 points)
Overall Score: 21 points
The eastern Mongol leader unsurprisingly failed to win any of the alternate histories but stirred up plenty of trouble as usual. Do you even need to look at the data to know who started the most wars in this bunch? Temujin initiated 59 conflicts, significantly more than anyone else, and had the most lopsided ratio of offensive to defensive wars at 59 to 22. He was a total wrecking ball that smashed apart the games of his neighbors over and over again; I don't think it's a coincidence that Tokugawa and Cyrus had the worst survival rates in the field. It was hard to do well when the eastern Mongols were repeatedly rampaging across the border game after game. Temujin was never in a position to win any of these games and had to settle for a handful of second place performances. It wasn't a case of him getting unlucky, he never had the economy to make a serious challenge for a victory. The performance of the eastern Mongols could largely be tracked by how many kills Genghis Khan was able to achieve since that was where he scored basically all of his points. Game #5 was perhaps the most typical of the bunch as Temujin scored three different kills by taking out Cyrus, Boudica, and Churchill... only to commit sucicide and score zero points for finish position by attacking the much more advanced Tokugawa in the extreme lategame. I'm honestly shocked that Temujin survived to the finish in more than half of these games. This guy is a total lunatic and performed exactly the way that everyone expected in the alternate histories.
De Gaulle of France
Wars Declared: 28
Wars Declared Upon: 35
Survival Percentage: 30%
Finishes: 0 Firsts, 4 Seconds (8 points)
Overall Score: 14 points
De Gaulle essentially had the same survival rate and low kill total of Cyrus without any of the first place finishes. He was the very definition of a "blah" leader, performing poorly without being interesting or noteworthy in any particular way. De Gaulle most frequently clashed with Boudica in these games and the Celts typically came out on top in those conflicts. He didn't expand well, or tech well, or fight well, or do much of anything to stand out. De Gaulle's only good performance came in Game #9 where he shared a religion with Boudica and he narrowly lost to her in the space race. But note the date of that Spaceship victory: Turn 382. De Gaulle was only able to challenge for the win because it was such a preposterously late victory date due to everyone having ruined economies from heavy warring. In more normal games, De Gaulle didn't have the research capacity or the conquered territory to compete. I think that De Gaulle has one of the worst AI leader personalities and his poor showing in this game did nothing to change my opinion. De Gaulle didn't have an amazing starting position by any means but surely he could have done more with it than this. In a weak field of opponents and with an excellent diplomatic position he still managed to achieve almost nothing.
Game Six ended up being one of the most unpredictable matches in Season Six. Churchill won on this map roughly half the time but if he didn't win then it was pretty much open season for anyone else to claim the top prize. As I mentioned above, I found it difficult to come up with a clear narrative to explain how this one would play out since the group of Boudica / Tokugawa / Kublai Khan / Cyrus were all able to win at the same rate in the non-Churchill games. My impression was that this was a weak field of leaders and without a strong AI present the door was open for anyone who had diplomatic luck break in their favor. These alternate histories also featured very heavy warring (average number of wars = 13.45) which often slowed the tech pace down to a crawl. It was a bit like watching a low Elo Livestream for League of Legends: lots of action, lots of stuff taking place on screen, no real strategic purpose behind most of it. I guess it was still entertaining in a mindless kind of way.
Thanks as always for reading, I hope you enjoyed this look back at Game Six!