Wildcard Game 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. The Wildcard game had seen Gilgamesh pull off an unexpected Diplomatic victory when Mao Zedong seemingly had things locked up. 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:
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 previous few games that I'd investigated with alternate history scenarios had largely confirmed that the game we watched on Livestream was a typical outcome. The Wildcard game did not match that pattern, however; Gilgamesh's victory had felt like a weird fluke and replaying the map confirmed that this was a rare outcome. In fact, Gilgamesh didn't win even one of the twenty replays of this scenario although to be fair he should have been the winner of Game #4 (where he lost a same-turn tiebreak to Hammurabi). Instead the dominant AI leaders proved to be Mao Zedong and a somewhat surprising Genghis Khan Temujin. Mao had the best track record in terms of victory points with eight different first place finishes and a quartet of runner up performances. He was almost always the top AI from an economic perspective and this allowed him to seize the technological lead in game after game. Aided by the complete lack of Financial AIs on this map, Mao frequently reached a more advanced tier of military technology than his rivals and then either escaped into space or ran over his backwards neighbors. This was a pretty sad lot of researchers outside of Mao and it was reflected in the late finishing dates of these games.
The unexpected top competitor with Mao ended up being Genghis Khan. The Mongol leader has been so self-destructive with excessive militarism over the course of past AI Survivor matches that it caught me off guard when he started winning a series of games. But Temujin won five different times including three games by the Diplomatic ending - what in the world?! The key to his success was having a peacenik AI neighbor nextdoor to bully in the form of Hammurabi. Time after time, Temujin would emphasize military technology and then overrun a more advanced Hammurabi who had skimped out on units to build wonders and city infrastructure. The Mongols would capture those big cities full of wonders for themselves and then snowball off the accumulated land. Temujin's economy was frequently abysmal but if he could grab enough territory for himself at some point it stopped mattering how badly he'd neglected universities and courthouses. And the Mongol leader compiled a stunning kill total of 28 over these games, double any of the other leaders despite getting eliminated in fully half of these matches. I can't recall seeing an AI leader who wasn't particularly dominant scoring that many kills in any of the other alternate histories. But then again, Temujin launched 80 offensive wars (!!!) and never stopped fighting someone from beginning to end. Against a field of generally weak opposition, he ended up being one of the strongest individuals.
Brennus was the best of the rest with three wins and a series of second place finishes. As we watched in the Livestream game, he benefited from having a more sheltered position that didn't attract as much barbarian activitiy. The same couldn't be said for the other leaders on the southern side of the map who all suffered repeatedly from the Raging Barbarians setting. This was one of the patterns that emerged from replaying the map and I don't think it was a coincidence that Mao and Genghis Khan ended up winning so many of the games. There was simply a lot more open space in the southern tundra where barbs could spawn and slow down the development of the nearby civs. All told, the three northern leaders scored 16 of the 20 victories while the four southern leaders could only manage four wins between them. Gilgamesh and Tokugawa were frequently weak and bullied across these matches, stuck behind the normal growth curve and eventually conquered by their peers. Tokugawa in particular was eliminated in 85% of these games despite a favorable diplomatic environment simply because he fell behind in the early stages of so many games. Bismarck tended to do better against the barbarians only to suffer the same fate as we saw in the actual Wildcard match: everyone else would hate him diplomatically and he would be killed off at some point in the midgame. This happened over and over again and was responsible for his horrible 90% elimination rate. Hammurabi similarly tended to have strong early games and then get dopgiled by a group of the low peace weight leaders. The Mongols attacked in basically every single game and there was further pile-on from Mao and even Brennus in many games. Brennus did a lot of cross-map fighting in some of the other alternate history games and it seems to be part of his AI personality. On the rare occasions where Hammurabi didn't get killed off, his economic emphasis gave him an excellent chance to win outright.
The presence of a bunch of mostly mediocre AI leaders resulted in some unusual aspects to these games even outside of the Raging Barbarian settings. I want to emphasize again how slow these matches were in terms of their finishing dates. A typical game of AI Survivor concludes somewhere around Turn 325 since there's usually at least one AI who's managed to finish the whole tech tree and launch the spaceship if another victory condition hasn't manifested yet. For this particular set of alternate histories, there wasn't a single game that ended before Turn 300 and the average finishing date came out at Turn 367! Individual games concluded on Turn 397, Turn 386, Turn 389, Turn 406, Turn 392, and Turn 427 in Game #14. This might not matter when you're reading a summary of the results but I watched these games play out and they were loooong because the AI leaders mostly kind of sucked at winning a victory condition. On that note, there were no Cultural finishes at all and no one made any effort to chase after that ending. Most of the games concluded by Domination and then there was a strange number of Diplomatic endings to go along with a few Spaceship finishes. These Diplo victories happened for the same reason that we saw on Livestream: there was a bunch of low peace weight leaders who had scored mutual military struggle bonuses by working together to kill off Hammurabi/Bismarck. Combine those friendly relationships with a lengthy endgame sequence to allow time for more UN votes and it makes sense. Better economic leaders simply win by Spaceship or Culture at a faster date and leave less time available for UN voting shenanigans to take place.
Now for a look at the individual leaders:
Mao Zedong of China
Wars Declared: 35
Wars Declared Upon: 45
Survival Percentage: 70%
Finishes: 8 Firsts, 4 Seconds (48 points)
Overall Score: 62 points
Mao Zedong was the leader who should have won the actual Wildcard game and he was clearly the top leader in the alternate histories that I ran on this map. In addition to his eight victories and the most finish points by a wide margin, Mao also the highest survival rate of any leader at 70%. He was undoubtedly the best economic leader on this map and ran out to a tech lead in seemingly every game. We've seen this before in other seasons of AI Survivor as Mao frequently out-performs his mediocre Expansive/Protective traits and often winds up as one of the better teching leaders. His AI personality just does a better job of internal development than many of his competitiors for whatever reason. On this particular map, Mao was often able to avoid early warfare and quietly grow a tech lead as everyone else engaged in violent bloodletting, then pounce with rifles or tanks that the other AI leaders couldn't match. Mao wasn't one of the more aggressive leaders and ended up with more defensive wars than offensive wars despite finding himself far ahead in many of these games. (The AI gets more aggressive when it has a power rating lead so this is pretty noteworthy.) It helped that Genghis Khan would usually go charging off against Hammurabi and leave Mao along to develop in peace. This pattern played out repeatedly and led to a series of Chinese victories.
Where Mao found himself getting into trouble was when he faced an early attack from Temujin or Brennus or Gilgamesh. This only happened a minority of the time but it did occur and usually led to a weaker outcome for Mao. Both of the two games where Mao was First to Die were a consequence of the Mongols invading at an early date. One advantage working in Mao's favor was relatively little harassment caused by the barbarians. The northern tundra didn't have too much space where they could spawn and Chinese units were typically able to keep them under control. Mao was able to build markets and libraries rather than get pillaged into the stone age from endless waves of barbarian interlopers. Because Mao didn't engage in as much fighting and generally wanted to be left alone, he ended up with notably fewer kills than would be expected from a leader who won this many games. The fact that he only had 14 kills kept this from being a total runaway in terms of the points scoring. But leaving that aside, Mao still collected his fare share of eliminations and no one else could match his record in terms of winning the game. I feel bad for the folks who had Mao as their selection in the picking contest because the alternative histories revealed that this had clearly been the correct choice for the Wildcard game's victor.
Genghis Khan Temujin of the Mongols
Wars Declared: 80
Wars Declared Upon: 27
Survival Percentage: 50%
Finishes: 5 Firsts, 3 Seconds (31 points)
Overall Score: 59 points
Genghis Khan was the most entertaining figure to watch from the Wildcard game and he somehow nearly equaled Mao Zedong in the scoring. I think that these results slightly overstate Temujin's performance because he pulled out Diplomatic victories in Game #6 and especially Game #17 where Mao had an uncatchable tech lead and would have won by Spaceship if the United Nations hadn't intervened. But even taking this into account, Temujin was definitely the second-strongest AI leader on this map and scored far more kills than anyone else which is a pretty big deal. The Mongols were aggressive to an absurd degree with 80 offensive wars against only 27 defensive ones. Again, that's not Temujin fighting 80 wars in total, that's 80 wars that he specifically started on his own, a perfect average of four per game. This is the most that I've seen to date in running the alternate history scenarios. He fought, and fought, and then fought some more. It's not surprising that this turned into 28 kills which was double anyone else's total. Temujin is a very simple leader to understand: he's going to train military units and attack and that's pretty much it. This strategy tends to fall apart in most AI Survivor games because the Mongol economy ends up comically weak as a result but it worked out on this map due to the presence of fairly unimpressive rivals. Temujin was essentially a parasite that leeched off the economies of other nations after conquering them, inherently a risky strategy that happened to succeed more often than not for this setup.
Temujin's most common path to victory came from overrunning Hammurabi and then snowballing off his captured cities and world wonders. I could almost set my watch by how often the Mongol invasion of Babylon took place, kicking off in the Turn 80 to Turn 100 range in almost all of these games. Temujin was often behind in tech and city development but usually succeeded anyway by the simple expedient of building units in every city while Hammurabi kept working on wonders until it was too late. This provided the Mongols with additional cities and population which could then be thrown against Bismarck or whoever else as the violence continued. It wasn't that dissimilar to the Mongol conquest of China in the real 13th century. When Temujin faltered, it was for exactly the reasons that everyone would expect: too much warring, too many enemies, shockingly poor research that left the Mongols a generation behind in military tech, etc. The great khan was knocked out in fully half of these games which is way too high for as strong as he tended to be, a reflection of how he never stopped clubbing everyone else over the head and making enemies. Somehow he also won diplomatically three times though when the stars were in perfect alignment; this game is weird sometimes. ¯\_(ツ)_/¯
The Mongols were definitely the most fun civ to watch in these games due to their penchant for insane violence. There were other wacky things associated with Temujin's performance that I have to mention here. In Game #6 he managed to lose his capital to the barbarians and still came back to win the game! It was very rare to see an AI capital get captured (I think it happened three times across all 20 games) and somehow this doofus won in the United Nations anyway. That would have been amazing to watch on Livestream. Then in Game #11 the Mongols managed to die when their last remaining city culture flipped to Mao. I remembered to capture that in a screenshot and awarded the kill to Mao in his ending tally. I demonstrated that this can happen in my Passive Aggressive Conquest game but if you didn't already know, yeah, culture flips can kill civs in Civ4. Anyway, this guy is a total psychopath and it was highly entertaining to watch him flip out and do crazy stuff in these games.
Brennus of the Celts
Wars Declared: 48
Wars Declared Upon: 27
Survival Percentage: 60%
Finishes: 3 Firsts, 4 Seconds (23 points)
Overall Score: 37 points
Brennus was the strongest of the southern AI leaders and generally made the most of his more sheltered starting position. He was in many ways a weaker version of Mao, finishing with the same number of kills and a slightly worse survival percentage. Unlike Mao, Brennus was notably more aggressive with almost twice as many offensive wars as defensive wars. Brennus frequently engaged in warfare with Gilgamesh and usually came out on top in the clash between the two neighbors. This was a major difference from the actual Wildcard game that we watched where Brennus and Gilgamsh shared a religion and never fought at all, eventually uniting hands and working together to hand Gilgamesh a Diplomatic win. That was an extreme rarity and it was far more typical for these two leaders to be at one another's throats. Gilgamesh suffered more harassment from barbarians and also seemed to pull more aggression from the other AI leaders which gave Brennus the leg up in their contest. When Brennus could conquer through Sumeria and gain another empire's worth of territory, he could snowball from there and work his way to a Domination victory. This was the path that the Celts followed in Game #1 and Game #7, the two best performances for Brennus.
Of course it wasn't unusual to see Brennus struggling either if he couldn't manage to break out of his corner starting position. He had a good survival rate and was only First to Die a single time (in Game #15 when Mao and Gilgamesh both attacked early on) but it could be tough to find a way out of that cramped location along the western ocean. Brennus didn't have the traits or the economic personality to win peacefully like Mao and instead he always searched for a military route to be powerful. I mentioned this before but Brennus engaged in an unusually high number of cross-map wars, enough that I don't think it can be chalked up to coincidence. He really liked to attack Hammurabi or Bismarck and especially if one of them had a competing religion. This just seems to be part of his AI personality. Overall, Brennus was a bit of a pleasant surprise on this map. Little was expected of him in the picking contest but he ended up being one of the better AI leaders if still a good ways short of Mao's China.
Hammurabi of Babylon
Wars Declared: 14
Wars Declared Upon: 51
Survival Percentage: 25%
Finishes: 3 Firsts, 1 Second (17 points)
Overall Score: 20 points
The other four leaders on this map were all more or less even from a scoring standpoint even though they arrived at those points in different fashions. It's a bit shocking that Hammurabi wound up with 20 points thanks to three different victories, and while I think that this is somewhat due to small sample bias working in his favor, the Babylonian leader had pretty good odds to take home a victory if he could avoid dying. Hammurabi's stats reflect what we would expect to see from the only true high peace weight leader on this map: few offensive wars, few kills, lots of defensive wars, and lots of eliminations. Babylon had the most sheltered starting position on the globe and Hammurabi was typically the score leader during the early turns since he faced little harassment from the barbarians. He wasn't the fastest at expansion and liked to stop to build additional world wonders, with the Pyramids being a particular favorite. This created a clear dichotomy: if Hammurabi could stay out of war, his economic and cultural development would accelerate him past his competitors and put him in excellent position to capture the victory. Of course, as a high peace weight AI in a field of low peace weight aggressors, this rarely happened and Hammurabi was eliminated again and again by his rivals. He won 3 times out of 5 survivals... but died in the other 15 games. Whoops.
Somewhat surprisingly Hammurabi was not the most likely to be First to Die in these games even though most of the other AI leaders hated him. This was likely due to his safe corner starting position as he didn't take much damage from the barbarians and the poor Japanese (who were hammered in game after game) instead saw the most early exits. Hammurabi did suffer the most invasions of anyone in the game however so it wasn't exactly fun and games in Babylon. Hammurabi's wins were also somewhat of a fluke; in Game #20 he was the smallest civ on the map and dying to Mao in a war when his spaceship landed, while in Game #4 the Babylonian spaceship won a *SAME TURN* tiebreak with Gilgamesh's Domination victory. Only in Game #18 was Hammurabi the dominant AI leader thanks to massive infighting amongst the low peace weight leaders. His true odds to win were most likely somewhere in the 5-10% range rather than the 15% that I observed here. That's still better than I expected to see and reflects the advantage that the northern AI leaders had on this map thanks to the lessened barbarian activity.
Tokugawa of Japan
Wars Declared: 40
Wars Declared Upon: 46
Survival Percentage: 15%
Finishes: 1 First, 1 Second (7 points)
Overall Score: 19 points
Tokugawa wound up with an almost identical score as Hammurabi while reaching that score in a completely different fashion. The Japanese leader had an even worse survival rate despite a much friendlier diplomatic environment and could only manage a single victory and runner up finish. Tokugawa was able to make up the difference by scoring far more kills, winding up with an even dozen that nearly matched Mao and Brennus. It's also worth noting that Tokugawa's scoring was due in no small part to the one strong game that he played, as Japan ran a clear sweep with a victory and four kills for a total of nine points in Game #9. This was a game where Bismarck was completely hamstrung by barbarians and Tokugawa managed an early conquest followed by snowballing across the rest of the map. Without that outlier result propping up the rest of the scoring, Japan very easily could have finished at the bottom of the scoreboard.
The big problem for Tokugawa was the barbarians boiling up out of the southern tundra. He seemed to struggle with them more than anyone else and it was commonplace to see Japanese territory lacking any improved tiles for long stretches of time. The poor Japanese starting techs (Wheel/Fishing) didn't match this map location at all which only put Tokugawa further behind the rest of the field. Japan was the leader most likely to be First to Die and this was heavily due to the weak openings that Tokugawa kept getting stuck with. The best path to success for Tokugawa came when he could ally with other low peace weight leaders to attack and annex Bismarck's or Hammurabi's lands. This tended to work out better for other leaders than Tokugawa though and his isolationist personality often left Japan on the outside of the alliance structure. Tokugawa has been a half-decent AI leader on some of the other AI Survivor maps but this clearly wasn't a good setup for him.
Gilgamesh of Sumeria
Wars Declared: 34
Wars Declared Upon: 47
Survival Percentage: 40%
Finishes: 0 First, 5 Seconds (10 points)
Overall Score: 15 points
Gilgamesh was the leader who had the biggest disparity between his typical performance and the game that we watched in Season Five. During the actual Wildcard game, Gilgamesh forged a strong friendship with Brennus and conquered the Mongol core territory in the midgame, then overran Bismarck in the lategame to amass enough population for an unlikely Diplomatic victory. This felt lucky at the time and the alternate histories confirmed that this had been very lucky indeed. More repetitions of the map proved that Gilgamesh was the weakest of the low peace weight leaders, rarely reaching a position of power and influence. While his survival rate was considerable better than Tokugawa (and he should have had a victory over Hammurabi in Game #4), Gilgamesh didn't have very many opportunities to pile up the additional territory that he needed to become dominant. Attacking Tokugawa probably represented his best chance but Gilgamesh usually found himself stuck in wars with Brennus, and the Celts normally came out on top in that matchup. This was another knock-on effect from the barbarians as Sumeria tended to suffer heavily at their hands. We thought in the picking contest that Gilgamesh's nearby copper and Creative trait would give him the edge there; however, for whatever reason Gilgamesh was pretty terrible at connecting his copper and getting out vultures. Perhaps it would have made a difference if the resource was north instead of south of his capital - he had enormous trouble keeping those tiles from being pillaged over and over again.
Gilgamesh therefore came out of the landgrab phase in one of the weaker positions on the scoreboard in most games. He was rarely in a position to be a real threat to win and backdoored his way into the second place on a number of occasions. Examples of this included Game #10 and Game #11 and Game #15; Gilgamesh had no chance to win any of those matches and fell into second place when a much stronger Bismarck was killed by Mao and Genghis Khan. Outside of Gilgamehs's one excellent performance in Game #4, he had his best result in the actual Wildcard game. I didn't see Gilgamesh conquer the Mongols in any of these matches and that outcome seems to have been a highly unlikely occurrence. Another sign of the Sumerican weakness was a lack of kills at only five, highly atypical for this usually aggressive nation. Gilgamesh never tried to chase after a Cultural victory either, a goal that he might have been able to achieve given how slowly some of these games finished. This was a clear case of the AI Survivor community being wrong: Gilgamesh was not a good choice at all on this map. My own pick of Gilgamesh was the equivalent of hitting an inside straight draw in poker - I didn't deserve to be that lucky!
Bismarck of Germany
Wars Declared: 33
Wars Declared Upon: 41
Survival Percentage: 10%
Finishes: 0 First, 2 Seconds (4 points)
Overall Score: 14 points
Bismarck graded out as the lowest scoring leader on my replays of this map at only 14 points and had the highest elimination rate at a whopping 90%. This doesn't mean that he was obviously the worst leader though, as the scores were very close for Hammurabi and Tokugawa and Gilgamesh and Bismarck. We would need a bigger sample size to determine if one of these leaders was clearly worse than the others on this map. (This ended up being a game where everyone scored double digit points and there was no true hopeless leader like Gandhi in that one playoff match.) Bismarck was the other leader with a high(ish) peace weight which caused the low peace weight leaders to be predisposed to dislike him. He was only First to Die three times and often came out of the landgrab phase in a fairly good position. Germany would get hit with significant barbarian activity but could usually take advantage of a neighboring Japan which was in worse shape. In fact, most of Bismarck's scoring came from kills that were disproportionately directed against Tokugawa. Just as we saw in the actual Wildcard game, Bismarck would conquer a weak Japan at some point in the midgame and emerge as one of the stronger AI leaders.
And then it would all fall apart afterwards. Either Mao or Genghis Khan or Brennus or some combination of all three would have finished up with their own conquests and then would direct their attention towards the German nation in the east. Bismarck was repeatedly the victim of diplomatic dogpiles and this was the main reason why he was eliminated so often. He didn't do himself any favors by often skipping out on Rifling research for long stretches of time and allowing a more backwards Temujin to invade with superior units. To Bismarck's credit, he did finish in second place in both games where he survived... but "both" means that he died 18 times in 20 games, yikes. It's impressive that Bismarck managed to score eight kills across those games where he didn't survive and I'm honestly surprised that he managed as many points as he did. The only really good game that Bismarck played was Game #20 where he was the score leader at the time that Hammurabi won by space. In pretty much every other game, Germany was either roadkill or an also-ran (for example, Bismarck was in the process of dying to Mao when he took second place in Game #8). For a leader who had zero points in the historic AI Survivor rankings coming into Season Five, we didn't expect a whole lot and we didn't get a whole lot.
In summary, Mao Zedong was the best leader on this map but it didn't have any truly runaway leaders or any completely hopeless individuals. There was less separation between the top of the AI rankings and the bottom of the AI rankings than we typically see when running these alternate history exercises. This was reflected in the lack of clear favorites in the Runner Up and First to Die categories where none of the leaders particularly stood out. My guess is that this was due to the nature of the leaders who find themselves in the Wildcard game; since it generally contains mediocre to incompetent AI leaders, there's less chance for the top performing leaders to separate themselves from the rest of the pack. The barbarians may also have been a random factor that dragged everyone down to some extent. I'll try to make sure that future maps don't have as much of a north/south imbalance in that regard. (This is a bit tough because I generated four different potential maps for the Wildcard game and didn't know which one we'd be using until after Game Eight ended. There's always room to do better though.)
Now there's one other question that I'm sure readers are wondering: what would happen if the barbarians could spawn all of their normal units? They can't produce axes or spears under AI Survivor rules because the observer civ doesn't research Bronze Working or any of the other Ancient Age techs. I went ahead and tested this as well by granting the barbarians all of the non-religious Ancient Age techs in the Worldbuilder just to see what would happen. This wound up being a big disappointment however: nothing really changed when I ran out the first 100 turns about ten different times. The AI civs all researched Archery tech and with their Deity combat bonuses they were able to hold their cities just fine against barb axes. I saw a capital city fall once or twice but there were no AI empires getting wiped out and the games basically played out the same as before. There was a bit more randomness, more cases of one individual AI getting completely crippled by barbarian activity, that was all. As much as I'd like to post screenshots and tell you that the barbs were wiping out whole nations it simply didn't happen. Oh well.
Once again, thanks for reading and following along with AI Survivor!