Civ4 AI Survivor Season 5: Game Six Alternate Histories


Introduction

Game Six Alternate Histories Spreadsheet

One of the recurring features of past seasons of AI Survivor have been our "alternate histories", running additional iterations on the same maps to see if the same events would play out again. Game Six saw the pregame favorite Huayna Capac snowball his way into a runaway Cultural victory at an early date. 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 Six

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 story that emerged from re-running this setup twenty times was the fact that we had *NOT* seen a typical game on Livestream. The match that we watched wasn't completely out of left field, and I did see something pretty similar to the actual competition in Game #2, but on the whole these replays diverged in several key respects. The most important difference was the role played by Boudica as the Celtic leader was almost always far stronger than the also-ran that we experienced on Livestream. We saw Boudica collapse her economy to a horrifying degree for no clear reason, still stuck researching the most basic worker techs some 80 turns into the game. While the Celtic economy was never particularly good in these replays, Boudica never experienced anything remotely like that flameout in the other games. She also engaged in an early war with Mao's China in the actual Game Six which didn't happen even once in the map replays. Boudica's early war targets were always Lincoln (due to peace weight difference) or Montezuma (due to religious difference), never Mao Zedong. Boudica's terrible economy and sputtering war with Mao left the Celts much, much weaker in the actual Game Six than in the replays of this map. It sent Game Six off in a variant direction that failed to capture Boudica's true strength in this setup.

Many other divergences spiraled out as a consequence from the unusually weak Celts in Game Six. Because Boudica and Mao clashed in the early game and hurt each other for no gain, Augustus ended up notably stronger in the real Game Six than in the map replays. Although Augustus did hang around long enough to pick up a series of second place finishes thanks to his remote location on the map, the Romans were rarely a strong contender for the victory and often found themselves eliminated in the midgame or lategame by rival low peace weight leaders. Game #17 and Game #18 were the only times that Augustus managed to grow as stong as he did on Livestream. Instead of the Chinese fighting with the Celts, it was Augustus who almost always clashed with Mao in the map replays. Those two fought constantly and stalemated with one another most of the time. This was a major reason why Augustus and Mao both graded out as second-tier leaders on this map, too much fighting with one another without a clear winner.

Over on the eastern side of the map, there was another gigantic difference that shaped the gameplay: Lincoln survived far longer in Game Six than he did in the map replays. Lincoln was the overwhelming favorite as First to Die in the picking contest and it was a shock when he outlived Montezuma and Boudica with a relatively late Turn 249 elimination. Even a cursory glance at the data above will reveal that this was highly unusual; Lincoln was First to Die ELEVEN different times and rarely made it past Turn 150. He was commonly attacked by Boudica, by Montezuma, by Gilgamesh, and occasionally by Huayna Capac as well. We noted while watching the game that Lincoln failed to secure a source of copper and remarkably managed to settle for iron up in the northern tundra. He was nowhere near as lucky in many of the map replays and often came under attack with no metal resources, resulting in one early elimination after another. Lincoln was a dead man walking on this map and his actual strength was significantly weaker than what we watched on Livestream.

There were a number of knock-on consequences from Lincoln dying so much sooner in these map replays. Boudica and Gilgamesh were the biggest beneficiaries as they were the ones who most frequently snowballed off captured American territory. It's no coincidence that they finished with the highest scores from vulturing Lincoln's land in game after game. Montezuma also experienced a number of strong openings thanks to devouring Lincoln only to collapse in the midgame because, well, it was Montezuma. I honestly thought that Gilgamesh would end up as the top scoring leader and I was somewhat surprised to find that he didn't win as often as Boudica. Gilgamesh seemed to be held back a bit by his corner start; while it resulted in him having the best survival rate and a lot of second place finishes, the Sumerians were also lacking somewhat in conquest opportunities. This was still a very good setup for Gilgamesh but I thought he would do even better thanks to having Lincoln around to snack on.

Finally, there was the winner of Game Six and AI Survivor juggernaut Huayna Capac. I did not pick him to win due to a cramped starting position and then felt foolish when the Incans walked to an early victory. Well, it turned out that Huayna Capac truly was a weaker option than Boudica and Gilgamesh for this scenario. Although he still did manage an impressive five victories, all five of them came by Culture and the Incans were rarely a dominant power even when Huayna Capac manage to achieve wins. I think that Huayna Capac was only leading in score in Game #2 and maybe Game #7 when his Cultural victory triggered, he was behind in all of the others. The weakness of the Incans was evident in their lack of second place finishes and a paltry 4 kills. Huayna Capac often came under attack and was thoroughly bullied in a way that we didn't see in the actual Game Six. He basically hit a gutshot straight draw in the game that we saw (the war declarations lining up such that he could absorb all of Montezuma's land and then all of Lincoln's territory) which rarely repeated itself. If you were skeptical of Huayna Capac's position going into Game Six, you were right to be wary!

Now for a look at the individual leaders:

Leader Summaries


Boudica of the Celts
Wars Declared: 57
Wars Declared Upon: 19
Survival Percentage: 65%
Finishes: 7 Firsts, 5 Seconds (43 points)
Kills: 31
Overall Score: 74 points

Boudica ended up as the top AI leader on this map and it wasn't particularly close. She had the most first place finishes, a strong collection of second place finishes, and by far the most kills of any AI leader. In fact, Boudica had roughly as many kills (31) as all of the other non-Gilgamesh leaders combined (32). This was a sign of the strength of the Celts across game after game, as Boudica was fighting and winning wars even when she didn't end up taking home the victory herself. The Celts were never the First to Die and were almost always one of the contenders at the top of the scoreboard. Boudica was blessed with lots of food resources in her core cities and easy access to copper for early game military power. She leveraged this into one war after another that succeeded more often than not. Boudica had two main targets for her early aggression: Lincoln and Montezuma. She generally did better by warring with Lincoln because the Americans were weaker and often fell apart in general dogpiles. Boudica struggled a bit more with Montezuma due to the Aztec penchant for training non-stop units and this was one of the biggest ways that her games could fall apart. If the Celts could get over the hump and conquer one of their two neighbors, however, the snowball was off and rolling down the hill, often not stopping until hitting the Domination limit.

Boudica's biggest weakness was over-emphasizing military and ending up with a poor economy. She didn't do a very good job of building infrastructure as she typically preferred to train more units over developing her cities. The Celts were frequently behind in technology as compared to the other leaders and had to compensate for this by having more territory from successful conquests. If Boudica couldn't manage to stomp all over someone else it rarely ended well for her. Of course, she usually did notch an early conquest thanks to having such a strong starting position. Domination wins were rare on this map and Boudica was the only leader that managed to achieve them; she won by Domination four different times while also taking home two Spaceship wins and a Diplomatic win for good measure. Boudica did get eliminated about a third of the time which reflected her dangerous position in the middle of the map. It was a high risk / high reward scenario: it made the Celts more likely to win overall but also more likely to get knocked out. Overall, Boudica proved to be a very strong leader and it's a bit of a shame that we didn't get to see her real abilities in the actual Game Six.


Gilgamesh of Sumeria
Wars Declared: 55
Wars Declared Upon: 15
Survival Percentage: 85%
Finishes: 5 Firsts, 6 Seconds (37 points)
Kills: 18
Overall Score: 55 points

Gilgamesh had an excellent starting position in the northeast corner of the map and he leveraged it into the second-best showing out of this group of AI leaders. Gilgamesh was almost always one of the top contenders in each game and he had by far the best survival rate with only three eliminations in total. The Sumerian leader was well positioned to expand thanks to having doormat Lincoln as his western neighbor and this was the main reason why Gilgamesh wound up with the second-highest kill total at 18. Gilgamesh's best games occurred when he was able to achieve a solo conquest of Lincoln; more frequently, he was part of a wider alliance that partitioned American territory between several different nations. Gilgamesh also found himself frequently engaging in wars with Huayna Capac with the Incans proving to be a tougher nut to crack. He won these conflicts more often than not but frequently took long enough that other leaders elsewhere on the map had seized control of the game in the intervening time. For some reason, Gilgamesh also charged across the world to attack Augustus more than a few times in a bizarre outbreak of fighting. I guess that was peace weight difference at work although it always felt weird when it took place.

All five of Gilgamesh's victories came via Culture which ended up being a theme for this game. This proclivity to turn off research and chase after culture proved to be a double-edged sword for Gilgamesh. On the one hand, it allowed him to sneak out narrow victories in Game #18 and Game #19 that he might not otherwise have won against larger rivals. On the other hand, Gilgamesh was one of the better researchers in this game and it felt like he was taking his foot off the gas pedal by stopping his teching at an early date to run the culture slider. I'm a little bit surprised that he didn't score more kills given that he declared war as frequently as Boudica did; I guess that he wasn't quite as suicidally aggressive as she was. If I had to summarize Gilgamesh's performance, I'd say that he did a fine job overall while still leaving a bit on the table. I think that he could have been a bit more dominant given the situation that he found himself in.


Huayna Capac of the Incas
Wars Declared: 18
Wars Declared Upon: 40
Survival Percentage: 40%
Finishes: 5 Firsts, 0 Seconds (25 points)
Kills: 4
Overall Score: 29 points

The pregame favorite and winner of the actual Game Six wound up being notably less successful across the course of twenty map replays. To his credit, Huayna Capac still achieved five victories and that's a pretty big deal for any leader. His wins were mostly of the unimpressive variety, however, won through turtling up and running the culture slider and praying not to be attacked by the other AI leaders. All five Incan wins took place via Culture and when combined with Gilgamesh's own love of the same victory condition it meant that this map had far more Cultural wins than any other type. Huayna Capac was held back by a cramped starting position that left him with relatively little room to operate. It was difficult for him to break free of the eastern seaboard and begin amassing territory in the same way that we watched on Livestream. He managed it only twice across these twenty replays, in Game #2 and Game #7. Not coincidentally, these were the two best games for the Incans, especially Game #7 which had a monstrously early Turn 257 finishing date. Huayna Capac managed a similar quick ending in Game #15 where his Cultural victory crossed the finish line on another blazingly fast Turn 266, albeit with quite a bit less territory. These were games where the Incans absorbed most of the Aztec territory right away and then stayed out of wars while the other leaders stagnated.

However, these games tended to be the exception more than the norm. Huayna Capac often came under attack and lacked the production base necessary to push back against larger invaders. He was the First to Die three times and the Incans were eliminated more often than they survived to reach the finish line. I mentioned this before but it bears repeating again that Huayna Capac had a paltry 4 kills to reflect his unimpessive status in warring. Here's another way of looking at this issue: while Boudica and Gilgamesh both won at roughly the same rate as Huayna Capac, they managed ten runner-up finishes between them while the Incans had a big fat zero in that category. The glass half-full way of spinning this is that Huayna Capac was pretty good at winning Cultural victories even from a subpar starting position. The glass half-empty approach would be that Huayna Capac was pretty bad at doing basically anything else on this map. He was fortunate to have the real Game Six play out the way that it did.


Mao Zedong of China
Wars Declared: 40
Wars Declared Upon: 17
Survival Percentage: 45%
Finishes: 2 Firsts, 3 Seconds (16 points)
Kills: 11
Overall Score: 27 points


Augustus of Rome
Wars Declared: 17
Wars Declared Upon: 46
Survival Percentage: 45%
Finishes: 1 First, 5 Seconds (15 points)
Kills: 9
Overall Score: 24 points

Mao and Augustus were what I started calling the "western pair" in my head while watching these games. Mao and Augustus were virtual mirrors of one another across the replays of this map, clashing again and again in mostly stalemated wars with a decisive winner rarely emerging. (This was a stark difference from the actual Game Six where Augustus wound up being notably stronger.) Mao and Augustus ended up with nearly identical scores, as China managed an additional win which was counterbalanced by Rome having somewhat better odds to finish in second place. If you really want to dig into the numbers, Mao had 40 offensive wars against 17 defensive wars while Augustus had 17 offensive wars against 46 defensive wars - nearly a perfect inverse of one another. A typical game had Mao initiating hostilities against Augustus for the first time and then the two of them fighting repeatedly for the rest of the mach. Often no one came out on top until a third leader from somewhere else on the map intervened. They also wound up with exactly the same survival percentage of 45% when all was said and done, with the peace weight disadvantage of Augustus canceled out by his more sheltered location in the southwest. These two matched one another almost perfectly.

The path to success for both of these leaders was having the other one weakened in some fashion. I went back and checked the table to see if the numbers backed up my impressions and there was an exact match here. There were five games where Mao had a top two finish and Augustus was eliminated in all five of those games. Similarly, there were six games where Augustus had a top two finish, and guess what, Mao was eliminated in all six of those games. The two of them couldn't both be strong, it had to be one or the other. (The actual Game Six was another outlier in this respect as Augustus finished second with Mao still alive to reach the Wildcard match.) For whatever reason both leaders won all of their victories via Spaceship, I suppose because neither of them had any real interest in chasing Culture and they didn't get strong enough to achieve Domination. They also had some of the slowest-finishing victory dates at Turn 347, Turn 355, and Turn 394 which reflects the weaker nature of their performances. Anyway, I think readers will get the picture. These two leaders were very similar overall and this was reflected in their near-identical results.


Montezuma of the Aztecs
Wars Declared: 35
Wars Declared Upon: 37
Survival Percentage: 15%
Finishes: 0 Firsts, 2 Seconds (4 points)
Kills: 8
Overall Score: 12 points

If we ever needed more proof that Montezuma is a bad leader for AI Survivor, the data from the replays of this map helps serve as confirmation. Montezuma had a favorable diplomatic environment with lots of other low peace weight leaders on his borders and a sitting duck neighbor to snack on in the form of Lincoln. Monty proved unable to translate this favorable start into even a single victory and he only had one game where he was even remotely competitive for the top spot. That was Game #4 where Montezuma partitioned the Incan domains at an early date and then shockingly overran Gilgamesh but still couldn't manage to overcome Mao's march to space. Montezuma could only muster up a pair of second place finishes across twenty games for a paltry 4 score points (the next closest non-Lincoln leader was Augustus at 15 points) and even that poor showing was overstating Monty's performance by a good bit. The Aztecs just found every possible way to die, over and over and over again. Monty only reached the finish date in three of these games for an elimination rate of 85%. He was First to Die on six different occasions and typically was out of the game before Turn 200. Almost any other AI leader would have done a better job than this.

Montezuma was dying repeatedly for the same reason that he always fails: self-destruction through excessive warring. Monty fought constantly with everyone else, even other nations that should have been friendly allies. This was aggression with no thought or logic behind it whatsoever. Monty's most frequent targets were Lincoln and Boudica (the latter because of religious tension) but he also invaded Huayna Capac and even Gilgamesh on many occasions. There were a number of games where the Aztecs managed a conquest and became surprisingly strong in the midgame... only for Montezuma to inevitably collapse after falling behind in technology from nonstop fighting. He has to be one of the worst leaders in the whole game at managing an economy. It's also worth mentioning that Montezuma managed to die via having his last city culture flip away to the Incans in Game #8, that's always worth a laugh whenever it happens. I suppose Montezuma did manage 8 kills thanks to being in so many wars, that was the best thing that anyone could say about this performance. While it was highly amusing to watch Monty's antics, the word is certainly out by now that his AI personality is a failure.


Lincoln of the Americans
Wars Declared: 2
Wars Declared Upon: 50
Survival Percentage: 0%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 0
Overall Score: 0 points

Game Six had one completely hopeless AI leader and his name was Lincoln. He had an impossible starting position, a high peace weight pacifistic leader surrounded by mostly warmongers, and the map played out pretty much exactly the way that one would expect. Lincoln was attacked early and often, usually by more than one AI opponent, resulting in an unbroken string of eliminations. He was First to Die eleven different times and never managed to survive to the end of a single game. In fact, Lincoln only made it past Turn 215 a single time and usually exited the stage by Turn 150 - these weren't just eliminations, there was a smoking crater where America used to be in game after game. Lincoln's lengthy survival in the actual Game Six was one of the most unusual features of the match. I think everyone was surprised that he lasted as long as he did and replaying the map conclusively proved that it had been atypical. Here's another way that the real Game Six was weird: Lincoln initiated two different offensive wars in the game played on Livestream. By way of contrast, he initiated two offensive wars across the entirety of these twenty replays. He started two wars and was attacked FIFTY times! So yeah, it was a little uncharacteristic that Lincoln did so well in the game that we watched.

Lincoln scored zero points in these replays, never finishing in the top two (or even surviving) and never managing a single enemy kill. I can't particularly speak to what his path to success would have been since he never did succeed in any of these games. Game #11 was his least bad performance, an unusual game where Montezuma collapsed early on and then Huayna Capac and Gilgamesh left the Americans alone for some time. Lincoln was still eliminated in that game but surviving to Turn 280 was better than what he managed in most of these attempts. He really didn't have much of a chance on this map and largely impacted the wider game by virtue of who acquired the most spoils from his inevitable death. Poor guy.

Conclusions

When Game Five took place, I was patting myself on the back for nailing basically everything in the prediction contest. Replaying that map had been a humbling experience as it turned out that I had largely gotten lucky and there had been all sorts of other possible outcomes that looked nothing like what I had predicted. Game Six was a case of exactly the opposite situation: I picked Boudica to win with Gilgamesh in second place and then flamed out with virtually nothing in the scoring contest. But replaying Game Six proved that I had actually read this game almost perfectly! Boudica was the most likely leader to win, Gilgamesh was the most likely to come in second, and Lincoln was the heavy favorite to be First to Die. (Montezuma was also a good pick for First to Die because he also had solid odds and most of the betting money was on Lincoln but the American leader was definitely the most likely to exit first.) As I watched Boudica march all over the rest of the field during these replays, I felt weirdly vindicated in my pick from the previous calendar year. See, I was right all along! Of course I hadn't been right about Game Five and had simply gotten lucky so this was a case of randomness evening out over time. One place where we all failed: only 6 out of 243 contest entrants picked Cultural as the victory type which was clearly the most common on this map. Whoops! Guess we all missed that one.

I hope this was entertaining to look back at. Thanks as always for reading along.