Civ4 AI Survivor Season 6: Game Five Alternate Histories


Game Five 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 Five saw heavy pregame favorite Willem fall apart at the hands of Bismarck thanks to an expertly timed German backstab. 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 Five

Game One | Game Two | Game Three | Game Four | Game Five

Game Six | Game Seven | Game Eight | Game Nine | Game Ten

Game Eleven | Game Twelve | Game Thirteen | Game Fourteen | Game Fifteen

Game Sixteen | Game Seventeen | Game Eighteen | Game Nineteen | Game Twenty

(Note : "A" column tracks the number of war declarations initiated by the AI, "D" the number of times the AI is declared upon, "F" the points for finish ranking, and "K" the number of kills.)

I wanted so badly to write about how Game Five had been an aberration and Willem was in reality the strongest leader on this map, thus justifying the huge price that I had paid for him in the AI Survivor fantasy auction. Needless to say, that was not how the alternate histories played out. Willem was completely ineffective on this map and rarely even managed to survive to the end of the game, much less be in a position to compete for a victory. Willem struggled to expand beyond five or six cities and he was never able to conquer one of his rivals to snowball ahead on territory. Worse than that, Willem repeatedly threw away strong positions in the midgame as the worst tendencies of his AI personality came to the fore. There were at least half a dozen games where he had an edge in both territory and technology yet still managed to lose a 1 vs 1 war to Bismarck or Washington because he refused to research Rifling tech and insisted on building wonders and missionaries rather than units. I'm confident that Huayna Capac manages to win at least three or four games from this position by virtue of having a better AI personality. Willem therefore turned out to be a total dud pick, with zero kills and only a single first place finish in a highly atypical game where he was never attacked once. He wasn't even a great pick for First to Die and this might be the single worst fantasy auction bid of the entire season.

Fortunately for me, I spent my final 13 gold by taking a flyer on Bismarck and the German leader wound up being an absolute juggernaut on this map. He won the actual Game Five in convincing fashion and backed up that performance by winning 12 more times across the 20 alternate histories. He added five more runner up finishes in the eight games where he didn't come in first place and only suffered elimination a single time. Bismarck similarly crushed his rivals in the kills department by nearly having as many kills (29) as the other five leaders combined (34). It was one of the most dominant performances that I've seen across the alternate history games and fell only a single point shy of breaking the 100 mark. And yet this overwhelming performance from Bismarck was utterly missed by the AI Survivor community, with only 5 people out of 245 contest entries picking him to be the winner. What in the world happened here? How did we all misread this result so badly?

As best I can tell, it was a combination of overvaluing the poor past performances from Bismarck and undervaluing his starting position. This was an exceptionally strong starting position for Germany with a bunch of floodplains tiles for food, a plains cow for production, a grassland gems tile that could be connected immediately (thanks to Germany's Hunting/Mining starting techs), and ample opportunities to expand into more high-quality land to the north. Bismarck had the good luck to start in a lush resource-heavy portion of the map with plenty of room to spread out. Bismarck didn't get much attention from the community because he'd been such a sad sack performer in past seasons based on a small sample size of outcomes. I know that I was also skeptical of Bismarck because he happened to start next to known loose cannon Ragnar, and indeed Ragnar would wind up fighting with Bismarck in most games. However, it can actually be a good thing having an insane AI as a neighbor if you happen to have a really strong starting position. Bismarck out-expanded and out-teched Ragnar in virtually every game; therefore, when the inevitable Viking invasion took place, Bismarck was able to counterattack and capture Ragnar's domains for himself. This provided a springboard to roll over the rest of the map and caused most of these games to end in Domination.

Bismarck was also aided by Willem's constant refusal to research the military techs which would keep him alive. Thus Bismarck had a secondary path to success by gobbling up his northern neighbor and claiming all of Willem's juicy world wonders for himself. Germany had a more reliable snowball path through Scandinavia but the Dutch were a frequent target as well. It's not an accident that Willem and Ragnar had the worst scores across the alternate histories: they had the misfortune to start next to the unstoppable Bismarck who absorbed their territories for Lebensraum again and again. Although there were a few games where Bismarck faltered - not many since he was a top two finisher in 17 of 20 games but it did happen - it required something unusual for the German leader to find himself outside of scoring contention. Bismarck almost always came out of the landgrab phase in first place and then secured his position by hitting key military techs before his competitors.

The other leader who drew a fair amount of community support in Game Five was Suryavarman. The Khmer leader ended up with a respectable second place scoring performance, taking home four victories and another four runner up finishes across the alternate histories. Suryavarman had the most isolated starting position on the map and appeared to have excellent opportunities for expansion through Ragnar and Lincoln. This did happen in some games and the best Khmer performances tended to take place in matches where Suryavarman was able to solo kill Lincoln and then keep rolling through the rest of the map. However, Suryavarman had a crippling weakness that prevented him from being more dominant: poor teching. Whether it was due to weaker economic traits or a lack of rivers in his territory, Suryavarman tended to fall far behind Bismarck in research and often lost lategame wars even when he had more cities under his control. There were several games where the two of them were close on the scoreboard but Bismarck was fielding tanks (panzers) against rifles/cavs for Suryavarman and we all know how poorly that goes for the AI.

As for the two American leaders, Washington wound up doing surprisingly well when he wasn't dogpiled into oblivion. Washington achieved three victories (only one less than Suryavarman) and shockingly had the second-most kills after Bismarck. This was heavily due to his success against Willem and I'd estimate that two-thirds of Washington's kills came against the Dutch. Washington's path to success was to conquer Willem and steal away all of his inevitable world wonders; when Washington pulled this off he was a major player, and when he didn't the red Americans were usually eliminated. Willem could have and should have beaten Washington far more often than events played out but the net result was a very solid set of alternate histories from Washington. Lincoln was much less impressive, never winning a match and collecting half a dozen runner up finishes through no great accomplishment of his own. Lincoln backdoored into second place six times for one reason: he was further away from Bismarck than anyone else. Lincoln almost always sat around in his northeast corner doing essentially nothing; his grand total of two kills really tells the story here. Washington was the far more effective of the two high peace weight leaders even though he was more likely to be eliminated.

Now for a look at the individual leaders:

Bismarck of Germany
Wars Declared: 47
Wars Declared Upon: 21
Survival Percentage: 95%
Finishes: 12 Firsts, 5 Seconds (70 points)
Kills: 29
Overall Score: 99 points

I covered most of the major points for Bismarck in the overall summary above. He was a true monster on this map and I was shocked at how he kept winning one Domination victory after another, even in games where his initial 100 turns were mediocre. It didn't seem to matter how the AI war declarations happened to shake out, it always felt as though Bismarck would find a path back to the top spot. Bismarck could have been even more successful than the numbers indicated here: he lost a SAME TURN victory tiebreaker to Washington in Game #18 when his spaceship landed on the same turn that the American leader triggered Domination. (Apparently the Domination victory gets precedent in Civ4's coding, likely due to alphabetical order in the victory conditions XML.) He basically had the teching performance that everyone expected Willem to enjoy and thanks to Bismack's sole tech flavor (MILITARY) he prioritized Rifling tech at an early date in every game. The best research rate in the field along with early rifles/cavs pretty much meant a curbstomping of a lot of medieval opponents. Bismarck seemed to find that sweet spot in these alternate histories between being overly passive like Pericles or Gandhi and being excessively militaristic like Ragnar or Montezuma. He also had the good luck to avoid getting targeted too often for invasions, facing only 21 defensive wars in total. Now this was heavily due to the fact that Bismarck was almost always very strong and the AI is less likely to attack neighbors who are much more powerful but it still helped in the early game. This wound up being the first game where everything truly came together for Bismarck and he showed what he could accomplish in the perfect scenario.

Suryavarman of the Khmer
Wars Declared: 51
Wars Declared Upon: 31
Survival Percentage: 65%
Finishes: 4 Firsts, 4 Seconds (28 points)
Kills: 13
Overall Score: 41 points

Suryavarman was also fortunate in having relatively weak neighbors to pick on. He had a great deal of success in conquering either Ragnar or Lincoln or both across many of the alternate histories. Unfortunately for Suryavarman, he also had many games where he stagnated for long periods of time against Lincoln, much as we watched in the actual Game Five on Livestream. Suryavarman's aggression against Lincoln also tended to bring in Washington as an additional opponent and frequently left him on the wrong side of 1 vs 2 conflicts. Suryavarman and Washington fought in significantly more total wars than any of the other AI leaders and the constant warfare helped to drag Suryavarman back to the pack in many of these games. He also tended to be a poor techer which was responsible for a number of defeats that really should have become victories based on the amount of territory under Khmer control. It wasn't a coincidence that all of Suryavarman's victories came by Domination since he usually lost out to someone else if the game in question came down to a tech race. Overall, I can't avoid the feeling that this was a somewhat underwhelming performance for Suryavarman given his starting position and a favorable diplomatic situation. Although he still had good results taken as a whole, I think that the top tier AI Survivor leaders would have won more frequently from his start.

Washington of America
Wars Declared: 27
Wars Declared Upon: 60
Survival Percentage: 55%
Finishes: 3 Firsts, 2 Seconds (19 points)
Kills: 15
Overall Score: 34 points

Washington was the most surprising AI leader in Game Five as far as exceeding low expectations, especially after he was First to Die in the match that we watched on Livestream. This turned out to be a highly unusual result since Washington was only First to Die once across the 20 alternate histories. I was shocked to see Washington win three different victories and with all of them being genuine, not flukey United Nations nonsense coming out of left field. Washington's main path to success was conquering Willem and taking all of the Dutch wonders/infrastructure for himself, then using their combined lands as a platform for further expansion. This was the roadmap in both Game #14 and Game #18 which played out in strikingly similar fashion: Washington solo conquered Willem, then took most of Ragnar's territory in a subsequent war, and at that point was too large to be stopped. Washington's AI personality tends to be ineffective in most games (he's a peaceful AI who likes to build military units) but if he gets far ahead of the field, he can snowball his position in reasonably effective fashion. That's not to say that Washington was unstoppable by any means, as he had roughly 50/50 odds to be eliminated and found himself getting dogpiled pretty frequently. Washington was attacked more often than anyone else in the game by a massive margin; he was invaded 60 times and Suryavarman was second-highest at 31 times. It was commonplace for Willem to be fighting with Washington over border tensions and then some combination of Bismarck/Ragnar/Suryavarman would leap in as well due to peace weight differences. But on the occasions where Washington wasn't ganged up on by his neighbors, he was surprisingly effective on this map. Suryavarman was slightly more effective, but only slightly, and that was pretty impressive.

Lincoln of America
Wars Declared: 12
Wars Declared Upon: 26
Survival Percentage: 40%
Finishes: 0 Firsts, 6 Seconds (12 points)
Kills: 2
Overall Score: 14 points

There was a major dropoff from Washington to the other American leader who wound up with the fourth-most points. Whereas Washington was a reasonably strong AI leader who kept running afoul of poor diplomacy, Lincoln was a completely ineffective leader who did very little across the alternate histories. Lincoln fought the fewest total wars of anyone in the field, only 12 offensive and 26 defensive wars, and yet he still wound up getting eliminated 60% of the time. His paltry 2 kill were indicative of an inert leader who sat in the corner building and teching along without interacting with the rest of the world. It was highly apparent across these games how Washington's AI personality was significantly more aggressive than Lincoln's AI personality despite their similar peace weights. I think the best thing that I can say about Lincoln is that he teched pretty well for someone who rarely had much territory under his control. Lincoln just does not show up much in my notes because he was so passive in these games. As I mentioned above, his second place finishes were all backdoor results that mostly stemmed from Bismarck starting out on the extreme other side of the world. However, Lincoln was never in contention to win any of these matches, even the ones where he finished in second place. Take a look at the endgame screenshots at the top of this page, he was never close to the top leader in any of the alternate histories. This was a poor performance that lucked its way into some theoretical playoff berths due to dumb luck.

Ragnar of the Vikings
Wars Declared: 28
Wars Declared Upon: 29
Survival Percentage: 15%
Finishes: 0 Firsts, 2 Seconds (4 points)
Kills: 4
Overall Score: 8 points

Ragnar had exactly the outcome in the alternate histories that everyone expected: wild, insane aggression that repeatedly flamed out without achieving much. He was significantly weaker in these matches than the actual Game Five that we watched on Livestream and in fact wound up First to Die far more than anyone else. Ragnar amusingly became a decent cheap purchase for AI Survivor fantasy purposes since he was First to Die 10 times in 20 alternate histories. This was certainly more than he accomplished in terms of real accomplishments, managing only a pair of runner up finishes and an unimpressive four kills in these repeated playthroughs of the map. Anyone reading this is likely well familiar with Ragnar's weaknesses: he fights too many wars and annoys too many of his neighbors and doesn't do enough to develop his own economy. Ragnar was perennially behind in technology in these games and suffered several big collapses (just as we watched in the real Game Five) due to his aggression backfiring against rivals with more advanced militaries. His best results came in highly atypical matches; Ragnar conquered Washington in Game #17 and then absorbed both Lincoln and Suryavarman in Game #19 to run out to big leads in both games. However, he couldn't close the deal in either one, with Bismarck executing a perfect backstab in the former game and Willem's small but vastly more advanced empire taking the Cultural prize in the latter. Ragnar was his usual wrecking ball self and he rarely managed to inflict much damage before getting hoisted on his own petard.

Willem of the Dutch
Wars Declared: 23
Wars Declared Upon: 21
Survival Percentage: 15%
Finishes: 1 First, 1 Second (7 points)
Kills: 0
Overall Score: 7 points

This was a terrible outing from Willem and most of the disastrous result was entirely due to his own AI personality. Willem is infamous for refusing to research the military techs needed to keep him safe (especially Rifling) and this self-destructive tendency was fully on display in the alternate histories. There were at least half a dozen games where he was ahead in tech and ahead in territory and still managed to die to Washington anyway in a 1 vs 1 conflict. That should never happen but it turned out to be readily possible when half the Dutch cities were permanently stuck on wonders or missionaries or some other nonsense. Although Willem had excellent land in his core he didn't expand particularly well and rarely managed to get more than five or six cities. We saw this in the Livestream game where he was doing too much wonder-building and that pattern was repeated time and again. Willem also wasn't helped in the religious realm either. He would usually found two or three religions and inevitably one of the minority faiths would spread to his neighbors to poison relations. Bismarck's gems resource also caused Germany to found its own religion in many games and that was similarly bad news, creating tension between Germany and the Dutch that overrode their shared peace weight. Peaceful coexistence is always in Willem's best interest but he almost never got it in the alternate histories - frequently due to his own stupidity! You might notice that Willem actually started more offensive wars than suffered defensive wars (23 to 21); he was often foolishly picking fights that he couldn't win rather than teching along in peace. There was exactly one game where Willem never fought a single war (Game #19), and sure enough, he won that match with a Cultural victory. Willem badly underachieved on this map with the flaws in his AI personality on full display.


The actual Game Five turned out to be pretty typical of the larger sample size in the alternate histories. Bismarck was indeed the most likely choice to win on this map, Suryavarman was a common runner up (it was either him or Lincoln most of the time), and Domination was definitely the best choice for the victory condition. The only truly unusual thing about the Livestream game was Washington getting eliminated first and Ragnar emerging temporarily as a major powerhouse. Otherwise the viewers of Game Five were treated to a fairly standard outcome for this scenario. These alternate history results were a bit embarassing as everyone missed Bismarck's true strength and most people highly overrated Willem. No one was more guilty of this than me as I staked my whole fantasy team on Willem dominating this map... only to get bailed out by nabbing Bismarck as well and stumbling into the true powerhouse. Well, you know what they say about being lucky and good and all that.

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