Civ4 AI Survivor Season 8: Game Two Alternate Histories

(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.)

Eauxps I. Fourgott: I think the general community feeling about Game 2 was that it was a return to normalcy after a crazy season opener. That certainly proved true for the alternate histories, which showed that we had seen a typical outcome for this map. This setup was absolutely dominated by the pairing of Justinian and Gilgamesh, who combined to take 70% of all top-two positions and nearly 60% of all kills, emerging as the strongest leaders in game after game and ultimately tying in their overall score total. Gilgamesh's strength had been easy to predict ahead of time, as he had a generous starting location that let him always expand to a good size and become one of the game's strongest leaders without lifting a finger. He also lifted a finger plenty of times, though, using his strength to go out and crack some heads and amass an impressive kill total; he was helped in this regard by bordering Khan and Kon, two of the map's best candidates for an early conquest. This led him to seven different victories, and even in games where he got the snowball going too slowly and couldn't keep up with Justinian, he was the "default" second-place finisher, duplicating his result from the real game another seven times. Actually, his real performance was one of his weakest ones, thanks to facing an early 2v1 that cost him one city and stunted his early development.

Meanwhile, Justinian had a less favorable starting position, but played the clear best series of games across the alternate histories. His early expansion was merely average, but in this map of mostly militaristic leaders, Justinian was able to regularly build the strongest economy while proving no military slouch himself. His early founding of Christianity would start him off on the right foot with the early score and GNP lead, and many times he was able to maintain this lead for the rest of the game, taking land off one or two rivals to reach the point where nobody could catch him in tech. Sometimes he'd have less land but a better economy than Gilgamesh, and leverage the latter to still come out on top in the end. Even in games where Justinian didn't do so great, he still fought well and proved a tough nut to crack. He would remain in contention until late in the game despite getting attacked ridiculously early and 2v1'd, and thus ended with a good survival rate even in the games where he didn't come out on top.

There was a huge dropoff after Justinian and Gilgamesh, and the results among the others ended up being more or less a crapshoot. Each of the remaining five leaders had their own issues on this map, all of them were eliminated in the majority of replays, and other than the truly hopeless Genghis Khan, they all had roughly even, low odds to advance to the playoffs. Only two unusual games had one of these five leaders win, and only a single one had neither Justinian or Gilgamesh advancing. Basically, it was extremely rare for the lesser leaders to accomplish anything against one of the two big dogs, and in most cases the best they could hope for was to slide into second place when one of Justinian or Gilgamesh knocked out the other. There wasn't even a heavy favorite for First to Die on this map, with Genghis, Bismarck, and Wang Kon all having roughly similar odds. While there were definite patterns to these leaders' performances, which one(s) would prove relatively successful on any given replay was largely random and couldn't be predicted with any reliability ahead of time.

To go into a bit more detail about each of the other leaders, Wang Kon was able to build a surprisingly competitive empire from what had looked like poor starting terrain, but was routinely dogpiled (or just attacked by a stronger Gilgamesh) and thus was eliminated the majority of the time. Bismarck also had land with high potential, but his slower start caused by more jungle and less early culture seemed to hold him back too much to be able to stand up to Justinian and Gilgamesh in the lategame, and that was in his good games. There were also plenty of times where he simply botched the opening and never had a chance. Tokugawa was able to expand much better than in the real game and was his usual prickly self in a fight, but his economy was terrible in these games which would always result in him falling behind. He usually relied on good relations with Justinian to skate by alive. Cathy's Wildcard result from the real game was borne out as she was the most likely leader to survive, thanks to a fairly sheltered position and frequently adopting Christianity, but she was indeed too cramped to have any shot of success in most of these games, and missed most of the opportunities she did get. Finally, Genghis Khan turned out to be always cramped badly by Cathy's culture and thus too small to have any chance of starting a snowball. His starting position ended up being much worse than anticipated, and he further played these games poorly to get eliminated time and time again without ever coming close to a win.

There were also two outlier results in these games that were completely unlike any of the other replays. One saw Bismarck take the world by storm, snowballing hard and winning the second-earliest win of any of the replays. In the other one, Justinian and Gilgamesh both failed to expand properly, leading to a wild game that ended with Wang Kon and Catherine as the only two remaining leaders. The community reactions if that had been the real game would have been a sight to behold.

I also wanted to discuss a couple of particular points of interest with regard to the early game. One was the location of everybody's second cities: we've seen in the past that there's some variability with this, some leaders always sending their second city to the same location and others choosing a variety of locations. On this particular map, almost every leader chose the same spot in every single game, which had some major effects on how the map played out. Justinian always ended up with a Holy City in the desirable river valley to his southwest, securing the most valuable territory between himself and Cathy, while Cathy always founded her first two cities to her south and east, badly cramping Genghis's starting location. Wang Kon would always found his second city to the northwest which ensured that he had sufficient room to expand. Gilgamesh was the only leader whose second city location varied, as while he'd always go to the north or northeast, there were a couple of different specific locations. I didn't notice any particular correlation between his choice here and overall performance, though.

The other general point to discuss is the founding of the starting religions, which had a major impact on the real game and was a big point of discussion. Most people, myself included, expected that Toku founding one of the opening religions was an unusual result, and this turned out to be correct. There was no true favorite in this category, and every single leader (including Justinian himself) founded the second religion in at least one game, but most often it was Gilgamesh or Wang Kon who actually did the deed. This didn't seem to have a major effect, though, as nobody seemed to perform particularly better with the early religion. The global adoption of Christianity in the real game also turned out to be an unusual result as the secondary religion(s) tended to spread a bit more, but this turned out not to have a major impact on the game dynamics. Religious lines mattered a bit less in this game where almost everybody was aggressive and would declare war at Pleased relations, and even with less diplomatic ties, Justinian was generally able to avoid disastrous dogpiles. A particular leader becoming Christian could help that leader stay safe from Justinian, but that was about it as far as major effects, since there would always be somebody whom he was willing to attack until he reached a point where it wasn't necessary any longer. So the real game had a fairly unusual result in this regard, but not one with a huge impact aside from the earlier elimination of Tokugawa.

Now for a look at the individual leaders:

Justinian of Byzantium
Wars Declared: 35
Wars Declared Upon: 49
Survival Percentage: 80%
Finishes: 11 Firsts, 3 Seconds (61 points)
Kills: 18
Overall Score: 79 points

There was a sizable minority of the fanbase, especially over on CivFanatics, that believed in Justinian for this game, while my own assessment was "if somebody bids a lot [in fantasy] on him this season, they're making a big mistake." Well, now we know who was right about this one, as the former champion backed up his real win with eleven more in the alternate histories and was head and shoulders above most of the other leaders. Justinian seemed to have the perfect AI package for this particular setup: a religious focus that helped him pull ahead in economy, sufficient military focus to remain competitive on that front, moderate aggression and refusal to declare at Pleased which helped him declare the right mix of aggressive wars without going overboard, and good religious spread and moderate peaceweight to keep himself from being a dogpile magnet. He still had close to the most defensive wars in the set, but it was a manageable amount, and with a higher peaceweight that number would have undoubtedly jumped up to absurd levels. As I mentioned before, Justinian was a tough fighter in these games and was hard to knock out even when at an disadvantage; one game saw him hold off an extended 2v1 against Cathy and Toku without ever losing any cities, and another saw him survive multiple wars with a base of just four cities. He only suffered four eliminations in the entire set, twice at the hands of a hostile dogpile and twice at the hands of a runaway AI. In general, for Justinian to NOT advance to the playoffs, he either had to have an unusually slow start (either because of his own slow expansion or from getting attacked very early) or get overwhelmed by a particularly bad dogpile. There was only one game that fit neither of these criteria, where he simply failed to pull in front for whatever reason and was outscaled by Gilgamesh.

Justinian's only failing compared to Gilgamesh was that he achieved significantly less kills, but what this essentially meant was that Gilgamesh would just keep going and going, while Justinian would content himself with the conquest or two he needed to win and stop there. This could potentially come up as a weakness in other games, but in this particular case I don't think Justinian lost many if any games through this sort of inaction. The other weakness of his performance was that all of his economic wins came fairly late, with Turn 360 as the earliest spaceship victory that he achieved. He was ahead of the rest of this pack in economy, but far from the level of, say, Mansa Musa, and that bears keeping in mind for when he faces a less warlike field. On the other hand, Justinian was versatile enough to win by all four victory conditions in this scenario; he only got a single Diplomatic and Cultural win each, but could have easily achieved more had he gone for them, and one game in particular ended with him within reach of all four victory conditions at the same time. Overall this was a very strong performance, and this is the third season in a row where Justinian has scored the highest in his opener's alternate histories. The evidence continues to indicate that he's one of the very best leaders for this competition, this season he was finally able to deliver in the real game, and I'm excited to see how far he can go in the playoffs.

Gilgamesh of Sumeria
Wars Declared: 54
Wars Declared Upon: 36
Survival Percentage: 75%
Finishes: 7 Firsts, 7 Seconds (49 points)
Kills: 30
Overall Score: 79 points

Thanks to our scoring system that rewards kill credits, Gilgamesh was able to tie Justinian in the overall scoring even though he was clearly outplayed by him in a majority of contests. I think this result slightly overstates Gilgamesh's strength (especially since with all the dogpiles that took place, just a few changes in the timing of city captures could have swung the total quite a bit in either direction), but it's clear that he was far stronger than anybody else on the map aside from Justinian. As mentioned before, he benefited from a generous backline area that only he could realistically settle, allowing him to possess an empire of competitive size even in games where he expanded slowly; Creative trait and the religion he often founded would also help his early game. By the midgame he would virtually always be one of the top leaders in score and food, and when he went to war that usually translated to a decisive advantage. He was also helped by bordering the two best war targets; he attacked Wang in almost every game, even when they shared a religion, and was often stronger on his own but also often got assistance from another leader to make for an easy conquest. He could also attack a Genghis who was busy fighting somebody else, which usually translated to a speedy kill. By the ends of these wars, Justinian would be the only leader capable of challenging him, and Gilgamesh could otherwise run over the field with little opposition.

Gilgamesh had been the favorite in the picking contest, and we were right to recognize that he had the best starting position, but he was only moderately successful in converting that to an actual win. Whether he placed first or second boiled down to how quickly he could move; the games he won were ones where he managed to get a snowball going before Justinian could do anything to secure his lead. If he couldn't manage a conquest fast enough (or see Justinian get crippled early on), he was consigned to second place. The real game ended up being one of his weakest performances, thanks to him suffering an early attack from Genghis Khan; this prevented him from really getting going until the midgame, and was an unusual result in the alternate histories, as Genghis instead tended to target Cathy first. As for when Gilgamesh failed to place at all, this required him to either lose a fight with Justinian in the lategame, or else get dogpiled (or in one unusual case, expand poorly and get solo killed by Wang). Dogpiles were rare for Gilgamesh though, as his low peaceweight and out-of-the-way location resulted in few leaders other than Wang being interested in attacking him or remaining committed to the attack. As a result, he only was eliminated five times in the set, and collected a top two finish in fourteen of his fifteen surviving games. Overall, this was a strong military performance from Gilgamesh, who collected at least one kill in all but three of the replays and averaged a kill and a half per game, but his relatively low number of wins also speaks to his economic weakness.

Wang Kon of Korea
Wars Declared: 26
Wars Declared Upon: 49
Survival Percentage: 20%
Finishes: 1 First, 2 Seconds (9 points)
Kills: 10
Overall Score: 19 points

Wang was the narrow favorite for First to Die on this map and only survived to the finish four times, yet he still finished the alternate histories in third place out of seven leaders. I think that speaks to how generally ineffective most of the field was and the random nature of their successes. Wang was unpopular in the picking contest due to his seemingly poor land quality and diplomatic situation, but he was actually able to build a solid empire with his available land. Sending his first settler straight towards Gilgamesh in every game helped him to secure a decent amount of territory, at least when he settled it in a timely manner, and he would often be one of the score leaders in the early to mid-game. However, sooner or later, he'd end up bogged down in wars and eventually eliminated. The biggest thorn in his side was Gilgamesh, who attacked him in virtually every game; even adopting the same religion as the Sumerians usually wasn't enough to save him! But Wang's high peaceweight made him a popular target for the rest of the map as well, and so he'd frequently suffer dogpiles that he couldn't recover from. Wang tended to do a decent job of fighting at first, especially in 1v1 wars, and sometimes managed to successfully fend off 2v1s as well like we saw in the real game. However, all this fighting would bog him down and cause him to fall behind, and so he'd eventually be cleaned up later down the road. He sometimes just died early as well, suffering the First to Die fate six different times, four of which resulted in Gilgamesh leveraging the extra early territory to win. While he made a good show of it, he was usually doomed by his high peaceweight, lower military focus, and inferior land compared to his western rival.

There were a few games where Wang was successful, though. He did actually manage to win one game, the bizarre Game 14 where everybody stagnated for a long time and he managed to eventually pull in front by conquering an unusually weak Gilgamesh, going on to score four kills before beating Cathy out via spaceship. He came very close to winning again in Game 17, another unusual one where he ended up working with Gilgamesh for most of the game, and in the endgame was on track to barely beat Justinian in the space race until the final turns. He also got one more second-place finish in Game 12 where Justinian helped him defeat Gilgamesh in a late battle. These games did show that there was some potential in Wang's starting position and that he was capable of taking advantage, but ultimately were very much the exception rather than the rule. For the most part, Wang spent these alternate histories getting kicked around by the others. He was pretty much doomed from the time his field was picked.

Bismarck of Germany
Wars Declared: 23
Wars Declared Upon: 52
Survival Percentage: 25%
Finishes: 1 First, 2 Seconds (9 points)
Kills: 6
Overall Score: 15 points

Bismarck had land with good potential, with lots of space and lots of grassland buried under jungles, but in the end proved generally unable to capitalize on this. Many of these games saw him simply fail to play well, expanding too slowly, launching an attack too early, or himself getting attacked before he could get up and running. There were also many games where he didn't have such a disastrous start, but also failed to really pull ahead before getting dragged down in warfare. All of these games saw him as an ultimately weak civ without a chance of success, and contributed to his 25% First to Die rate. Other games, like the real Game 2, saw Bismarck play better and get out to a strong position, in contention for a top two spot, only to eventually get in a fight with a stronger Justinian or Gilgamesh and get knocked out that way. He didn't seem to be able to outscale those two, since they were able to get up and running much faster, and so he would end up getting knocked out despite having been within spitting distance of a playoff ticket. Other than his generally slow starts, Bismarck had two main problems on this map. One was the fact that he lacked a category in which he could gain a real edge over the rest of the field; Bis has a military focus, but was stuck playing against a bunch of other leaders with that same focus, and so this afforded him no advantage at all while costing him the chance to get ahead economically. The other big problem was that Bis couldn't stay out of trouble. His relatively high peaceweight and central location made him a popular target and he was invaded more times than anybody else in this game; as a result, he couldn't just skate by with a slightly weaker empire, as somebody else would always come calling and knock him out. There were only two games in the entire set that did not end with Bismarck either out or in the process of getting conquered.

Bismarck managed to salvage his score in the alternate histories (relatively speaking) by winning a single, highly atypical game. This game saw him peacefully expand to a much larger size than anybody else, join in an early dogpile of Gilgamesh(!), and snowball from that point on to a blazing-fast Turn 270 Domination win. He never came remotely close to duplicating this feat, though, and only finished twice in a distant second place, achievable only because Justinian or Gilgamesh had been unusually knocked out of contention. Ultimately, Bis had an iffy location on this map and didn't have the skill needed to leverage it effecively. He was a poor bet for this one.

Tokugawa of Japan
Wars Declared: 41
Wars Declared Upon: 35
Survival Percentage: 35%
Finishes: 0 Firsts, 3 Seconds (6 points)
Kills: 9
Overall Score: 15 points

Tokugawa ended up being stronger in the alternate histories than he was in the real Game Two, but not by a whole lot. I didn't ever see him repeating the same pattern of atrocious expansion as he did in the real game - this seems to have been caused by his decision to found an early religion, as he also performed quite poorly the one time he repeated that move in the alternate histories - but better expansion wasn't enough for him to be truly competitive. Toku was able to expand to a decent size in many games and proved a scrappy fighter as always. He managed to secure a kill in nearly half of the games, usually against Bismarck; the two fought a lot and which one was stronger varied, but Toku was able to get the upper hand a fair amount of the time. The problem was, by the time he was finished fighting Bismarck, he would be horribly behind in tech and never come close to catching up. There were a good number of games where Toku was competitive in land and power, but his terrible economy resulted in him eventually fighting wars while a generation behind in military tech, and that just doesn't cut it. Tokugawa isn't a very good economic leader in the first place, and this was compounded by his available land on this map being fairly poor. I think a better leader could possibly have made something of this start, but Toku in this position was incapable of competing for the win. He managed to secure a trio of second-place finishes in games where he avoided coming to blows with a stronger leader, but more often he'd eventually be crippled or eliminated due to his poor tech. His First to Die result was at least a rare outcome - he tended to exit later on in the game and only repeated this result once in the set - but the overall picture was still pretty similar. Toku didn't have much of a chance here.

Catherine of Russia
Wars Declared: 40
Wars Declared Upon: 25
Survival Percentage: 45%
Finishes: 0 Firsts, 3 Seconds (6 points)
Kills: 4
Overall Score: 10 points

Cathy was the favorite for First to Die in the picking contest, but this proved to be a poor prediction, only occurring twice in the alternate histories. However, the overall read that she didn't have enough space to be competitive was absolutely correct. Cathy was stuck with a poor position squeezed up in the corner of the continent, and to make matters worse, Justinian always founded a holy city that claimed him a lot of the contested territory between them. As a result, she could almost never expand out to more than five or six cities, and especially for a military-focused leader in a field full of them, this wasn't enough to work with. She wasn't helped at all by the fact that Genghis Khan usually attacked her very early on; the two were pretty evenly matched and she was rarely in danger of actually dying from this, but it did hamper her and keep her on the bottom. (It also set up both of her First to Die results, when Justinian dogpiled her to ensure an early exit.) The overall result was that Cathy was more or less stuck waiting in the corner, hoping for an opportunity to poach some cities and move up a little bit, and unable to do much herself to escape her predicament. (One specific point to discuss is the very close settling race that she lost to Justinian in the real game, when he settled Nicaea to the northeast to cramp her further. It turned out that this was an unusual result in the alternate histories... but ultimately didn't matter, since she in turn founded Yaroslavl to his south in a region that he normally settled in these games. It more or less evened out.)

Cathy also didn't play this game as well as she could have, though, and often failed to seize the opportunities she did get to move up. There were several games where she was able to get some extra land, only to fail to war at the times she needed to and solely pick unhelpful fights, ensuring that she remianed a weaker leader. Perhaps the most egregious example was one game where she and Gilgamesh were set to conquer Genghis together... only for Cathy to sign peace, leave him completely to Gilgamesh, and instead send her army over to unsuccessfully attack Wang Kon. The Wang Kon win also saw her as the other remaining leader at the end, and I think there was a real opening for her to win here... but after the game's early stages, she chose to sit back in peace, watching as instead Wang went out and conquered the remaining leaders to ensure his win. I think these lapses in judgment were caused as much by the inherent randomness of the AI as anything else, but they do make it clear that Cathy played worse than most of the pack and deserved her poor finish. At least she was decent at surviving; she accomplished this in nearly half the games, as her position was fairly sheltered and she didn't face any close threats if she shared a religion with Justinian. She did manage a trio of second-place finishes as well, when she got opportunities to pull ahead a little bit and actually took them (and, notably, Justinian was knocked out of contention, giving her more space to breathe). For the most part, though, this was a poor performance from a poor position. Cathy will have to hope to do better in the Wildcard this year.

Genghis Khan of the Mongols
Wars Declared: 50
Wars Declared Upon: 23
Survival Percentage: 10%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 5
Overall Score: 5 points

Genghis had attracted a lot of attention prior to this game; he had a very strong capital location that made some fans think that this might finally be his time to shine. However, this prediction proved to be a complete miss, as the Khan was the only leader to secure zero top-two finishes across the alternate histories. The biggest problem for Genghis, and one that I don't think many people anticipated (I certainly didn't), was that his position was super cramped. It didn't look like it at first, but Cathy always sent her first two settlers to the same two locations in Genghis's direction, and her free Creative culture coupled with his own refusal to tech Mysticism ensured that this would strictly limit the amount of land he had available. It was unusual for Genghis to be able to found more than five cities, and that simply wasn't enough to be competitive. From that point on, he was his usual hyperactive and psychotic self, launching a lot of wars (the second-most of the set, despite his weak position and tendency to be dead) but accomplishing very little with them. Unlike in the real game, he most often attacked Cathy first, and could sometimes take a city or two this way, but she was usually able to hold him off and this wasn't a reliable way for him to get out in front. Everybody else was even stronger and so his attacks on them would see even less success. He could occassionally grow stronger by taking part in a dogpile and getting some cities that way, but he didn't have the initial strength needed to ever properly snowball. His general weakness and proximity to Gilgamesh also made him a solid pick for First to Die.

While Genghis's start was a big factor in his poor performance, his own overaggression also played a role; there were a few games where he managed to become genuinely competitive for a top two position, only to eventually get himself in a fight he couldn't handle and knocked out of contention anyway. His strong starting land did allow him to keep up economically in games where he could expand to a decent size, making up for what is normally one of his biggest weaknesses, and I think a more balanced leader could have made the playoffs from this start at least a couple of times. Genghis was the wrong man for the job, though, and so only scored points from a small handful of kills; he couldn't even skate by to the Wildcard in most cases, as he just kept getting in fights until he got killed. At least he turned in some entertaining performances; perhaps my favorite was a game where Wang was conquering him, only to get backstabbed by Gilgamesh. Genghis repaid the favor by himself attacking Gilgamesh, who had just saved his butt, about 15 turns later, then peacing out just as Wang was starting to turn things around, leaving Wang to his fate and instead attacking a runaway Justinian to get himself killed. Anyway, the biggest reason that the community was fooled about Genghis's chances was that his start was much less favorable than it had first appeared to be, but his own poor AI personality sealed the deal to make him the biggest dud on the map.

Conclusions

I've recently seen it complained that the alternate histories aren't very reliable because they tend to diverge too much from their games' real results - that something about running the replays must have an effect on the AIs, since we've gotten as many outlier results as we have. Well, perhaps this set will offer a good counterargument to that complaint. This time they rather aggressively indicated that we had received an expected result: Justinian was the most common winner, Gilgamesh was the most common runner-up, Cathy was the most likely other leader to survive to the wildcard, Domination was the most likely victory condition, and even the finish date and number of wars were pretty average. A few details were different, but the overall dynamics of the game we saw were repeated over and over again in the alternate games. Not only that, but the alternate histories did produce two completely wacky results in games that looked completely unlike any of the other replays, showing that yes, that DOES just happen from time to time!

After such a wacky Game 1, it is perhaps for the best that we actually got a normal result that fell so well in line with the expected outcomes. In any case, I hope you enjoyed this look back at the setup for Game 2.