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 one of the closest finishes in AI Survivor history, with Pacal leapfrogging Hatshepsut on the final turn of the game turn snatch a Runner Up finish by a mere 4 points on the scoreboard. 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.
Special additional note for this game:
This was a rare occasion where I couldn't wait until the end of the season and felt the obligation to run the alternate histories right away to see how the map most commonly played out. My picks for Game Five were based on three main assumptions: Pacal would struggle with his seafood-heavy start, Brennus would be diplomatically unpopular and fall behind technologically, and those first two factors would cause Darius' Financial + Alive combination to make him the favorite. It was clear that Pacal and Brennus could only do well by mutually supporting one another, as the only two low peaceweight individuals in this field, and either one of them being weak would likely torpedo both of their chances. Thus it was enormously frustrating to watch how Pacal did struggle for the first 100 turns, how Brennus did fall behind technologically and have the worst diplomatic position in the game... only to see none of that matter as Pacal somehow climbed to the top of the scoreboard anyway while Brennus inexplicably survived to the end of the game. Meanwhile, Darius didn't expand well and was bullied by his neighbors into a First to Die finish. I had to know - how much of this was typical? What would happen from replaying the map a bunch of additional times?
This was what I found:
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 alternate histories revealed that my pregame analysis was a case of "close, but no cigar" to use the old phrase. The first thing that I watched during the map replays was the performance of Pacal. His weak opening in the actual Game Five was in no way an aberration, as Pacal sat at the bottom of the scoreboard over the initial 75 turns in game after game. The AI genuinely struggles with a seafood heavy capital for all of the reasons that we identified on Livestream: building too many work boats and then not using the seafood tiles after improving them, failing to build enough settlers, not having the happiness cap to make use of all that food, etc. The rest of Pacal's land was high in quality with a lot of river tiles and oodles of grassland after he was able to chop through the northern jungles. Even his capital was great later on once he could reach Hereditary Rule civic and get enough happiness to work all those water tiles. The problem was Pacal's early game where he inevitably started slowly each time, and in a normal game, this likely would have doomed him to second-tier status or First to Die candidacy.
Fortunately for Pacal, this was not a normal game, and his slow starts were almost never punished just as we watched in the real Game Five. Darius turned out to be the weakest AI leader on the board (more on this later) while Hatshepsut invariably found herself getting attacked by Brennus in virtually every game. This created space for Pacal to recover from his underwhelming openings, and by the time that Turn 150 rolled around, Pacal was almost always near the top of the AI rankings. With Darius a non-factor, it was indeed the Financial + Alive combination that dominated these games, only it was Pacal who was sitting in the driver's seat instead of the Persian leader. And it's not as though Pacal was running over the map in these games or putting on super impressive performances; while he did do that a couple of times in Game #12 and Game #17, most of the time Pacal didn't get much past 10 cities. This was reflected in him having fairly pedestrian Spaceship finishing dates which tended to fall around Turn 330; certainly not bad but also not exactly a blistering pace from a Financial leader. Pacal frequently never made it out of the southeast corner... and that was frequently enough to win by default against this suspect group of competitors.
Pacal also benefited enormously from the presence of Brennus as a low peaceweight ally to his northwest. It became clear almost immediately that Brennus was acting as Pacal's attack dog in game after game, roughing up the other strong economic leaders (Hatshepsut and Darius) and dragging them down into the mud while Pacal settled the jungle and merrily teched away. In this field of otherwise mostly peaceful leaders, Brennus might as well have been from a completely different genre of video game. He fought again and again and again, fighting 108 combined wars when Asoka was the next-highest at 73 total wars. This is even more impressive given that Brennus also had one of the highest elimination rates, since he couldn't start more wars when he was already dead. Brennus had the most offensive wars by a country mile and nearly had the most *DEFENSIVE* wars as well (42 compared to 44 for Hatty and Asoka). That should be impossible, having the most offensive and defensive wars simultaneously, but he nearly pulled it off.
Thus I was correct that Brennus would be drawn into a ton of wars and also that he would fall far behind the other leaders economically. This should have made Brennus the overwhelming favorite for First to Die instead of the Celtic leader essentially tying with Hatshepsut and Darius (all of whom were within one First to Die of each other in the alternate histories, 6 to 5 to 5). However, Brennus was repeatedly saved by the fact that this group of AI leaders was catastrophically bad at fighting offensive wars. The epic fails from Hatty that we watched on Livestream were absolutely typical in this regard as Brennus was saved over and over again in situations where he should have been dead. I can't tell you how many times I watched Hatshepsut or Asoka or Darius have a crushing edge over Brennus only to sign peace immediately rather than push their advantage. There were multiple games where Brennus faced an early 1 vs 3 situation and survived because the Hatty/Darius/Asoka trio were all peeing their pants with fear and desperate to sign treaties at the first opportunity. And of course these leaders all avoided military techs like the plague which repeatedly allowed Brennus to draw level with his opponent or even win wars against foes who had hundreds more scoreboard points and were more than a full era ahead in technology. Brennus absolutely should have been dead meat from his strategic position but instead was able to claw out a respectable showing in the alternate histories, largely driven by a whopping 18 kills.
Then there was Darius who simply flopped in this game when he came under the least amount of pressure. His central starting position proved to be weaker than most of the community had anticipated (myself definitely included), as he was never able to expand to the rich floodplain valleys to his east and west. Instead his territory always stretched out into a vertical line running from north to south, meaning that he lacked any safe backline cities and was always subject to being attacked anywhere at any time. I thought that Darius might have had some poor barbarian luck in the actual Game Five which wouldn't be repeated in the alternate histories, but nope, he was a true minnow in this scenario and always seemed to be at the bottom of the scoreboard. Darius was out-expanded by the other peaceful AI leaders and constantly disrupted by attacks from Pacal and Brennus which he could ill afford to fight. His own AI performance was dreadfully bad as well, failing to build settlers and refusing to research military techs as if he were trying to get himself killed. This certainly did nothing to disrupt the perception that Darius is a poor AI leader being carried by his amazing traits.
This should have left Hatshepsut an opportunity to grab the mantle of top-performing high peaceweight AI leader, and to her credit, she was definitely the strongest of the non-Pacal AI leaders in this field. Hatty won six different times across the 20 map replays, every time by Culture (not counting her Diplomatic victory which hit literally three turns before her third city was about to go Legendary), and she was the only leader aside from Pacal who was able to win with any kind of regularity. Given that Hatty was under much more pressure from Pacal and Brennus than her western compatriots, this was a far more impressive performance than anything they managed to achieve. However, because Hatshepsut was also under 2 vs 1 pressure repeatedly, she also suffered a lot of eliminations and was a frontrunner for First to Die status. Hatty could have done significantly better than this if she would have simply researched military techs and crushed Brennus rather than endlessly letting him off the hook with peace treaties. She played like she was terrified of her own shadow in the same frustrating fashion that we watched on Livestream. This might be the best stat highlighting her playstyle: Hatshepsut had six victories but only four kills. How is that even possible?! 
As for Ramesses and Asoka, they mostly tended to be irrelevant to the outcome of each game which was largely decided by what was going on in the east. The two of them spent a lot of time fighting one another which usually stalemated to no one's advantage, though there were a few games where one or the other was able to crush their rival and become a major player. Asoka was the boom-or-bust candidate of the pair, with a better chance at winning an overall victory tied to also having a higher chance to be outright eliminated. He fought with Brennu a fair bit and was usually stronger than the Celtic leader, though of course Asoka almost never managed an outright conquest of Brennus because he would leap to sign peace at the first opportunity. Ramesses was the only one of the six leaders who didn't win at least once in the alternate histories that I ran, ironic for the winner of the real Game Five, and instead wound up with a whole bunch of Runner Up finishes. This was almost entirely due to the fact that he was further away from Pacal and Brennus than anyone else, not so much due to his own actions. I would caution anyone jumping aboard the Ramesses train after watching Game Five as he was thoroughly mediocre across these alternate histories and didn't do a whole lot.
Now for a look at the individual leaders:
Pacal of the Mayans
Wars Declared: 37
Wars Declared Upon: 28
Survival Percentage: 90%
Finishes: 9 Firsts, 5 Seconds (55 points)
Kills: 12
Overall Score: 67 points
The alternate histories proved that Pacal was definitely the best performer in a rather shaky field. As I wrote above, Pacal winning by Spaceship was the default outcome for this map if nothing out of the ordinary took place. This was truly the Financial + Alive phenomenon in action since Pacal frequently didn't have a territorial or population lead, he simply out-teched the rest of the field and reached the end of the tech tree first. With that said, it was genuinely impressive how Pacal was able to recover over and over again from his weak starts and find his way into a pole position. Repeatedly I would start the AI playtest, step away from the computer for a minute, and come back to see Pacal in first place somehow. He had a clear edge in winning the most outright victories, and in games where he didn't win, he nearly had the most Runner Up finishes as well. Pacal therefore had about double the odds to advance to the playoffs as compared to anyone else in this field, and also lapped the competition in terms of pure survival with his 90% survival rate. I think that he would have been in dire trouble if there had been any competent fighters among the rest of this group, any opponent who could start snowballing across the map with conquests, but outside of two Brennus games that basically never happened. Although this wasn't a particularly flashy performance from Pacal, it was more than enough to get the job done in the middle of a clown fiesta.
Hatshepsut of Egypt
Wars Declared: 22
Wars Declared Upon: 44
Survival Percentage: 65%
Finishes: 6 Firsts, 1 Second (32 points)
Kills: 4
Overall Score: 36 points
All of the strengths and weaknesses of Hatshepsut as a leader were fully on display in the Livestream game that we watched. She has a great civ at her disposal in Egypt, she has excellent starting techs, she's reasonably good at expanding and developing her territory, etc. And then all of that good stuff largely gets wasted because she's so dismally bad at anything involving the military side of the game. I genuinely believe that she's worse than Gandhi in this regard as we've seen Mr. Non-violence score more than a handful of conquests over the years. Hatty was egregiously bad at snowballing her position across the alternate histories, allowing a weak and hostile Brennus to sit untouched on her border in game after game rather than absorbing his land. This put her in needless danger as Hatshepsut was frequently the victim of 2 vs 1 conflicts when Brennus joined together with Pacal. Hatty also suffered a series of First to Die knockouts when Brennus connected his copper resource at an early date and then blindsided Egypt with extremely early invasions, several of them taking place before Turn 60. These attacks often either crippled Hatshepsut or eliminated her entirely, which would more or less seal those games as Pacal victories. When she could successfully fend off attacks or find herself left alone for a while, Hatty was much more effective at scoring Cultural victories than Ramesses on this map, with the actual Game Five being a bit of a false positive in that regard. Still, it's hard to avoid the sense that this was another setup where Hatty could have and should have performed better; she's been unlucky not to score any points but has also been her own worst enemy too.
Brennus of the Celts
Wars Declared: 66
Wars Declared Upon: 42
Survival Percentage: 55%
Finishes: 2 Firsts, 3 Seconds (16 points)
Kills: 18
Overall Score: 34 points
Brennus played this game a bit like he was a rabid animal lashing out at everything in sight. As mentioned above, he launched 66 offensive wars despite also dying about half the time which limited the time available for military crusades; Pacal was next-highest at 37 offensive wars and the Mayan leader was basically always alive to start new conflicts. This resulted in the most kills in this set of alternate histories by a wide margin at 18, half again the total of the far-stronger Pacal. Brennus did much better than his strategic position should have warranted, largely thanks to focusing heavily on military technology and exploiting the glass jaws of his high peaceweight neighbors. He really should have died about 75% of the time instead of 45% of the time, it was insane watching how many times he squeezed out of dogpiles by getting Hatshepsut or Asoka or Darius to sign treaties when they were easily winning their wars. Of course, Brennus' aggression could still backfire and he was First to Die about as often as Hatshepsut and Darius. There's also a real chance that this data set is overstating the performance of Brennus, as he scored 27 points (with two victories and two Runner Ups) in the first 6 games that I ran, then the Celtic leader scored a mere 7 points and a single Runner Up result over the final 14 alternate histories. His true performance level is probably more akin to Asoka and Ramesses in terms of our scoring, though arriving there in a completely different fashion (more kills and fewer top two finishes). Even re-running the map 20 times remains a relatively small sample size.
Asoka of India
Wars Declared: 29
Wars Declared Upon: 44
Survival Percentage: 45%
Finishes: 2 Firsts, 3 Seconds (16 points)
Kills: 8
Overall Score: 24 points
Asoka spent most of his games locked in a duel with Ramesses which typically neither of them was able to win decisively. He didn't fight with Brennus as much as I'd been expecting, and when they did engage in warfare (almost always started by Brennus), the two of them tended to peace out without either side achieving a full conquest. Asoka somewhat surprisingly didn't win either of his games by Culture, taking once unexpected Diplomatic victory in a close Game #11, and then out-racing Hatty's culture with a Spaceship ending in Game #16. He wasn't too impressive across these alternate histories, as Hatshepsut was usually the strongest of the peaceful AI leaders even though she faced more attacks. Most games saw inconclusive fighting between Asoka and Ramesses that went nowhere, and then eventually Pacal would win by space or Hatty would win by culture. Asoka has been described as the bargain bin version of Gandhi, playing the same style of game but doing everything worse, and these alternate histories backed up that sentiment.
Ramesses of Egypt
Wars Declared: 35
Wars Declared Upon: 22
Survival Percentage: 75%
Finishes: 0 Firsts, 6 Seconds (12 points)
Kills: 4
Overall Score: 16 points
Ramesses was clearly the lesser of the two Egyptian leaders on this map. He always settled the same eastern floodplains spot for his second city, the same one that we watched on Livestream, but it was pretty random in terms of whether he landed an early religion. Ramesses had the most sheltered starting position on the continent as he was attacked less than anyone else, largely due to being the furthest away from the psychotic Brennus. This did not translate into him being a strong high peaceweight leader, however, as Hatshepsut clearly did better from a tougher spot and even Asoka was probably a little bit stronger as well. Mostly Ramesses found himself locked into lengthy and pointless conflicts with Asoka, wars that went nowhere and rarely snowballed one side into a strong position. The Ramesses that we watched on Livestream was more powerful than the Ramesses in most of the alternate histories, and of course Ramesses won the real Game Five while none of his clones managed a single win in any of the 20 map replays. Nor did they really get that close either; while Ramesses had a ton of second place finishes, he wasn't a true competitor to beat Pacal to space or Hatty to culture. He was Runner Up behind Brennus in both of the Celtic leader's wins just because Egypt was the furthest nation from the Celts. Overall, this was a pedestrian series of performances from Ramesses, not great and not terrible, somewhere in between.
Darius of Persia
Wars Declared: 25
Wars Declared Upon: 34
Survival Percentage: 35%
Finishes: 1 First, 0 Seconds (5 points)
Kills: 1
Overall Score: 6 points
These alternate histories might have sealed the deal on Darius being the most underperforming leader in Civ4 AI Survivor. Despite his awesome traits and strong Persian civilization, Darius simply cannot seem to perform unless he gets handed a godly corner starting position where he can be safe from aggression, as in Season Seven when he made his run to the Championship game. Starting in the middle of the map was utterly disastrous for him in this game, as Darius was pressured from all sides where he wilted like a garden in drought. He failed at settling the map in timely fashion, he failed at researching military techs, he failed at defending himself from attack, and on the rare occasions where he had a military edge, he failed to press his advantage because he usually signed peace at the first opportunity. His one victory came in the atypical Game #15 where Brennus was crippled at an early date and Hatshepsut fought a long war with Pacal to slow the Mayan teching and clear Darius' path to space. Darius scored all six points in that one game or else he would be sitting on a big fat zero from the alternate histories. It was a really, really bad set of games from him and the Persian leader couldn't even manage the "Alive" part of the Financial / Alive combo. Good grief!
What a bizarre game this one was; I'm still not entirely sure how to feel even after watching 20 map replays. Credit certainly goes to Pacal who made the most of an awkward spot and proved once again that he was the best performer out of this dubious crowd. As for everyone else... it honestly felt like they weren't even trying to win across most of these games. Hatshepsut refused to conquer her weak neighbors, Darius refused to defend himself, and Brennus engaged in the most self-destructive behavior imaginable while somehow not getting punished for it. I have less respect for everyone in this game as a result of running these alternate histories. Maybe I need to go watch some Mansa Musa highlights to remind me that there are a few capable AI leaders in Civilizaton 4. 
Thanks as always for reading, I hope you enjoyed this look back at Game Five!
