Civ4 AI Survivor Season 5: Game Seven Alternate Histories


Introduction

Game Seven 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 Seven featured one of the all-time most dominant performances in AI Survivor history, a triumphant conquest for Alexander similar to his romp through the real-world Persian empire. 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 Seasons Three and Four of AI Survivor, I had gone back and investigated some of the previous 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 specific 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 5 seasons, we'll have a 60+ games 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 Five Game Seven

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 Seven's big question was whether or not Alexander's lightning conquest over Zara was an outlier result or something predetermined to take place. The answer to that question very quickly became obvious on repeated playthroughs: Ethiopia was doomed. Zara's demise was not an isolated incident but rather something that happened over and over and over again. Alex killed off Zara in almost every single game, before Turn 100 fully half of the time, and poor Zara wound up being First to Die an astonishing 17 times in 20 games. This was even worse than the hopeless position of Gandhi in Season Five's Playoff Game Three where the Indian peacenik only wound up being the first eliminated 16 times in 20 games. These deaths from Zara were also shockingly fast in nature, with half a dozen of them even beating out the Turn 80 elimination in the actual Game Seven (which itself is currently the record for fastest-ever elimination in AI Survivor history). Alex killed Zara on Turn 75, Turn 67, Turn 80 (again), Turn 72, Turn 77, Turn 79, Turn 66, and the unbelivable date of Turn 60 in Game #9:

That's a picture of what it looks like for one of the AI civs to get knocked out in 1600 BC. Half of the map hadn't even been settled yet, sheesh! Although there were a handful of games where Zara managed to last into the midgame (and two unusual games where he made it all the way to the finish), the rapid exit of Ethiopia was one of the defining features of this scenario. It influenced everything else that took place in these matches and it was the biggest reason why Peter and Alexander overwhelmingly graded out as the top leaders. In game after game, Alex would eliminate Zara during the early turns and then work together with Peter to partition Churchill's lands. The unfortunate English leader also had a truly awful outing on this map and looked good only in comparison to Zara. If there was a "Second to Die" category, Churchill would have won it in nearly every game. The pattern of Ethiopia disappearing first followed by England disappearing second probably took place 15 times in these 20 games, it was extraordinarily common. The Game Seven that we watched was completely typical in this respect.

As far as who actually won the game, the results were split pretty evenly between Peter and Alexander. They dominated the victory screen between them with 9 wins for Peter and 7 wins for Alexander. Based on watching the games, I think that this slightly overemphasizes Peter's actual strength and slightly underestimates Alex; the Greek leader had some bad RNG in terms of war declarations that cost him victories while Peter pulled out two victories that he really shouldn't have won in Game #15 and Game #16. My sense is that they were about equally likely to take first place. The two leaders had different paths to victory, however; Alexander could only win in exactly one fashion by snowballing a series of conquests and running over the whole map. Alex had seven victories and all seven of them came through Domination. Peter was more versatile than the Greek warmonger, scoring a number of Spaceship victories thanks to his role as the best techer in this field. Peter absolutely could get out to a military lead and run the Domination playbook but he had other options to fall back upon if that didn't work out. Alexander by contrast was a total one-trick pony who had no idea how to do anything other than attack.

This difference was reflected in the scores for the two leaders. Peter largely ended up with a higher total score than Alex thanks to landing a whole bunch of second place finishes. Alex had seven firsts but only a single second place finish, reflecting his "first or die" mindset. Peter also survived to the end of the game 90% of the time while Alex was eliminated fully half of the time. This was no accident but rather a direct consequence of their AI personalities. Alex would fight, and fight, and then fight some more. If he managed to plunder enough territory and population to get over the hump, he could become unstoppable and keep going until hitting the Domination limit. But it was also easy for him to get stuck in unproductive wars and fall behind in tech, eventually getting eliminated by other leaders who had reached more advanced military units. This game did a nice job of highlighting the weaknesses inherent in the hyper-aggressive AI leaders like Alexander and Shaka and Genghis Khan. Even with absolutely ideal map conditions, Alex still flamed out a disproportionate amount of the time with his endless warring.

As far as the other remaining leaders, Qin Shi Huang put up a very solid performance in the third place spot. He was well behind Peter and Alex with only three wins but quietly accumulated the most second place finishes of the bunch. Qin's best games took place when he was able to eliminate Justinian and absorb the Byzantine territory on the west side of the map. The Chinese could also make headway by conquering Churchill to the north - the one problem was that Peter and Alexander usually devoured Churchill before Qin could get in on the action! He proved to be a fairly good researcher despite having poor economic traits and was usually hanging around in second or third place on the scoreboard looking for an opportunity to make a move. Again, the word I keep coming back to for Qin is "solid", not great and not terrible, a real workmanlike performance. Qin had about 50/50 odds to make the playoffs based on these alternate history scenarios which feels like a good showing for him.

Finally, the heavy pregame favorite Justinian was mostly a nonentity on this map just as he had been in the actual Game Seven. It's easy to forget this but Justinian was the pick to win the game from 75% of the contest entrants, a major misread for the AI Survivor community (myself included!) I think that we may have been thrown off when Huayna Capac won the previous Game Six from a similarly underwhelming start, something that proved to be a bit of an unlikely result when we ran the alternate histories for that match. Anyway, Justinian just didn't manage to accumulate enough territory to be competitive in these games. His seafood start was a poor fit for Justinian's starting techs and the Byzantine leader kept running his Deity starting settler into the tundra in every single game. Justinian was always squeezed on land and he was nearly always sitting at the bottom of the scoreboard. He did manage to win by Spaceship in the very atypical Game #8 (where Justinian conquered Peter and Alexander stalled out with long unproductive wars) but that was an outlier result to be sure. Justinian mostly hung around in the southwest corner of the map and tried to avoid becoming the target of the more powerful leaders. He was isolated enough to get a series of distant, trailing second place finishes but this was very much an unimpressive performance. Even a strong leader like Justinian couldn't do much with this position.

Now for a look at the individual leaders:

Leader Summaries


Peter of Russia
Wars Declared: 57
Wars Declared Upon: 23
Survival Percentage: 90%
Finishes: 9 Firsts, 6 Seconds (57 points)
Kills: 23
Overall Score: 80 points

Peter was the most well-rounded leader of this group and deservedly graded out as the top performer on this map. He had the most wins (even if a few of them were somewhat unlikely), the best survival percentage by a wide margin, and the most top two finishes. Peter almost always finished in either first or second and was a near-lock to move on to the playoff round. He was helped by having a low peace weight on a map where the only two high peace weight leaders were total sitting ducks, making it easy for Peter to work together with Alex and Qin against their common foes. Peter also had the good luck to be drawn into a field of AI leaders with poor economic traits and little interest in infrastructure. He wound up being the best researcher largely due to process of elimination: Justinian was too small, Alex was too insane, and Qin's traits were even worse than Peter's Expansive/Philosophical pairing. This allowed Peter to launch the first Spaceship and achieve some unexpected victories even in games where other leaders had larger empires.

Peter's path to victory most typically came from conquering Churchill, either solo or together with Alex, and then looking to make a move in the midgame against Justinian or Qin. Oddly enough, whether Peter won the game was often determined more by what was happening with Alexander than by what he himself was doing. Sometimes Greece would land a quick knockout blow over Ethiopia and simply become unstoppable, smashing across the rest of the world map in exactly the fashion that we watched in the real Game Seven. However, it was also quite common for Alex to stall out somewhere along the way, either by taking too long to overcome Zara or by picking a bad invasion target or by simply getting dogpiled. In those scenarios the door was open for Peter and he was savvy enough to walk through it most of the time. I have a whole bunch of notes from the alternate histories saying something along the lines of "Alex finds a way to commit suicide and Peter's teching wins the game for Russia". When Peter didn't win the game, he still managed to hold on for a second place finish on many different occasions. He was tough to eliminate and nearly always sat at the top end of the scoreboard. While Peter isn't one of the best leaders for AI Survivor purposes, he's certainly competent and he took full advantage of a favorable setup to become the top-scoring leader on this map.


Alexander of Greece
Wars Declared: 69
Wars Declared Upon: 24
Survival Percentage: 50%
Finishes: 7 Firsts, 1 Second (37 points)
Kills: 29
Overall Score: 66 points

Alex was by far the most interesting AI leader to watch in this game and made for an incredibly compelling figure. If Game Seven had been an anime, he would have been the protagonist with the crazy spiky hair. This was the rare game where an AI leader with a personality bent on extreme aggression ran smack into ideal conditions to make an early rush possible. Alex had copper within the borders of his capital, the right tech preferences (and Mining tech) to ensure that he would research Bronze Working at an early date, and a high peace weight neighbor to the north with no copper resource available. Zara's Creative borders pushed them into early tensions and Alex usually responded with an incredibly early war declaration. There were multiple games where Alex pulled off the same feat that we watched on Livestream, killing Zara before the Ethiopians even had Archery tech researched. Zara would get stuck with warriors and chariots... against a Greek nation that had a Phalanx unique unit which uniquely countered chariots. Whoops. The result was a plethora of fantastically early eliminations for Zara and the chance for Alex to snowball ahead via a successful conquest. He killed Zara in almost every game, I think 15 or 16 times across these 20 games. I can't ever remember one leader eliminating another leader this consistently across repeated playthroughs of a map. Greece's conquest of Ethiopia was baked into this setup before it even started and almost everyone in the AI Survivor community missed seeing it ahead of time.

Once Zara was dead, Alex would almost always continue onwards with an invasion of Churchill. This was also typically successful as can be seen with Churchill's own 90% elimination rate. However, after that things tended to become a bit tougher for Alex. He had a bad habit of picking wars with Justinian on the other side of the map, conquests that often yielded little of value since the cities would be swallowed up by Chinese or Russian culture. It was also commonplace for Alex to find himself in wars with multiple opponents at once, with his insane militarism causing the Greeks to be unpopular with everyone else in the game. The Greek economy similarly tended to be pretty sad since Alex trained endless units instead of building infrastructure in his cities. His tech preferences similarly caused Greece to research military stuff and ignore key developmental techs in favor of churning out more swords and guns. Now when this all worked out, it succeeded in spectacular fashion. Alexander never backdoored his way into an unlikely victory, every single one of his wins came through Domination with blue borders swallowing up the rest of the map. Many of these wins were also impressively early in terms of finishing dates: he won by Domination on Turn 270, Turn 298, Turn 284, and a scorching Turn 253 in the final Game #20. On these occasions when Alex managed to get ahead, he never took his foot off the gas pedal and absolutely annihilated the rest of the field.

But unfortunately this was the only thing that Alexander knew how to do, a one-hit band who was totally lost when his aggression didn't work out. Everything started with the inevitable attack against Ethiopia and if that conquest failed to materialize it never meant good things for Greece. Zara had two games where he survived to the finish and both of them saw Alex eliminated with zero points scored. (Alex had at least one kill in 17 out of 20 games and these were two of the three games where he came up emptyhanded.) Alexander could also stall out to a surprising degree even when he did pull off the early invasion of Ethiopia. In Game #1 Alex found himself mired in a quagmire of a war with China, in Game #9 that amazingly early conquest fell apart with a dogpile of Greece in the midgame, and in Game #18 Alex was brought down by a Russia/China alliance thanks to the two of them sharing a religion. Alex was simply too violent for his own good and this ended up hurting him in many of these matches. He was involved in by far the most wars and impressively scored a gigantic 29 kills so it was never boring to watch Alex but there was enough self-sabotage here that Peter wound up with more total points.


Qin Shi Huang of China
Wars Declared: 32
Wars Declared Upon: 38
Survival Percentage: 55%
Finishes: 3 Firsts, 7 Seconds (29 points)
Kills: 12
Overall Score: 41 points

Qin was the leader who ended up somewhere in the middle for this game, not reaching the heights of Peter and Alexander while remaining far more competitive than the other three laggards. He managed a tough position in the middle of the map much better than Churchill and usually found himself in the role of a second-tier power a little bit behind the leaders. Qin was able to achieve three victories across the replays of this map and narrowly missed out on several others. He actually reached 62% population and land area in Game #16 before coming up just short and getting beaten in the space race by Peter. Qin seemed to get tangled up in more wars with Justinian than anyone else, and the best games for China took place when Byzantium was eliminated or neutralized at an early date. Qin also seemed to pair better with Peter than he did with Alex; there were a lot of games where Peter and Qin took the top two spots working together with a shared religion of some kind. Alexander's path to Domination usually ran through England and China which made it harder for Qin to have a strong finish when Alex was the top dog.

This was a leader who exemplified the concept of being average. Qin had roughly the same amount of offensive and defensive wars, unlike the other leaders who were all hilariously skewed one way or the other. Qin survived to reach the end of the game right around half of the time. Qin was right in the middle of the pack in terms of number of kills scored, well below Peter and Alex but far above the other weaklings. You get the idea. Qin was never First to Die in any of these games and he was strong enough that he wound up with the most second place finishes at seven. That was somewhat of an artifact of RNG since there were several different leaders who had similar runner-up finishes but it also spoke to the fact that Qin was usually in the position of being a secondary leader in terms of strength. I think that this was a pretty good performance for him overall and Qin made the most of his opportunities. He actually had better odds to make the playoff round than Alex, albeit doing so in far less impressive fashion.


Justinian of Byzantium
Wars Declared: 18
Wars Declared Upon: 28
Survival Percentage: 45%
Finishes: 1 First, 5 Seconds (15 points)
Kills: 3
Overall Score: 18 points

Justinian is a good leader for AI Survivor purposes but he did not find himself in a strong position on this particular map. The biggest problem was his capital's seafood-heavy location which synergized poorly with Justinian's starting techs. (I think this would have been a decent place for someone who starts with Fishing tech like Julius Caesar but it was a poor fit for Byzantium's Mysticism/Wheel pairing.) Justinian liked to send his initial Deity settler off into the southern tundra and he never really recovered from there in most games. Even with the Imperialistic trait, Justinian typically found himself locked out of the juicy jungle region to his north and was left to fight uphill battles against China and/or Russia. Justinian's religious emphasis also did him no favors here, as he frequently founded a religion that didn't spread very well and then pulled aggression from the practitioners of other faiths. He didn't have a lot to work with and Justinian typically struggled to have much of an impact on the wider gameplay. While he was never First to Die, Justinian was usually sitting in last place on the scoreboard for long stretches of time after Zara and Churchill had been purged, a weak and bullied civ with few options at hand.

Justinian managed to achieve one victory across these twenty games, a thoroughly unlikely Spaceship win in Game #8. I mentioned above that this game saw a really bizarre outcome where Justinian solo conquered Peter and then slowly limped to a late Turn 371 Spaceship victory as Qin and Churchill engaged in long unproductive wars. The fact that Churchill was still alive and kicking in that game should tell the reader something about how much of an outlier result it was. Outside of that game, I can't recall a single instance of Justinian being in a genuinely strong position. Although he had five runner-up finishes, these were generally games where Justinian backdoored into the second place spot thanks to one of the other contenders getting conquered in an endgame showdown. For examples of this, take a look at Game #3 and Game #4 and Game #14 and Game #19. Justinian was far, far behind in all of these matches and did very little to deserve his second place finish. A better indicator of the weakness of Justinian was his lack of kills, scoring a measly three of them across twenty games. Justinian is no military slouch and if he couldn't do better than three kills then something was very wrong with his Byzantine civ. This was simply a case of a bad match between an AI leader and their starting position. The Season One champ will have to look for a better draw in the next year of competition.


Churchill of England
Wars Declared: 11
Wars Declared Upon: 56
Survival Percentage: 10%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 0
Overall Score: 2 points

Hoo boy, this was a rough scenario for our boy Churchill. It wasn't a historically bad setup like it was for Zara next door, just a routinely bad location for England to be situated in that resulted in Churchill getting pasted game after game. We shouldn't let the horribleness of Ethiopia overshadow how awful things were for Churchill as well. He was eliminated in 18 out of 20 games and failed to score a kill in any of them. Zero kills at all, nada, zilch - yikes. Churchill was in the unenviable position of having low peace weight leaders on all sides, between Peter to the west and Qin to the south and Alex to the east. I mean, technically Zara was located to the east but Ethiopia never lasted for very long in these games. Churchill found himself being the "Second to Die" in almost every match, greasing the wheels for the snowball of either Peter or Alex along the way. You can't have powerful runaway AI leaders without having nations to be run over and Churchill was firmly located on the underside of the steamroller.

There was exactly one game where Churchill did anything of note, the weird Game #8 that I mentioned a couple of times already. With Peter falling to Justinian there was no strong leader with a low peace weight in the west, and over in the east Alexander managed to get himself stuck on the wrong side of a 2 vs 1 war against Churchill and Qin. The two Protective leaders were able to eliminate the Greeks and that left Churchill alive to hang around and collect his one and only runner-up finish. Churchill also survived to the ending date in Game #6 but there was nothing impressive about his finish in that game, clinging to life with a couple of cities remaining, far behind the victors. Otherwise it was a long series of eliminations for Churchill as he wound up First to Die twice while mostly exiting the stage somewhere in the Turn 100 to Turn 175 range. Alexander would attack from one side and Peter from the other side and that would be all she wrote. It was a brutal game overall that Churchill surely would like to forget about as soon as possible.


Zara Yaqob of Ethiopia
Wars Declared: 2
Wars Declared Upon: 26
Survival Percentage: 10%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 1
Overall Score: 1 point

But not as brutal as it was for this guy, holy cow. Zara's situation in Game Seven has to be on the short list for the most thorough ass-kickings in AI Survivor history. He died at the earliest date that we'd ever seen in the real Game Seven and then proved it was no fluke by dying even earlier in many of the alternate histories. I don't want to belabor this point too much but Zara was knocked out *TEN* different times before Turn 100! It's extremely rare for me to write down a double-digit turn number on the First to Die counter and it happened in fully half of these matches. Even that's underselling things because Zara set the record with his Turn 80 elimination on Livestream and then died at a still earlier date in seven of these alternate histories. I would type in the AI autoplay command, get up to grab a drink of water, and Zara would be dead before I sat back down again. Blink and you'll miss him, sheesh. Yes, the map conditions were perfectly suited for Alexander to make an early conquest, but even so, this was really pathetic on Ethiopia's part. Zara did have horses and iron available and his cities were all getting free Creative culture for the extra defensive bonus. There's no excuse for an AI leader to get blasted off the map this often at such an early date.

Even when Zara managed to stalemate Alex's inevitable invasion, he still seemed to find himself as First to Die more often than not. He had late First to Die exits on Turn 153, Turn 155, Turn 194, and even on Turn 282 in the wacky Game #15. This was largely due to Ethiopia being hamstrung and unable to expand thanks to the Greek invasions even when they weren't immediately successful. Zara made it to the finishing date twice, in Game #6 and Game #14, both times as a weakling down near the bottom of the scoreboard. Zara was in the process of dying to Peter in Game #14 when the Russians fortunately hit the Domination limit. Zara had zero top two finishes and only scored a single kill, an unlikely last hit against Qin's final city in the southern tundra in Game #14. I specifically remember this because Zara landed the kill despite having far fewer units in the area than Peter. In other words, Zara was extremely lucky not to get shut out completely and take a zero for the whole series of games. He was utterly annihilated on this map in spectacular fashion and the evidence is really starting to pile up by now that Zara's Season One performance was largely the product of dumb luck.

Conclusions

Some of the alternate histories produce inconclusive results without any clear favorites and suggest that the game in question could have branched off in many different directions. Game Seven was not one of those cases; it seems pretty clear that either Peter or Alex would win in the vast majority of cases, with Qin as a strong secondary leader and the other three AIs as non-factors. Zara and Churchill were walking directly into a buzzsaw on this map and there doesn't seem to have been much that they could have done to avoid their respective dooms. I should also mention here that there were some other unusual aspects to this map, such as the very low number of wars that took place. The average number was about 9 wars per game and none of them went higher than 13. That was partly due to the fact that these tended to be fast games overall, lots of early Dominations achieved by Peter and Alexander. Note as well that there were zero Cultural or Diplomatic finishes on this map, everything concluding with either Domination (12) or Spaceship (8). Future picking contest entrants should try to remember that certain AI leaders have a high chance to pursue culture (like Gilgamesh, Willem, Huayna Capac, etc.) while other fields of AI leaders simply don't care about the victory condition at all. With mostly poor economic leaders present, this wound up being a heavily militaristic scenario.

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