Civ4 AI Survivor Season 6: Game Eight Alternate Histories


Game Eight 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 Eight saw Victoria settle 19 cities without ever being attacked and then run away with an insurmountable economic lead. 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:

Season Six Game Eight

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.)

I knew ahead of time going into the alternate histories for Game Eight that Victoria was going to have a difficult time repeating her economic tour de force. She was somehow never attacked by Montezuma, never attacked by Hannibal, never attacked by anyone as she spent some 200 turns peacefully settling NINETEEN cities in extremely aggressive locations across the map and then sat back and relied on her Financial trait to claim an easy Spaceship victory. Needless to say, her path was nowhere near as smooth in the alternate histories. In contrast to the real Game Eight where she was never attacked even one time, Victoria wound up tying with Montezuma for the most defensive wars faced at 44 and this was responsible for her having a survival rate just under 50 percent. The biggest change here was her relationship with Wang Kon; rather than cooperating in peaceful harmony, the two of them wound up fighting in nearly every game. It was common for the two of them to wind up with different religions and even when they shared the same faith their immense border tensions were often enough to drive them into warfare. The net result was Victoria being unable to repeat her peaceful REXing and she only claimed two victories, neither of them anywhere near as dominant. Victoria turned out to be a solid performer on this map but nowhere close to one of the favorites.

Instead the alternate histories for Game Eight were completely dominated by two leaders: Justinian and Darius. The Byzantine leader was a heavy favorite in the picking contest and the alternate histories confirmed that the community was correct as Justinian won 9 times in 20 games while also scoring the most kills. Justinian suffered from bad luck in the actual Game Eight as his self-founded religion failed to spread to anyone else and then he was blindsided by an invasion from the incredibly-advanced Victoria which knocked him out of a likely second place finish. On the repeats of this map Justinian fared much better and he was always sitting at the top of the scoreboard coming out of the landgrab phase. Zara Yaqob almost always picked up Justinian's religion and the two of them paired together as allies in many of these games (thus resulting in Zara having the most Runner Up finishes). But it wasn't uncommon for Justinian's religion to spread further afield than that and sometime he had as many as three or four other leaders practicing his Christianity. Justinian was simply unlucky in terms of religious spread in the game that we watched.

Justinian's success often seemed to be tied to where the barbarian cities sprang up on the map. Sometimes they would appear right along his northern borders and that would slow down the expansion of Byzantium across the map; this is of course what happened in the actual Game Eight as can be seen in the overview screenshot from Turn 50. However, there were plenty of other games where the barbs appeared elsewhere and in those cases Justinian was able to push further north and west before the warring began. Sometimes he was 300 or 400 points clear of anyone else on the scoreboard at the end of the landgrab phase purely off of his expansion. Justinian also usually benefitted from strong local diplomacy: Montezuma and Zara nearly always went to war and similarly Victoria and Wang Kon usually fought. That allowed Justinian to jump into a war of his choosing on the friendly 2 vs 1 side and acquire more territory for himself. Many of Justinian's best games saw him overrun the weak Wang Kon and then snowball from there with double the territory of anyone else. Justinian tended to falter when diplomacy occasionally put him into two front wars or when he simply couldn't keep pace economically with one of the Financial leaders. But he won about 50 percent of the time and it was nearly always by Domination which should give readers some idea of the Byzantine ass-kickings that we didn't see in the Livestream game.

The other top leader on this map was Darius who pursued a completely different strategy. Rather than seeking to conquer the whole map, Darius simply capitalized on the weakness of Hannibal in game after game and then turtled on their combined territory while racing to space. One of the major questions that I had for the alternate histories was whether Hannibal would repeat his disastrous jungle-choked settler build from the actual Game Eight and I can confirm that yes, this was indeed a pattern that repeated. Although Hannibal occasionally settled Utica somewhere else, he usually picked that same spot something like 80 percent of the time and crippled his civ's development. This was not something that I anticipated when designing the map and it had a massive effect on how the western part of the continent played out. Darius inevitably conquered Hannibal in basically every game but how well Darius finished was largely tied to how quickly he could achieve that conquest. For example, in Game #13 Darius eliminated Hannibal on Turn 113 and cakewalked to victory from there. By contrast, in Game #7 the elimination didn't take place until Turn 223 and by that point Justinian was unstoppably strong leading to a Byzantine victory. The faster Darius could take control of Carthaginian territory, the better his odds were of winning the game. This did not mean that Darius suddenly became a dynamic leader though, as he basically sat back and took no further steps to develop his position after eliminating Hannibal. He simply won a bunch of games because having Financial/Organized traits and two nations worth of territory was enough to launch an early spaceship. Nothing about this setup changed my opinion of Darius: it was the starting position that was strong, not the leader himself.

As for the other leaders, Zara largely played the role of Robin to someone else's Batman by claiming the most second place finishes. He paired most frequently with Justinian as a shared religious duo although sometimes he also finished in second behind one of the Financial leaders. Zara's biggest problem was insane lunatic Montezuma next door who kept attacking him in every game; this tended to slow down Zara's development enough that he wasn't a serious contender for first place. Only in Game #9 where Monty spent the whole early and midgame fighting Victoria was Zara able to come out on top. Wang Kon and Hannibal were the minnows of this game: almost always at the bottom of the scoreboard, both with high elimination rates, each of them with only a single top two finish. Wang Kon was pulverized by heavy warring and seemed to lack much room to expand. His single victory in Game #17 only took place because he managed to avoid the lategame wars while everyone else crashed their economies with too much fighting. Hannibal was simply hopeless on this map thanks to that jungle start as he was First to Die 11 different times and had an elimination rate of 90 percent. I'm honestly shocked that he had a single Runner Up finish in Game #10 thanks to a truly embarassing performance by Darius.

And then there was Montezuma. This game may have shown the deficiencies of Montezuma as an AI leader more so than any previous iteration of AI Survivor. In the actual Game Eight, Monty suffered from unlucky diplomacy as he was invaded by three different leaders and quickly knocked out of the contest. This was highly unlikely though since Montezuma was only First to Die in a single repetition of this map. The Aztecs genuinely had strong territory at their starting position and Montezuma often came out of the landgrab phase as one of the top leaders on the scoreboard... only to inevitably crash and burn later on thanks to his truly putrid economy. I've watched hundreds and hundreds of these AI games over the years and I cannot overstate how poorly Montezuma ran his economy in these alternate histories. His AI personality essentially built nothing but units in every city, forever, and therefore the Aztecs lacked the infrastructure needed to be competitive in the long term. After about four or five of these games it became clear that Montezuma would never, ever, ever win a game on this map because his economy was just too bad. The Aztecs ballooned up to massive size in several of these games and it didn't matter because Monty would be fielding rifles when his top competitors were driving around in tanks and mechs. This is the reason why Montezuma had so many eliminations in the Turn 200-300 range as his abysmal teching would squander promising starts again and again. It was absolutely the classic behavior that we've come to expect from Montezuma with insane aggression, terrible diplomacy, and cover-your-eyes awful research. What a clown show.

Now for a look at the individual leaders:

Leader Summaries

Justinian of Byzantium
Wars Declared: 54
Wars Declared Upon: 36
Survival Percentage: 70%
Finishes: 9 Firsts, 3 Seconds (51 points)
Kills: 24
Overall Score: 75 points

Justinian showed us once again why he remains one of the best leaders for AI Survivor purposes. He was significantly stronger in the alternate histories as it turned out that we saw one of his weakest performances in the Livestream game. Justinian had the most victories, the highest kill total, and narrowly lost out on having the best survival percentage to Zara. He started the most offensive wars and won eight of his nine victories by Domination, a sign that Justinian was aggressively pushing his advantage whenever he pulled ahead. Once again the combination of Imperalistic expansion together with a self-founded religion to make diplomatic allies proved to be successful. Justinian was on pace to produce one of the truly rare 100+ point alternate history scores for a while before he cooled off and "only" managed two firsts and two seconds over the last seven games. This was an incredibly dangerous leader who was always in contention for the win on every replay of the map. I have no doubt that Justinian will be a popular pick once again whenever we do the next season of this competition.

Darius of Persia
Wars Declared: 32
Wars Declared Upon: 28
Survival Percentage: 70%
Finishes: 7 Firsts, 3 Seconds (41 points)
Kills: 19
Overall Score: 60 points

Darius won nearly as often as Justinian and accumulated almost as many kills even though his style of gameplay was completely different. Darius is a pacifistic AI leader who prefers to sit back and tech his way to a Spaceship ending; this is why he only found himself involved in 60 total wars as opposed to the 90 wars that Justinian fought. Unlike a lot of the other peaceful AI leaders, Darius has no interest in chasing after a Cultural victory condition and indeed none of these games saw that victory type. Darius was extremely lucky in this game to find himself starting on a western subcontinent with the hopelessly crippled Hannibal and the two of them inevitably ended up fighting in every map repetition. Darius is a poor warrior but he didn't have to do much to defeat Hannibal. Eventually his Financial/Organized traits would pull him ahead in tech and he would conquer Hannibal's territory for himself. This is how Darius wound up with 19 kills on the game as nearly all of them were scored against his northern neighbor. Darius made few attempts to expand outside of the western subcontinent and even in his victories he largely remained confined to that same portion of the map. Only in Game #2 and Game #5 did he manage to acquire any kind of territory elsewhere. Sitting back and turtling normally doesn't produce a lot of victories in AI Survivor but Darius was in such a strong position that he was able to win repeatedly with this tactic. Generally speaking, if Justinian wasn't able to snowball over the rest of the map militarily, the ticking clock of Persian research would eventually deliver Darius a Spaceship victory. It was an unsatisfying and undramatic way to win games but it did repeatedly work for Darius.

Victoria of England
Wars Declared: 33
Wars Declared Upon: 44
Survival Percentage: 45%
Finishes: 2 Firsts, 4 Seconds (18 points)
Kills: 7
Overall Score: 25 points

There was a massive dropoff from Justinian and Darius to the remaining five leaders in the field as everyone else combined could only manage four victories between them. Victoria and Zara graded out very similarly under our scoring system, with Victoria more likely to win individual matches but also significantly more likely to be eliminated. Victoria essentially tried to play the same game as Darius by pushing for pure economy while lacking the sheltered starting position of the Persians. Victoria sometimes could get some breathing space to push out Imperialistic settlers, however she never found herself landing the dream scenario of the actual Game Eight in the same way that we watched on Livestream. More commonly she found herself tied up in warring with Wang Kon or Montezuma or Justinian. If she could managed to avoid two front wars and get some freedom to concentrate on teching, Victoria could prove to be quite successful. She survived to the finish in nine games and she was first or second in six of those matches. However, the fatal problem was the fact that she only survived to the finish in nine game which meant an awful lot of eliminations at the hands of her militaristic neighbors. I think that Victoria's score is slightly understated here since we know that she did win the actual Game Eight and she also *TIED* for a Spaceship victory with Darius in Game #11 when their spaceships reached Alpha Centauri on the same turn and Darius won by virtue of player order. (This would have caused a meltdown in the chat if it happened in an actual game and I'm sorry we missed it!) Anyway, I think that Victoria's true strength was probably a little bit higher than Zara's if we could run an unlimited number of these test cases. But she was clearly nowhere near Justinian and Darius in terms of power and the alternative histories confirmed that the outcome of Game Eight was fairly improbable.

Zara Yaqob of Ethiopia
Wars Declared: 34
Wars Declared Upon: 26
Survival Percentage: 75%
Finishes: 1 First, 6 Seconds (17 points)
Kills: 6
Overall Score: 23 points

Zara was the most popular choice as First to Die in the picking contest leading up to this game, probably because of the way he was blitzed out of the previous season by Alexander. Amusingly, he was not only never the First to Die in any of the alternate histories, he actually had the highest survival rate of any leader. Whoops. There was a feeling that Zara would run afoul of Montezuma and the community certainly wasn't wrong about that as the Aztecs attacked in virtually every game. However, Zara always seemed to be able to connect the iron resource at his capital quickly enough to defend himself against Montezuma's inevitable incursions and stall out those wars into stalemate. The constant attacks from Monty weren't enough to eliminate Zara but they did slow him down enough that he was almost never able to win a game in his own right. Instead, Zara usually partnered together with Justinian over shared religious faith and they finished first and second four different times (with Justinian always in the victor's seat). Zara's best result took place in Game #9 where Montezuma left him alone to spend the whole game fighting Victoria while Darius inexplicably charged across the map to invade Justinian. With his neighbors stuck in unprofitable wars, Zara was able to research into a tech lead and then conquer Montezuma to snowball the game. This was a true rarity though and more often Zara found himself playing second fiddle. He was sheltered enough on the east side of the map that he rarely died but also rarely could get over the hump into true contention. (One random side note: Zara died in Game #10 when his last city culture flipped away to Justinian - and yes, that can eliminate a civilization in Civ4!)

Montezuma of the Aztecs
Wars Declared: 47
Wars Declared Upon: 44
Survival Percentage: 30%
Finishes: 0 Firsts, 3 Seconds (6 points)
Kills: 11
Overall Score: 17 points

As bad as he was on this map, Montezuma still ended up coming in fifth place largely thanks to picking up almost a dozen kills. We've seen this pattern play out in other alternate histories with leaders like Ragnar and Genghis Khan Temujin, as their insatiable bloodlust gets them into a tremendous number of wars and therefore leads to at least a few points for eliminating their rivals. It's not enough to get these madmen anywhere close to an actual victory, mind you, but it does pick up a few random points. This was one of the best starting positions that we've seen for Montezuma, a lush river valley with multiple happiness resources and copper in the immediate vicinity. The stronger AI leaders likely would have been able to turn this position into multiple victories. Instead, Montezuma frittered away his advantages with his usual pattern of too many wars and some truly inept research. It wasn't just that he fared poorly in comparison to the Financial leaders, Monty was routinely far behind his neighbors Justinian and Zara as well. The Aztecs did grow quite large on several occasions, either through conquering Zara or Victoria, yet still ended up getting popped like a balloon due to their awful research. Monty was able to take second place behind Justinian several times thanks to a religious alliance (in the rare games where the Aztecs didn't found their own religion) but had no shot to win due to technological backwardness. I doubt that Monty would ever be able to conquer enough territory to win by Domination before his atrocious research allowed someone else to claim the victory. This was basically the best-case scenario for Montezuma and he was still far, far away from winning.

Wang Kon of Korea
Wars Declared: 19
Wars Declared Upon: 40
Survival Percentage: 35%
Finishes: 1 First, 0 Seconds (5 points)
Kills: 6
Overall Score: 11 points

Wang Kon was somewhat surprisingly the most peaceful leader in the alternate histories with only 19 offensive wars initiated. He was essentially an even weaker version of the Darius/Victoria strategy of attempting to push for the spaceship by focusing on the Financial trait and lacked the resources needed to pull it off. Wang Kon tended to find himself caught up in wars with Victoria or Justinian or both and his neighbors simply had stronger starting positions. Even assuming that Wang Kon could survive without being invaded, his strategy of going pure economy tended to lose out to one of the other Financial leaders chasing after the same target. It was only in Game #17 where Justinian collapsed from endless warring and Zara slowly conquered Victoria that Wang Kon could put together his own victory, aided by acquiring half of Hannibal's territory in earlier fighting. More commonly, Wang Kon was one of the weakest leaders in the field and suffered a number of early eliminations. Outside of that one flukey victory, Wang Kon's scoring was based on half a dozen random kills that he picked up here and there. He didn't have a particularly great starting position and the value of having Financial trait was greatly diminished in this game where three other leaders had the same advantage.

Hannibal of Carthage
Wars Declared: 23
Wars Declared Upon: 24
Survival Percentage: 10%
Finishes: 0 Firstz, 1 Second (2 points)
Kills: 0
Overall Score: 2 points

Hannibal had no chance at all in this game thanks to the AI deciding to pick that jungle spot for the second city, then refusing to train another settler until the first one was laboriously constructed over 90 turns. I had no idea that this would happen ahead of time and would have made map edits to prevent such a dud result from taking place. Even with this monstrous handicap, Hannibal was often able to frustrate Darius for long periods of time as we watched on Livestream. Sometimes Darius would run him over quickly but there were also times where it took more than 200 turns for the Carthaginians to submit. Nevertheless, Hannibal basically always lost that struggle because he had such a crippling disadvantage coming out of the gate. He only survived to the finish in two games although incredibly one of them was a Runner Up finish in Game #10. Hannibal was actually beating Darius in that game and would have eliminated the Persians eventually if the match had run on long enough. This was such a weird situation that it's impossible to draw any larger conclusions about Hannibal as an AI leader based on what we saw here. I apologize for screwing up the map - I didn't expect Hannibal AI to pick that particular spot for his second city!


Game Eight turned out to be highly atypical in the sense that neither Justinian nor Darius was the winner. Along with Montezuma being First to Die instead of the vastly more likely Hannibal, we had a true oddball result that didn't match how the scenario usually played out. This was also apparent in the number of wars that took place: we saw 7 of them in the actual Game Eight while the median value was 12 wars from the alternate histories. The warring total was more than two standard deviations outside the norm, a true outlier result and fantastic news for Victoria. Season Six of AI Survivor had quite a few of these unusual results in the opening round: Pericles not winning Game One, Napoleon not winning Game Two, Gilgamesh not even finishing top two in Game Three, and so on. The back half of the opening round went more according to form (Catherine, Bismarck, Churchill, and Mansa Musa were all confirmed as expected winners) but there's a reason why so many of the picks were all over the place in the contest. I'm looking forward to seeing the results for the Wildcard and playoff games where the favorites were much more dominant and the contest scoring picked up markedly.

Thanks as always for reading, I hope you enjoyed this look back at Game Eight!