Civ4 AI Survivor Season 5: Playoff Game Two Alternate Histories


Introduction

Playoff 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. My operating theory was that the actual Playoff Game Two that we saw had been a wildly unlikely sequence which was unlikely to occur again. Asoka settling right on top of his neighbors without ever getting attacked felt like a bizarre outcome that wasn't likely to be repeated. 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 Playoff 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.)

The biggest takeaway conclusion was that we had indeed seen a highly unlikely result take place in the actual playoff game. Asoka faced at least one war declaration in every single one of the 20 alternate history scenarios, facing an average of almost three invasions per game. The community was also correct that Asoka was the overwhelming favorite as most likely to be First to Die. The Indian leader was eliminated in 80% (!) of the alternate history scenarios and he was First to Die in half of those games, 10 out of 20 iterations of the map. Asoka being ignored by everyone else felt like an outlier result and it was indeed proven to be an outlier. (To his credit, Asoka did win outright in two of the four games where he wasn't eliminated so he was indeed a threat to win when he wasn't swallowed up at an early date. The problem for Asoka was the fact that he was out of the game before Turn 200 well over half of the time.)

At the other end of the spectrum, the community was also correct that Julius Caesar was the best choice to be the winner of the game. The western Roman leader won nine times in those twenty games and racked up more than double the power ranking scoring points as compared to anyone else. He was an absolute monster on the replays of this map and I had a lot of fun watching those red borders swallow up the other civilizations again and again. More on this in a minute in the individual leader sections. Willem and Suryavarman were about equally likely to take a victory, usually whichever one of them came out ahead in their death struggle on the western side of the continent. Like Asoka, Mao managed to win twice but was rarely in competition for the victory and has to be seen as a weaker option than his low peace weight compatriots. Augustus couldn't manage to win even one game but emerged as the overwhelming favorite for the Runner Up category, finding himself in second place an amazing nine different times. He was isolated enough to hang in there until the end while almost never emerging as the top dog.

Now for a look at the individual leaders:

Leader Summaries


Julius Caesar of Rome
Wars Declared: 64
Wars Declared Upon: 35
Survival Percentage: 75%
Finishes: 9 Firsts, 3 Seconds (51 points)
Kills: 29
Overall Score: 80 points

Julius Caesar was by far the best performing leader on this map and it wasn't particularly close. He had three times as many victories as anyone else, also managed three runner up finishes, survived to the end of the game at a higher percentage than anyone else, and had more than double the kills of the next-closest leader. In fact, it was a close contest for a long time as to whether Caesar would finish the alternate history games with more kills than everyone else combined. He didn't quite hit that mark but came shockingly close (29 kills as compared to 35 kills for everyone else together). Julius Caesar's path to victory usually involved an early attack on Asoka which would snowball him into a position where he could begin a methodical conquest of the rest of the map. The ideal situation for Caesar saw him attack Asoka together with Mao while Suryavarman and Willem were tied up in warring with one another and Augustus did his usual inert thing. Caesar attacked Asoka in a strong majority of games and his best performances almost always unfolded in this fashion. He attacked Willem much less often and tended to do worse in those scenarios, partly because Willem defended better than Asoka and partly because leaving the Asoka/Augustus combo alone tended to be bad news for the low peace weight leaders. Caesar almost never attacked Mao and when he did those were his worst-performing games. Naturally that unlikely result is what played out in the actual playoff game and allowed Asoka to emerge as a major power.

Caesar loved to stir up trouble by starting an amazing 64 wars, far more than any of the other leaders. This is partly due to the fact that he was eliminated less often than the other leaders and was therefore alive for more total turns. Caesar was also attacked a lot when considering how many wars he started, with 35 incoming war declarations faced in total. A huge number of these wars came from his fellow Roman leader Augustus, and I'd estimate that something like 70% of the Augustus war declarations were directed against Julius Caesar. Eastern Rome repeatedly sabotaged western Rome while Caesar was off fighting someone else, and the outcome of the game often turned on whether Julius Caesar could manage to shake off the resulting 1 vs 2 situation. He succeeded more often than not and once Caesar took control of the score lead, he never took his foot off the gas pedal. All nine of Caesar's victories were by Domination as he stomped the life out of his competitors. This is a ruthless AI leader who does an exceptional job of finishing off kills, a far cry from some of those other AI leaders who don't know how to go for the juggular. If Caesar ran over someone else in the midgame it was almost a foregone conclusion that he would be the overall winner.

The most impressive thing about this performance was the fact that Caesar didn't have a very good starting position. In fact, I think there's a good case to be made that this was the weakest capital on the whole map. None of Rome's starting techs corresponded to this rice/cows/ivory position and there was only a single hill tile to make use of Rome's Mining tech. Caesar did not have guaranteed access to iron and in fact Willem would end up settling the western iron in about half of these games. There was room for two or three cities to the north before the land filled up and then Caesar could plant tundra cities that lacked enough food to go beyond size 10 - that was it. Caesar was also located in the center of the map with potential enemies on all sides and two mostly hostile high peace weight leaders working in tandem off to the east. His only option to be relevant in this game was to fight his way out and that was exactly what happened again and again. Whatever brain aneurysm Caesar suffered from in the actual playoff game, I didn't see much sign of it in these alternate history games. Julius Caesar is an exceedingly dangerous AI leader and I'm disappointed that we didn't get to see him in the championship.


Suryavarman of the Khmer
Wars Declared: 45
Wars Declared Upon: 20
Survival Percentage: 55%
Finishes: 3 Firsts, 4 Seconds (23 points)
Kills: 13
Overall Score: 38 points

Suryavarman flamed out without accomplishing much in the actual game that we watched but I agreed with many of the other community members that he was a solid choice for this map. Running the same setup twenty additional times confirmed that Suryavarman was indeed a strong leader who performed well more often than not. He survived to the finish in more than half of the games played and ended up with seven total first or second place finishes. Suryavarman almost always found himself getting locked into early combat with Willem and the outcome of their struggle usually dictated his fate. The two western leaders both came out on top about half the time and this was more or less a true 50/50 proposition that always had massive implications for the rest of the match. Suryavarman performed the best when one of two different things happened: either Julius Caesar or Mao joined him in partitioning Willem, or alternately he was able to take part in a general dogpile of Mao and claim some of the Chinese territory. There were occasional games where Suryavarman managed to crush Willem on his own in a 1 vs 1 scenario and take all of the Dutch spoils for himself; this was a rarity but it was the path to overall victory in Games #6 and #14. More often though, there would be a lengthy stalemate between the two leaders that allowed other civs elsewhere on the map to take over the game.

It was also notable how aggressive Suryavarman was on this map. He initiated 45 wars while only being attacked 20 times, an even bigger offensive disparity than what we saw for Julius Caesar. This seemed to be partly due to the sheltered corner starting position of the Khmer as well as the local neighbors in the area. Willem isn't a terribly aggressive leader so most of the Khmer/Dutch conflicts were started by Suryavarman, while Mao tended to attack Asoka in most games and left the Khmer alone. This was likely the main reason why Suryavarman was declared upon less than any other leader in this game. The Khmer leader also tended to chase after Cultural victories again and again, something that everyone should keep in mind for future picking contests. Even his Domination victory took place in the context of Suryavarman running the culture slider and then tipping over the land threshhold via culturally acquired tiles. Suryavarman ended up running the culture slider something like seven or eight different times in the lategame, and this ended up costing him bigtime in Game #11 where he might have won if he'd continued going down the traditional spaceship path. Overall though, the general community sentiment was correct. Suryarvarman was one of the strongest leaders on this map and he absolutely had a decent shot to take one of the top spots.


Willem of the Netherlands
Wars Declared: 21
Wars Declared Upon: 41
Survival Percentage: 30%
Finishes: 4 Firsts, 2 Seconds (24 points)
Kills: 4
Overall Score: 28 points

The alternate scenarios confirmed the general sentiment that the community has formed about Willem: he's a high variance leader with strong odds to either take first place or get eliminated. Counting the actual Playoff Two game that took place, Willem only survived to the finish 7 times while being eliminated in the other 14 games. However, amongst those seven games where he managed to live, Willem came away with four victories and three runner up finishes. In other words, if he didn't get crushed he took home a podium spot every time. This makes perfect sense since Willem's economic-heavy gameplan will almost always see him gain a tech lead if given enough time, while also putting him in severe danger until he can run away with his superior teching prowess. Willem's well-known refusal to research Rifling tech was on display again in these alternate scenarios, and it was a major factor in him getting wiped out in a number of different matches. He would be ahead in tech and then get smashed by Julius Caesar or Suryavarman or Mao because he was off teching Corporation or Physics or something econ-focused instead of picking up Gunpowder or Rifling. Willem won three of the initial five games that I tested and I thought that maybe we were overlooking a possible juggernaut on this map. But then he was eliminated in nine of the next ten games and that confirmed his high-risk / high-reward strategy that most of us had him pegged for.

As I mentioned in the section on Suryavarman, Willem tended to end up in a conflict with the Khmer in almost every game and how that went typically dictated whether he would be strong or weak. I saw just about every outcome possible from those wars, everything from Suryavarman crushing the Dutch to protracted stalemates to easy Willem victories. There was significantly less conflict between Willem and Julius Caesar, probably because they share exactly the same peace weight value (4) and this tended to keep relations on the friendlier side. That's not to say that they never fought, just that it was much less common than clashes with the Khmer. Willem was not an aggressive leader at all with only 21 war declarations against 41 attacks faced, and the fact that he only picked up four kills despite winning four games indicates how he doesn't have much interest in sweeping across the map. All four of Willem's victories were Cultural in nature and Game #3 was the only time that he even came close to the Domination limit. Despite Willem's many eliminations he wasn't First to Die very often at all - this turned out to be not as good of a pick as the community thought it would be. Anyway, Willem performed more or less exactly as you would expect in these games. He would nearly always win if he was left alone long enough, and in most games his aggressive neighbors didn't give him that opportunity.


Mao Zedong of China
Wars Declared: 51
Wars Declared Upon: 40
Survival Percentage: 50%
Finishes: 2 Firsts, 1 Second (12 points)
Kills: 12
Overall Score: 24 points

Mao was the leader who ended up being the most cramped on this map. It certainly didn't help being a leader with no innate culture-producing abilities stuck in between Creative Khmer on one side and religious-obsessed Asoka on the other side. In the actual playoff game that we watched, Mao responded to this cultural pressure by founding a bunch of tundra fishing villages in the north and then getting dogpiled by three other leaders. That was a highly unusual circumstance that almost never happened in other games. Most of the time, Mao responded to Asoka's intruding cities by declaring war and invading India, often with a great deal of success. The many early eliminations suffered by Asoka in these alternate history scenarios could frequently be attributed to Chinese attacks. The most successful scenarios for Mao saw him attacking and conquering Asoka, often together with Julius Caesar, and then snowballing off those victories into conquests of other targets. Mao was much less successful in the games where he attacked Suryavarman or Willem or Caesar. He was involved in an enormous 91 wars across these 20 games, almost as many as Caesar despite being eliminated far more often than the Roman leader. The net result of all that fighting was a complete mixed bag, as Mao sometimes rose to be powerful off his conquests and sometimes found himself getting crushed by the invading hordes.

We saw a dogpile of Mao in the actual playoff game and I found that this was occasionally repeated in the alternate history scenarios. It wasn't exactly common but it did pop up several more times, with Mao being ganged up upon more often than any of the other leaders. I suspect that peace weight was responsible for this; while Suryavarman had an equally low peace weight score of 1, he was more isolated than Mao and therefore less likely to pull aggression from the eastern high peace weight leaders. The bulk of the war declarations from Augustus were targeted at Julius Caesar but if he wasn't attacking the other Romans he was invading the Chinese. And while Asoka almost never fought anyone, his rare wars almost entirely came against Mao. It was always a bad sign for the Chinese when Asoka was still alive in the midgame, as a strong India and eastern Rome meant trouble for Mao. He also had a tendency to fall apart in the midgame, often coming out of the initial fighting in a solid position and then collapsing afterwards. Mao was frequently sitting in third or fourth place in these games, not the first one to be eliminated but also not a serious threat to win the game. He was looking like one of the worst leaders for a while until rallying for two victories and eight kills in the last eight matches that I ran. Overall Mao was definitely the weakest of the low peace weight leaders on this map. He's a strong enough AI Survivor leader that he was still a factor in this game but it wasn't a great setup for him and he wasn't one of the better choices in the picking contest.


Augustus of Rome
Wars Declared: 34
Wars Declared Upon: 30
Survival Percentage: 50%
Finishes: 0 Firsts, 9 Seconds (18 points)
Kills: 5
Overall Score: 23 points

Sheesh, what I can say about this guy? Augustus is the leader who did the least with what he was given on this map. I think there's a strong case to be made that he had the best overall starting position of the group: a fish resource that hit his Roman starting techs, a plains hill to mine that paired with Rome's other starting tech, a wet wheat, and double Animal Husbandry resources. More importantly Augustus had the most isolated location by a wide margin, with enough space to spread out and establish a core of strong cities in complete peace. Most of the other leaders ran out of room after founding about 5-7 cities while Augustus was easily getting 10 cities uncontested in most scenarios. Some of those were weak tundra locations but there was also a lot of space in the fertile lands to the north. Augustus also benefited significantly from having a friendly Asoka as his only early game neighbor: Asoka *NEVER* declared war on Augustus even one time across the 21 total matches that we watched. Asoka also served as a perfect magnet for aggression, drawing the attacks from Julius Caesar and Mao while allowing Augustus to build away in peace. It's hard to imagine a setup much better than this for an AI leader.

Unfortunately for him, Augustus largely squandered this dream setup and failed to win even one game. He did come close several times but had to settle for being the bridesmaid instead of the bride: a staggering nine runner up finishes in total. In fact, Augustus survived to the end of the game ten times and was the runner up in every one of those games except Game #4 (when he was clinging to life with a single city left at the time the game ended). He was often the score leader coming out of the landgrab phase because he had more territory than anyone else, and Augustus then went on to do very little with the lead that he had been gifted by natural geography. This was a sign of the mediocrity of his AI personality as Augustus is eternally undermined by his conflicting traits. He has Imperialistic for expansion but he also wants to sit back and build wonders with Industrious. He has one of the game's best unique units in praetorians but he's too peaceful to use them most of the time. Augustus wasn't all that peaceful in this game however as he declared war 34 times, most frequently against Julius Caesar and almost always occuring in the midgame. Sometimes these wars hamstrung Caesar by hitting him while he was in the middle of another war and sometimes they served to annoy the game leader while backfiring in spectacular fashion.

The key to success for Augustus was keeping Asoka alive as long as possible. The two of them worked together as a team and an early exit for Asoka almost always meant trouble for Augustus as well. The eastern Roman leader took second place in all four of the games where Asoka survived to the finish, including running second behind Asoka in both of the games that Asoka won. Their shared identical peace weight score (8), inability to declare war at "Pleased" relations, and general pacifistic tendencies made the two of them a perfect pairing. I believe that Augustus only attacked Asoka one time across all of those games, and as already mentioned, Asoka never attacked him at all. The fact that Augustus only managed five kills across the 64 wars that he was involved in is another indication that he lacks a killer instinct. He was much more willing to end a war with a peace treaty than his low peace weight counterparts, even signing treaties for no reason when he was winning conflicts. In the end, the excellent starting position of Augustus was strong enough to land him a bunch of runner up finishes but his AI personality couldn't capitalize with any victories.


Asoka of India
Wars Declared: 5
Wars Declared Upon: 54
Survival Percentage: 20%
Finishes: 2 Firsts, 0 Seconds (10 points)
Kills: 1
Overall Score: 11 points

The picking contest had Asoka identified as the runaway leader as First to Die and the alternate history scenarios proved that the community was correct! Asoka was the punching bag of this world as he was eliminated in 80% of the alternate history scenarios. He was also First to Die in fully half of these games, ten different times in all, suffering eliminations on Turn 177, Turn 179, Turn 135, Turn 124, Turn 103, Turn 109, Turn 164, Turn 175, Turn 154, and finally a blazing Turn 99 in the last match. After not facing a single war declaration at all in the official playoff game, Asoka was attacked 54 different times in the alternate history scenarios, far and away more than any of the other AI leaders. It was even more impressive how many times India was invaded considering how few turns there was an Indian leader alive on the map to be invaded at all. Asoka was attacked almost three times per game on average and he was invaded at least once in every single one of the twenty alternate history games. The fact that he was never attacked in the actual Playoff Two game was therefore an extremely unlikely outlier result, something that appears to have had odds of less than 5%. We thought that this was bizarre at the time and we were completely right.

Asoka died over and over again in these games and was almost always off the board by the midgame. He was hampered by his setup for all of the reasons that the community identified prior to the match: several aggressive low peace weight neighbors, no access to early copper resources, and tech/flavor preferences which ensured that he'd be pumping out culture and creating lots of border tension. It was rare to see a game where Caesar or Mao or both of them together didn't attack Asoka. His best games always involved these leaders getting tied up in wars with other neighbors and leaving him alone. Asoka's two wins in Games #11 and #12 both saw long stalemates in the middle portion of the game where no leaders emerged as powerhouses. These were scenarios where the warmongers leaders were stuck in factional disputes with one another that went nowhere and Asoka could do his passive building thing for long centuries on end. Asoka was never a dominant leader and all of his victories came on limited territory rather than smashing his rivals. To his credit, Asoka won three times in the five matches where he wasn't eliminated (counting the actual playoff result) even if those victories were a bit flukey. He can best be summarized as a poor man's version of Willem, someone who was even more pacifistic but who also had decent odds to win if he somehow wasn't killed during the early stages of the game.

Conclusions

Running the alternate history scenarios proved that the actual result from Playoff Game Two was a low-odds outcome. The low peace weight leaders dominated the scoring by taking home nearly all of the victories, especially Julius Caesar who was a total beast on this map. Asoka was an unlikely, if not completely impossible, leader to emerge with the victory. It could and did happen but certainly not very often. One of the surprising results was the observation that Augustus was the clear best choice for the Runner Up pick, suggesting that an isolated spot far away from the top contenders may be one of the best qualities for that picking contest category. I'm personally disappointed that we won't get to see a more dynamic leader like Julius Caesar or Suryavarman in the championship game but that's the way that things shake out sometimes. Upset wins are a part of any competition. At least we can all rest assured that we're not crazy - Asoka getting completely ignored during his farmer's gambit was *NOT* a common outcome at all!