Civ4 AI Survivor Season 6: Game Two Alternate Histories


Introduction

Game Two 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 Two was expected to be a Huayna Capac stomp but instead turned into a Pacal runaway with lots of backstabbing on the part of the game's two French leaders. 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 Two

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.)

It turned out that we saw a highly unusual variant of Game Two as compared with how the map usually played out. Pacal was not the dominant leader on this map; not only did he fail to win a single time in the 20 map replays, he failed to score so much as a single point! There was a bit of poor luck in Pacal's zero score since he was in position to pick up a few kills and simply missed out on the finishing city capture to another leader. Nevertheless, Pacal turned out to be a dud leader on this map who was never in position to claim a victory in any of the alternate histories. He was eliminated fully 75% of the time and wound up being First to Die more than anyone else, albeit by a narrow margin. Repeated replays of the map revealed that Pacal's peaceful initial 200 turns in the actual Game Two had been extremely unusual. He was nowhere near as lucky in the alternate histories and I don't think he had a single game where he was able to avoid wars for such a long period of time. It was basically a complete fluke that Pacal was able to sit around building in peace while everyone else fought one another, and outside of that bizarrely fortunate bit of luck, he was one of the weakest AI leaders on this map.

Instead it was shockingly Napoleon who was the true dominant leader on this map. Napoleon won 12 times in 20 games and racked up a staggering 43 kills, the most that I had seen to date in any of the alternate histories that I've run. While he didn't win as often as Willem back in Season Five Game Eight, the French leader was able to score even more kills thanks to his relentless, neverending aggression. Napoleon benefited from having several weak neighbors that he could devour and then use their captured territory to snowball ahead. His best route to success was going to war with Pacal and rolling through the Mayan domains. This was simultaneously why Pacal had so many eliminations and why Napoleon took first place so often; I repeatedly wrote "Napoleon solo kills Pacal and snowballs to victory" in my notes while watching these matches. Napoleon heavily emphasized military technology while Pacal did not and there were a bunch of games where Napoleon hit key timing windows on the tech tree to maximum effect. He was able to attack Pacal with elephants/catapults from early Construction at a time when Pacal lacked longbows and many readers will know from Civ4 Single Player that this is a particularly good moment to launch an offensive. Solo conquest of the Mayans put Napoleon in a position where he was virtually locked into a top two finish and he could keep fighting and fighting from a position of advantage.

Napoleon's other juicy target was Suleiman and he often used a conquest of the weak Ottomans to put him into a leading role. This wasn't as successful since Suleiman's jungle-heavy territory was smaller and more crushed by nearby Incan culture but it could and did work in a number of games. The one target where Napoleon's aggression didn't work as well was Hatshepsut since any conquests made to the south stretched France's territory into a weird position and resulted in lots of culturally-crushed cities. Naturally this was the direction where he headed in the actual Game Two and most of Napoleon's bad outcomes flowed from an attack to the south. Napoleon was almost always one of the strongest AI leaders no matter what he did though, with all of his eliminations coming in the lategame and with a bunch of kills even in the games where he was knocked out. I think that his starting position wound up being a lot better than we realized before the game:

The French core had an irrigated wheat that the Agriculture/Wheel French workers could connect immediately, then a plains cow for production and a grassland gems tile for research. Napoleon always sent his initial Deity settler to the south where he claimed an irrigated corn, horses, another grassland cow, and ivory. He would always pick up the nearby copper resource to have access to all of the early game strategic resources, multiple luxuries, lots of high food tiles, and a river + gems tile for lots of early commerce. Napoleon was able to hit military technologies much earlier than your typical crazed warmonger as a result and then he snowballed off successful early attacks. Only 7% of the picking contest thought that Napoleon would win so this was something that the community completely missed, myself very much included. Napoleon was absolutely the right choice for first place and it was an oddity that Pacal emerged instead as the top AI leader.

The other strong leader on this map was Louis who came in second place 11 different times along with a trio of first place finishes in his own right. Louis benefited from having the most isolated spot on the map where he had room to spread out without too much pressure from his neighbors. He was also a major beneficiary of Napoleon attacking Pacal in so many games, a diplomatic development that opened up more of the north for Louis to settle. A typical game saw Napoleon snowballing ahead first through Pacal and then later via Suleiman/Huayna Capac on the northern side of the map while Louis spread out in the southeast followed by eventually absorbing the unpopular Saladin. He was the classic Runner Up finisher, rarely strong enough to win the game outright while very often sitting in second place on the scoreboard behind Napoleon or less often Huayna Capac. Louis lost a city to the barbarians in the actual Game Two which caused his true strength to be underestimated. He still wound up being one of the top powers despite that rare barbarian setback which should indicate how much better he was when that flukey outcome didn't take place.

The only other leader who had any success in the alternate histories was Huayna Capac and that was mostly because, well, he's Huayna Capac. The Incan starting position turned out to be not that great overall and a non-Huayna leader in that spot would have struggled to win any matches at all. Repeated matches made it clear that there simply wasn't a lot of room for Huayna Capac to expand and this limited his ability to emerge as a top leader. There was a lot of variation in terms of where Suleiman sent his early settlers and in some games Huayna Capac wound up getting stuck on as little as four or five cities. (Huayna Capac didn't help himself here by going wonder-crazy in some of these games.) Even with his amazing traits there wasn't a lot that Huayna Capac could do when stuck on a handful of cities. Sometimes Huayna fought repeated wars with Suleiman and sometimes they worked together as allies, it seemed to be pretty random how their relationship would shake out. The Incans still won four times because of how amazing Huayna Capac is but he needed fortune to break his way and that didn't happen consistently. He won once in the United Nations from a position of parity in Game #6, he was part of repeatedly successful 2 vs 1 coalition wars in Game #13, and he had one really strong performance by solo killing Suleiman in Game #12. Huayna Capac was a lock to win if he could pick up about two civ's amount of territory, however this happened to be a game where it was very difficult for him to pull that off.

As for the remaining four leaders, none of them had a serious chance to win the game and they rarely found themselves in a strong position. Saladin was the only one of the group who scored a single victory and he had a few second place finishes where he ran behind Napoleon. His problem was diplomatic in nature as he was dogpiled over and over again; Saladin was attacked only slightly less often than Hatshepsut for reasons that weren't entire clear to me. I think that Louis didn't have any other good targets for aggression and Saladin found himself pinched between Louis and Hatshepsut and Napoleon. On the few occasions where he wasn't dogpiled he did pretty well but the Arabians found themselves staring down the wrong side of a 1 vs 3 conflict far too often. The other three leaders were in even worse shape; Suleiman's jungle-heavy start didn't leave him with a lot of options and I should have done some more tinkering to make that location more viable. He did manage a pair of second place finishes but the odds were too stacked against the Ottomans being a real player in this game. Hatshepsut had a lot of excellent openings and sat near the top of the scoreboard on Turn 100 multiple times. Then her high peace weight would inevitably result in her getting attacked from multiple sides and she would die at some point in the midgame. Poor Hatty never surived to the finish of even one game though she wasn't First to Die all that often. Finally, Pacal was largely a punching bag for Napoleon and wasn't able to land even one kill. This just wasn't his game for the reasons detailed above - it's amazing how far he made it this season when he probably should have been out in the opening round.

Now for a look at the individual leaders:

Leader Summaries


Napoleon of France
Wars Declared: 70
Wars Declared Upon: 21
Survival Percentage: 75%
Finishes: 12 Firsts, 2 Seconds (64 points)
Kills: 43
Overall Score: 107 points

I made the case above for why Napoleon had such a good performance on this map: he was almost always able to conquer one of his neighbors and then refused to let up the pressure by attacking again and again from a position of advantage. While there were some games where he wound up in wars against multiple opponents at once or fell too far behind in technology to be effective, these tended to be unusual outcomes that fell outside the norm. One of the eye-popping stats for Napoleon was his number of offensive wars: 70 attacks initiated versus only 21 wars caught defending. This guy absolutely would not stop invading other civs and on this particular map he was strong enough to pull it off. Napoleon's 43 kills were also a record for the alternate histories that I've run thus far as he averaged more than 2 kills per game. That's exceedingly difficult to achieve and Napoleon had at least one kill in 19 out of 20 games. He had as many games with FOUR kills as he had with zero kills. It turns out that a leader with 10/10 unit training emphasis becomes a terrifying sight if he can manage to snowball far enough ahead of the rest of the field. If you ever wanted to know what it would look like if Shaka or Montezuma had the perfect map setup, this is what you would get.


Louis XIV of France
Wars Declared: 51
Wars Declared Upon: 19
Survival Percentage: 80%
Finishes: 3 Firsts, 11 Seconds (37 points)
Kills: 16
Overall Score: 53 points

The other French leader on this map was the perfect second banana to run behind Napoleon, finishing in second almost exactly as many times as Napoleon finished in first. They finished in those exact two spots nine different times in what was by far the most likely outcome for the two playoff positions. Louis was attacked less often than any other leader at only 19 defensive wars and this served him well as he was generally able to pick his spots for aggression and spend lots of time building up in peace when he wasn't looking to fight. Louis also had a highly favorable diplomatic situation, with his northern neighbor Pacal routinely getting smashed by Napoleon and his western neighbor Saladin frequently falling afoul of dogpiles. The net result was Louis having the best survival rate of any leader at 80%, even better than Napoleon since Louis didn't take on nearly as many opponents. This was enough to put him in second place again and again while falling short of outright victory. The one other thing that I'll point out is that success for Louis was inversely correlated with success for Saladin. Louis was only eliminated in four games but Saladin had his lone victory and two runner up finishes in those four games. The map could only have one strong leader in the southeast corner and conditions were ripe for that to be Louis in nearly every game.


Huyna Capac of the Incas
Wars Declared: 41
Wars Declared Upon: 26
Survival Percentage: 60%
Finishes: 4 Firsts, 1 Second (22 points)
Kills: 8
Overall Score: 30 points

As I mentioned above, Huayna Capac was the only other leader who had any consistent chance to win on this map and it was purely due to the fact that he's one of the top leaders in Civ4. His starting position wasn't as great as it seemed before the game due to a lack of space to expand. Most of us who picked Huayna Capac to win the game expected Hatshepsut to collapse immediately and open up more room for Incan cities, and yet that rarely occurred in these games since Hatshepsut typically played a strong opening. Suleiman was also a bit of a wild card for Huayna Capac, sometimes dragging him down with repeated warring as in the actual Game Two and sometimes working together with him to defeat Egypt. The craziest outcome in this regard was Game #6 where Huayna Capac took first and Suleiman was his junior partner in second place. Whether or not they could work together largely came down to the randomness of how the religions spread and sometimes nothing but pure RNG based on who attacked whom. While Huayna Capac's 60% survival rate was significantly better than the four leaders that finished below him, he was still knocked out of many games and often limped to the finish as a weak, irrelevant leader far behind Napoleon and Louis. Much like in the actual Game Two, there were a couple of different times where he was in the process of dying as the match concluded or survived with one or two cities remaining. This was a tough starting position and I suspect that most other leaders would have been lucky to manage a single victory, much less four of them. Huayna Capac shouldn't have been the favorite on this map but he proved yet again that he's always a threat to win in any game.


Saladin of Arabia
Wars Declared: 25
Wars Declared Upon: 58
Survival Percentage: 25%
Finishes: 1 First, 4 Seconds (13 points)
Kills: 8
Overall Score: 21 points

Saladin and Hatshepsut were the two leaders who were most affected by poor diplomacy on this map. Both of them were attacked relentlessly by the other leaders with Hatty getting invaded 63 times and Saladin coming under assault 58 times. Hatshepsut was easy to understand since her peace weight score was by far the highest on this map but Saladin's unpopularity was a bit tougher to unpack. Perhaps he randomly rolled a higher-than-normal peace weight when the map was generated or maybe his position in the middle of the map caused too much border tension with his neighbors. Saladin usually founded a religion but he didn't seem to have much success in making allies via religious diplomacy. Whatever the case, Saladin repeatedly collapsed in 3 way or even 4 way dogpiles as everyone sought to take a bite out of Arabian territory. On the few occasions where he didn't come under multiple atacks he could be pretty strong, with four runner up finishes and a single victory in Game #7. That game was highly atypical as Saladin was the one to solo kill Pacal and then absorb Louis afterwards to become unstoppable. Outside of that match, Saladin's best outcomes tended to take place when he could pick up some land from a partitioning of Egypt and then win a lategame duel with Louis. I mentioned above that only one of Louis/Saladin could have a successful game and whoever won their struggle would typically end up finishing in second place. When the diplomacy wasn't enormously stacked against him, Saladin had a fighting chance to take that second place spot; otherwise he was another casualty on the road to the double-France outcome.


Suleiman of the Ottomans
Wars Declared: 29
Wars Declared Upon: 22
Survival Percentage: 15%
Finishes: 0 Firsts, 2 Seconds (4 points)
Kills: 4
Overall Score: 8 points

Suleiman's fate was pretty easy to understand in this game: he had a rough jungle-choked starting position and it was almost impossible for him to dig out from that initial hole. To his credit, he had a number of games where he expanded well and picked up land that everyone expected to be claimed by Huayna Capac. The Ottomans were often a burr in the side of the Incans and could on rare occasions win the struggle with Huayna Capac for the western side of the map. However, the best outcomes for Suleiman took place when he instead partnered with Huayna Capac to work together and take down Hatshepsut. The two second place finishes for Suleiman in Game #6 and Game #16 both occurred when the pair of them scored an early kill over Egypt, though in Game #16 Suleiman eventually snowballed further ahead by killing the Incans afterwards. Long warring between Suleiman and Huayna Capac generally resulted in both of them becoming irrelevant in the wider game and snapped up by stronger leaders elsewhere on the map. Another huge problem for Suleiman was starting next to Napoleon who was the strongest leader on the map. There were a bunch of games where Suleiman launched ill-fated attacks against Napoleon which resulted in him getting run over by a superior French empire. It was only in the very rare games where Suleiman had a solid landgrab, worked together with Huayna Capac to conquer Hatshepsut, and also avoided ticking off Napoleon that he could have good results. That required a lot of things to line up correctly and the Ottomans weren't that fortunate in most games. This was a weak leader who only finished in fifth place by virtue of two other leaders coming off even worse.


Hatshepsut of Egypt
Wars Declared: 13
Wars Declared Upon: 63
Survival Percentage: 0%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 5
Overall Score: 5 points

Hatshepsut was the only high peace weight leader in a field full of militaristic low peace weight leaders while also starting in the middle of the map. This placed her in an impossible diplomatic situation and she failed to survive to the finish line in a single game. The war declaration counter tells the story here: 13 offensive wars and 63 defensive wars - yikes! That wasn't quite as bad as some of the alternate histories where Gandhi had zero offensive wars and 60+ defensive wars but it was still terrible nonetheless. Hatshepsut had pretty good starting terrain around her capital and she could sometimes emerge from the landgrab phase as one of the stronger leaders, for example in Game #11. Ultimately it didn't matter though as poor Hatty was inevitably dogpiled and her lands partitioned between competing conquerers. She was able to pick up a few kills here and there but that's all that she could do as Egypt was never in a position to claim a top two finish. We've been discussing how Hatshepsut isn't as bad of a leader as past performance in AI Survivor might indicate and she simply needs to get drawn into better fields to be more effective. She definitely isn't the worst despite scoring zero points to date across the first six seasons. However, Hatshepsut's Creative borders have a tendency to annoy her neighbors and her research preferences cause her to ignore military techs to a dangerous degree. Toss in her high peace weight and a lack of Financial trait for economic heft and this is a recipe for trouble. Sooner or later she'll find herself in a better starting situation but it's also not a total coincidence that Hatshepsut has struggled this much to date.


Pacal of the Mayans
Wars Declared: 14
Wars Declared Upon: 34
Survival Percentage: 25%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 0
Overall Score: 0 points

The actual winner of Game Two did precisely nothing across the 20 map replays and finished with a humiliating zero points. While he was a bit unlucky not to score any kills, Pacal was also sluggish and passive in his gameplay. He almost never initiated wars on his own and didn't take advantage of favorable diplomatic environments to snowball his civ ahead with additional land from conquests. Pacal was largely content to sit in his corner of the map and build endlessly; this worked out in the real Game Two because everyone left him alone while fighting mutually destructive wars. But it wasn't a strategy for sustained success and Pacal's nation usually collapsed when it came under attack from larger or more aggressive neighbors. Napoleon was the main culprit as he killed Pacal about ten different times across the alternate histories, then used the Mayan territory to dominate the rest of the field. Pacal was slow to research military techs and Napoleon just crushed him with elephants and catapults over and over again. Less frequently it was Louis or Saladin who attacked Pacal but someone pretty much always came calling. He was not able to tech away undisturbed in any of the alternate histories and the actual Game Two looks like a major outlier in that regard. This was a good map for Pacal and he played it very poorly by not being more proactive.

Conclusions

This was one of the more stunning results from running the alternate histories. I expected that Pacal wouldn't be quite so dominant on the map replays but his zero point outcome was pretty crazy. My expectation was that Huayna Capac would be more dominant in games where he wasn't getting backstabbed constantly and yet that wasn't the case either. I really did not see the Napoleon/Louis combo steamrolling the map to the degree that they did. Obviously both leaders were strong in the actual Game Two only to be overshadowed by Pacal thanks to the monstrous tech advantage held by the Mayans. It turned out that Napoleon/Louis weren't the outlier in that game, it was Pacal who was unexpectedly powerful! This had massive ramifications for the rest of the tournament as Pacal took his first place finish and cruised into the playoffs and then into the championship game. However, in truth he had about 75% odds to be eliminated in the opening round and essentially no chance at all to take a top two spot; at best he would have needed to advance via the Wildcard game. We would have had a completely different season of AI Survivor if this had been a more typical game. (Of course, Huayna Capac was also eliminated 40% of the time in the alternate histories and knocking him out of the tournament here would also have produced a vastly different season.)

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