Sullla's AI Survivor Season Three Alternate Histories

[Genghis the Barbarian]

S3P2 Writeup

A shocking result, to say the least. I also encourage readers to ponder the Diplomatic victory question.

Overview

To use my best Sullla voice: Whaaaaaaaaaaaaaaat?!?!?!

I mean, seriously. Wat?!? Gandhi winning the most games by a good amount? Against this hostile field? With Mansa Musa in the game, who looked like a massive disappointment here? De Gaulle being the 3rd best leader in this setup? Five Diplomatic finishes, all from the same two leaders? Minimal strong teching performances despite this star studded cast? The Actual Game turning out to be atypical? What the heck happened here?

Map Dynamics

This was not your typical 2v4 Good vs. Evil showdown. For starters, most of these leaders had builder-oriented personalities, and although they fought early and often, their overall approaches to these games were centered around internal development rather than external conquest. The only true warmonger in this field was Peter, and, well, the results made it blatantly obvious that he was the odd man out here. The first tsar was a major wildcard in these games despite, or perhaps because of, his pathetic showing. He could be useful to the evil leaders in two ways. First, he could do his part working with the evil leaders to help take down the Gandhi/Mansa tandem, like in Games 18 and 19. Alternatively, he could be so incompetent that he becomes an easy source of extra territory for one of his evil neighbors, whether it was from gifting over culturally crushed cities, fudging his expansion, or suiciding into a much stronger leader. Unfortunately for the baddies, Peter was more often than not a troll, doing nothing except attack his own brethren and being a roadblock between evil leader cooperation. Team Good more often than not overcame this hostile diplomatic environment, winning eleven games altogether and coming quite close to winning a few more, and I honestly think the evil leaders would have won more games without Peter (i.e., just in a 3v2).

Of course, it helped that Team Good happened to consist of the best economic leader and the best culture-monger in Civilization IV, both of whom had the two best starting positions in the map regarding land quality and quantity. Mansa's land was rich, while Gandhi was all but guaranteed a nice peninsula to his east that secured him 3-4 extra cities. The elephant in the room: it was Gandhi who got the king’s ransom of the win-share. Thankfully, there is an easy explanation: Mansa’s central position was a deathtrap, giving all four of the bad guys easy access to him. In many games, Mansa would be so embroiled in war that all he could do was serve as a meat shield for the Indians. As Gandhi’s eight wins can attest, Mansa was a fantastic meat shield, but it came at a cost nonetheless: Mansa only won three games and was either dead or utterly broken in most of the other games. Surprisingly, the biggest hurdles to high peaceweight success were Mansa and Gandhi themselves. The two frequently held different faiths, sparking conflict between the two that derailed multiple of their games.

Leader Dynamics

There were three primarily antagonistic pairings between leaders that affected these games. The first one was the Pacal-Peter pairing. In most games, Peter attacked Pacal early in a desperate attempt to gain some territory. This usually backfired on the Russians, and Peter was Pacal's springboard to the Championship game. Also, Peter’s propensity to avoid culture haunted him greatly, and there were multiple instances of him gifting over culturally crushed cities to the Maya. However, there were a couple of scenarios where he was instead the bane of Pacal’s presence. Sometimes, Peter was the catalyst to a dogpile that crippled the Mayans, like in Game 5 when he sparked an early 4v1 that led to Pacal’s one death in this set:

Now you know how it feels!

Almost equally as disastrous for Pacal’s winning prospects was if another leader, say De Gaulle, were to conquer Russia instead. Game 13, in fact, saw Mansa conquer Peter instead, permanently relegating Pacal to the role of second fiddle.

The next pairing was between De Gaulle and Mansa, and oh boy. I would never harbor such sheer hatred for my worst enemy. These two fought, frequently to the death, in every single game, with De Gaulle more often coming out on top. There were two main reasons. First, Mansa was THE dogpile magnet, and there were many instances where Mansa’s army was struggling all the way in lalaland while De Gaulle had a fresh army at the ready to topple Mansa for once and for all.

There is an imposter among us here…

Secondly, De Gaulle’s land had a TON of production, while Mansa’s land was on the flatter side. De Gaulle's production and terrain advantage was a major struggle for the Malinese in these games.

That’s enough hammers to give one Gaullestones

Like Hatty in Opening Round Game Eight, it was almost always a De Gaulle invasion that broke Mansa’s back. Upon killing Mansa, De Gaulle’s success hinged on if he could take out Gandhi early enough, or if Gandhi was already too far ahead to be stopped. Mansa would only come out on top if he had an unusually strong opening, if De Gaulle had an unusually weak one, or if De Gaulle attacked Mansa too late for his massive numbers to overcome Mansa’s tech disparity. Less commonly, De Gaulle merely served to soften up Mansa for Pacal to take the spoils instead, a major factor in Pacal's best games.

Finally, there was the Qin v Gandhi conflict, one that often determined if anything else even mattered. Qin and Gandhi’s rivalry began long before the first war horns blared out announcing their conflict, as the two would embark on a settling race, especially over the aforementioned peninsula and the barbarian settlements that sprouted there. Gandhi usually won this race, as Qin's expansion was generally slow due to his isolated coastal start and his penchant to go after early wonders (amusingly, Qin built The Great Wall in all twenty games). To say the least, it was a disaster for Team Good if Qin won the race instead. Like De Gaulle and Mansa, Gandhi and Qin fought in virtually every game, although Qin was a slightly more forgiving opponent than De Gaulle. If Gandhi was able to stalemate (or outright win), then his victory was all but assured.

How Typical Was the Actual Game?

4/10. In most games, Mansa's diplomatic situation was too much for him to overcome. A more typical game would see Pacal kill Peter, followed by the three remaining evil leaders to continuously deal hit-and-runs to Mansa until he collapsed or became permanently crippled. From there, two scenarios could take place: either the evil coalition took Gandhi down, with the victor generally being the largest leader no matter who the tech leader was (more on that later), or Gandhi was just too far ahead to be stopped.

Any deviation from the script, however, usually led to a Mansa victory. Mansa's livestream victory is easily explainable: Mansa and Pacal worked together as religious allies to take down the rest of the world (the two almost always had differing religions in the replays) while Gandhi turned out to be the dogpile magnet and did a shoddy job defending himself even from cross-map wars. Regarding his Alternate Histories wins: in Game 5, Mansa was the prime beneficiary of Pacal's only demise, while his Game 13 win was a strange game where everyone generally played poorly and Mansa just kind of won by default. Meanwhile, his Game 15 win was the single most spectacular performance I have, and will have, ever witnessed by a leader in Civilization IV: he mustered a victory despite getting attacked an unfathomable ELEVEN TIMES.

An Addendum: Diplomatic Victories

Most Alternate Histories sets have 1-2 random Diplomatic finishes which stem from two scenarios:

1. Coronating the obvious game winner 30 or so turns before that leader would have won anyway – this is the more common scenario.
2. A backdoor troll scenario where a much less deserving leader wins because he had better diplomacy with the non-ballot leaders, while the non-ballot leaders were relevant enough to play kingmaker – we always remember these cases, but they occur far less frequently than we think.

However, this was one of the rare Alternate Histories in which a Diplomatic finish actually felt like a legitimate win condition rather than an interruption or a chaos agent, as there were five Diplomatic wins split between De Gaulle and Qin which were achieved in exactly the same manner. For readers, my hope is that this set shines a spotlight on the debate of if the Diplomatic victory should remain a win condition for AI Survivor. The argument against the inclusion of the Diplomatic victory: had it been disabled, the Runner Up finisher would have won at least three of the five Diplomatic finishes. Even the two “rightful” UN victories were questionable: in Game 1, it looked like De Gaulle’s much larger empire would help him catch up to Pacal in tech, but that was no guarantee, while in Game 10, Qin’s election as World Leader interrupted a down-to-the-wire race to Alpha Centauri between the French and the Chinese; both leaders had launched their spaceships on the same turn, and perhaps this should be a major ding against the inclusion of the Diplomatic victory condition as it robbed us of an exciting space race.
However, with the exception of Game 12 where Gandhi was legitimately shafted out of a win (although it was partially his fault for beelining Mass Media before Rifling ), every game with a UN finish saw the most deserving winner come up on top. De Gaulle and Qin are not the types of leaders who can just sit back and coast on Financial like Pacal. They had to go out and earn every scrap for themselves, without being born with a silver spoon in their mouths, and while Pacal arrogantly sat back thinking that his Financial trait and a weak Peter suiciding into him was enough, De Gaulle and Qin put in all the hard work taking down Mansa and/or Gandhi. They may have been too far behind in tech to win in any other way, but they still played the best, and I could not conceive of any Diplomatic finish as a troll ending save for Game 12.

Religion had a minimal impact on games, as in the vast majority of cases the Evil leaders all shared a religion. If anything, religion had the potential to sow the seeds of division for Team Good. This was a violent world filled with stalemated wars, as these leaders were generally evenly matched in military tech, considerably slowing down the tech pace despite this field of elite techers. Mansa’s economy was especially mediocre by his lofty standards, although that had a lot to do with his nearly 5.5 wars per game average. With Gandhi and Mansa winning eleven games, it was no wonder that Cultural finishes dominated, but I still found it crazy that there were as many Diplomatic finishes as Spaceship and Domination combined.

Onto the leader summaries:

Pacal II of the Maya
Offensive Wars: 37
Defensive Wars: 33
Survival Rate: 95%
Finishes: 4 Wins, 10 Runner Ups (40 Points)
Kills: 16
Overall Score: 56

I thought this would be a Pacal romp, with a few Mansa or Gandhi wins sprinkled here and there and the occasional random De Gaulle or Qin victory. However, this was only a romp as far as Runner Up finishes were concerned. Regarding actually winning, Pacal had to overcome three hurdles:

1. Gandhi having a Gandhi game and winning
2. Mansa having a Mansa game and winning
3. De Gaulle or Qin getting large, building camaraderie, and then using the United Nations to pull the rug underneath Pacal.

In sixteen games, the three above factors proved too much for Pacal to overcome, forcing him to settle for second or for outright elimination. In the most nightmarish of cases, Pacal found himself completely shut out of the Championship despite being 2-3 techs away from building all of his spaceship parts. The Mayan leader did little to dispel the notion that he is an otherwise mediocre leader blessed with a golden package. His overall inactivity cost him dearly in many cases and was a crucial factor in the UN dynamics of this set, as he found himself in the awkward circumstance of having a small empire yet also being the tech runaway. As a result, when he inevitably reached Mass Media first and built the UN, the election would be between him and a leader like De Gaulle with, say, 35-40% of the world population, while there was a third party, say Qin, with around 25-30% of the world population who thus held the kingmaking power. Since Qin and De Gaulle had done most of the heavy lifting against Mansa and Gandhi, they had built up an immense amount of mutual military struggle bonuses, making them into steadfast allies, and one would crown the other as victor right under Pacal’s nose.

All of this was avoidable if Pacal was able to muster any sort of military initiative. Yes, he had 16 kills, but a good chunk of those came from Peter having drank too much vodka in these games. Naturally, Pacal’s four wins came in the games where he actually got his hands dirty and conquered some fools. Pacal is one of the game's very best economic leaders, but this set demonstrated why he does not quite belong in the absolute top tier of leaders. For a Financial leader with one of the best packages in the game, a dream diplomatic setup, and two easy neighbors to kill, Pacal should have performed far better than 56 points and four wins. There is a reason why he is a leader that we root against in every game.

Best Performance: Game 6 demonstrated Pacal’s sky high ceiling, as he actually leveraged his economic advantage and murdered the rest of his competition.

Worst Performance: I can give a pass to his one death – there is nothing anyone can do about an early 4v1. More embarrassing was his Game 12 performance, where Peter caught him without metals, leaving Pacal a complete afterthought.

Hare Award: Pacal should have easily won Game 11 had he not taken a nap with his Mechs and Modern Armors and allowed a backwards Gandhi to win by Culture.

Democratic People’s Republic Of Wang Kon Award: De Gaulle was for once the tech runaway in Game 10… but Pacal baited the French into attacking him and devoting his entire production to military rather than research, allowing Qin to catch up in tech (with some help from the Internet). Then, when De Gaulle was about to run over Pacal, the Mayans used the UN to end De Gaulle’s conquest, keeping it in Mayan hands. De Gaulle then voted Qin as world leader on the same turn he launched his spaceship, and Pacal had completed a successful troll.

Gandhi of India
Offensive Wars: 13
Defensive Wars: 59
Survival Rate: 50%
Finishes: 8 Wins, 2 Runner Ups (44 Points)
Kills: 2
Overall Score: 46

For once, Gandhi was not a lamb to the slaughter in a playoff game! On the contrary, he was the dominant leader, even if the score does not indicate this. To start, Mansa Musa should never have to pay for his drinks in Delhi ever again. Mansa had Skirmishers to shelter Team Good from early dogpiles, and his economic skills kept his tech going despite being mired in a brutal struggle for his survival from Turn 100 and on. Mansa’s resilience was crucial in helping Gandhi get those precious few extra turns that meant the difference between victory and destruction. Put literally any other leader in Mansa’s spot, and Gandhi would have been dead meat in this setup. This is not to say Gandhi did not deserve his success He executed his part of the bargain to near perfection and proved actually capable of defending himself if need be (unlike in the livestream). Normally, being in culture mode from Turn 0 is a weakness, but here, every turn mattered for Gandhi. Game 20, for example, saw Gandhi get three legendary cities just as Tanks from the Sino-French coalition army were outside of his 3rd city – one more turn, and Gandhi was done for. I noticed some utterly absurd cultural beelines: getting Meditation AND Polytheism, going for Philosophy before getting the crucial defensive Longbows, and which I had already mentioned, going for MASS MEDIA before RIFLING, a move that directly threw away a certain victory in Game 12. Despite this extreme display of culturephilia, Gandhi’s teching was excellent (perhaps aided by the immense amount of room for cities he had), and he was more than capable of helping with military matters; he really was more aggressive than his 13 offensive wars may have suggested.

Unsurprisingly, Gandhi either won, almost won (he would have won Game 18 had he turned on the slider), or died. There was three ways Gandhi could falter. First, he and Mansa could come to blows due to religious differences, fracturing Team Good and making it easy for Team Evil to win. This was not necessarily a death knell for Gandhi’s chances, as there were two games where the two fought and Gandhi still won, one of them being the oddball Game 2 where Indira took the helm. Nevertheless, fighting his only ally was not good for his prospects. Second, Mansa could fall apart too quickly, meaning that Gandhi was next on the chopping block – this was the most common cause of Gandhi’s failures. Finally, Gandhi could be his own worst enemy at times. To start, he was prone to crashing his economy if he combined over-expansion with the neglection of essential development techs like Wheel and Pottery (Game 13). To mitigate these early research struggles, I noticed Gandhi quite effectively utilized a failgold economy (for those unfamiliar, this means to use gold from incomplete wonder builds to fuel research – a common technique on higher difficulties) in these games. Gandhi could also fail to properly defend himself, like in Game 15 where he died to a cross-map invasion from De Gaulle. Nonetheless, this set showed why Gandhi is an elite culture-monger who would be a top five leader in the game if it were not for his extreme peaceweight. Unfortunately, that version of Gandhi did not show up in the livestream.

Best Performance: Game 14 was a well-executed Cultural victory, especially coming from a non-Financial leader.

Worst Performance: Crashing his economy in Game 13, failing to expand, and dying one turn before Mansa’s victory triggered. Dishonorable mention to Game 15, for reasons already mentioned.

Tortoise Award: With some help from Pacal’s inactivity, Gandhi won a Culture victory in Game 11 without ever turning up the Culture slider.

No More Mr. Nice Guy Award: Look at his Game 2 statline! At first glance, his zero kills may have made one think that Gandhi should stick to his pacifist ways, but he should have had at least two kills in that game. His first one was stolen when Peter troll sniped Qin in a last-minute vulture of a dying civ, and Gandhi was in the process of running over De Gaulle and Peter at the same time when he hit three legendary cities.

Why You Should Never Give The Nice Guy A Chance Award:

De Gaulle of France
Offensive Wars: 60
Defensive Wars: 20
Survival Rate: 90%
Finishes: 3 Wins, 5 Runner Ups (25 Points)
Kills: 17
Overall Score: 42

Qin Shi Huang of China
Offensive Wars: 54
Defensive Wars: 19
Survival Rate: 80%
Finishes: 2 Wins, 3 Runner Ups (16 Points)
Kills: 4
Overall Score: 20

Although there was a large disparity in score between these two leaders, I grouped them together as they were for all intents and purposes on the same team in this setup. Even outside of map dynamics, these two leaders share many similarities. Both leaders have excellent starting techs and like-minded personalities, both being Industrious low peaceweight backstabbers who prefer more builder-focused strategies. Qin has the blah Protective while De Gaulle has the decent Charismatic as a second trait, but Qin compensates for this trait disparity with better uniques. Regarding the map, both leaders had Fishing starts and at least one dogpile candidate as a neighbor. These two were not strange bedfellows in these replays – despite being conniving backstabbers situated on the opposite corners of the map, their objectives and pathways to victory were in perfect tandem with each other. All of their wins followed the exact same pattern – run over one or more of their neighbors, wait for Pacal or Gandhi to build the UN, and then rig the UN in their favor. Yet another sign of how similar the two were: they had virtually identical offensive to defensive war ratios of 3:1. At the very least, one leader could get into the championship game by killing one of their neighbors and leveraging that into a Runner Up finish. Their strategies reminded me of a human Deity player – they tended to break out with Cuirassiers (De Gaulle’s two move Musketeers were quite useful with this), before splitting enough territory to win through the UN. Their failures stemmed from the following: taking too long to conquer, being unable to coordinate dogpiles, and eventually getting wrecked in the late game by a much more advanced enemy.

The main reason why De Gaulle’s score was so much better: his land and neighbor situation. While Qin had to contend with Pacal and the hyper-cultural Gandhi as neighbors (and was First To Die a couple of times due to this tough neighbor situation), De Gaulle had the hapless Peter and the more vulnerable Mansa as neighbors, the latter of whom was a more consistent dogpile magnet than Gandhi. Moreover, De Gaulle’s capital and surrounding land was better, especially regarding growth and production. Qin either needed to conquer Gandhi in due time or effectively use a Great Lighthouse-Colossus economy in order to have a chance. Switch Qin and De Gaulle’s starting positions, and their results would likely be flip-flopped. Altogether, this was a better performance from the two than I was expecting, and perhaps the community especially underestimates De Gaulle. The French leader has flailed around more often than not, but this season did demonstrate that there are situations where he can succeed.

Best Performances: Games 1 and 10, respectively, as those were the games where the two would have had a chance to win without the UN.

Worst Performances: De Gaulle’s expansion was moribund in both the games he died, and Qin tried to backstab Pacal in Game 18, only for it to completely backfire in his face.

Civil Disobedience Award: In Game 14, De Gaulle (and Pacal, and the United Nations) were about to finish off Mansa… when his last city became ensconced in Gandhi’s borders. Since Gandhi refused to sign Open Borders, De Gaulle was stuck in a forever war, and Gandhi was safe to pursue his patented Cultural victory.

South Park City Wok Award:

Gotta keep out those darn 北京人.

Mansa Musa of Mali
Offensive Wars: 26
Defensive Wars: 93
Survival Rate: 40%
Finishes: 3 Wins, 0 Runner Ups (15 Points)
Kills: 4
Overall Score: 19

Poor, poor Mansa. 93 defensive wars says it all. His games were painful to watch, as the Malinese leader would frequently explode out into what appeared to be a dominant position, until the first of what would become a relentless barrage of attacks took place, taking down the titan through death by a thousand cuts. In every game, Mansa was teetering on the knife’s edge, never sure if the latest invasion would cause the whole moneybags machine to collapse. Even if he withstood these invasions, he was more often than not a husk of himself, failing to advance in more than half of the games he survived.

With that said, it was still clear why Mansa is one of the best leaders in AI Survivor. I do not think any other leader is capable of winning 3/20 games while being attacked 93 times. In fact, they would be lucky to win one game. Mansa only had one easy game, that being Game 5 where Pacal uncharacteristically died early. In the other 19 games, Mansa had one major issue: this was not the right field for culture-monging, not in this hostile field and especially not with Gandhi in the game. Mansa could have done better had he played more like he did in his opening round game, rather than trying to build missionaries while being in a 3v1 and still missing out on the cultural milestones. When Mansa did exhibit aggression, it was not always smart – he attacked Gandhi more often than I thought he would, and taking out his only potential ally tended to backfire on him. At the end of the day, Mansa’s central position proved to be a deathtrap.

Best Performance: In Game 15, when he faced 11 invasions, Mansa had to contend with a 3v1 on five separate occasions. Also:

Worst Performance: Game 1 was perhaps the one game where Mansa looked extremely pedestrian, and was deservedly First To Die.

Living Long Enough To Become The Villain Award:

Peter of Russia
Offensive Wars: 40
Defensive Wars: 6
Survival Rate: 25%
Finishes: 0 Wins, 0 Runner Ups (0 Points)
Kills: 1
Overall Score: 1

I have read Alternate Histories sets where leaders scored zero points, where leaders died in all twenty games, and where leaders squandered starting positions that would make a human Deity player jump in joy. Yet, there is a valid argument that this may be the single worst Alternate Histories performance ever. To start, look at the war counter. Despite only being attacked a total of six times – that’s 0.3 invasions per game – Peter died in three quarters of the replays, was the overwhelming favorite for First To Die despite being in the peaceweight majority, did not ever come remotely close to advancing to the championship game, and would have laid a complete egg had he not randomly sniped what should have been Gandhi’s kill in Game 2. This was an utterly inexcusable performance for a warmonger with a dream diplomatic setup and golden dogpile opportunities.

There were two major problems with Peter here. First, in this field, he might as well have been Ragnar or Genghis Khan. Indeed, if you replaced him with one of those leaders, there would have been little difference (actually, I think those guys might have performed better). For whatever reason, Peter would stop expanding after five cities in order to plot a war he had no chance of winning, whether it was against one of his neighbors or a fruitless cross map venture. Most games saw him either launch random attacks without rhyme or reason, get absorbed by one of his neighbors, or derail the evil gameplant with troll war declarations. His only usefulness to the bad guys was his being so incompetent that he became a mechanism for extra territory for his evil neighbors. The second issue, and perhaps one that mitigates this horrific output: Peter had BY FAR the worst land. He had a coastal capital without coastal resources, little room to expand (exacerbated by his poor cultural output), and lots of jungle with land that was not good enough to compensate post-Iron Working. His land was so bad (especially relative to his rivals), I think Huayna Capac would have struggled in his position. Nevertheless, although there is some debate over if Peter is an underestimated or overrated leader, I see him perfectly rated as a mediocrity.

Best Performance: Surely, you must be joking…

Worst Performance: Game 1, where his expansion was so awful that De Gaulle was able to jam border cities right next to Moscow.

Temujin Award:

Three cities on Turn 73…

Conclusions

This was yet another fascinating set, full of twists and turns, some shocking results, and some of the most exciting and awesome individual games I had the pleasure of watching. Some people enjoy seeing pure dominance, while others like to see evenly matched Alternate History sets – as I have done more, I personally find my favorite sets to be the ones with clashes to the death between Good and Evil, like in both this and the first playoff game. Some final food for thought: maybe we underestimate high peaceweights a little bit? This was not the first set in this season where the high peaceweights were able to overcome long odds to find success. For readers who have not yet made it to future seasons: do not be surprised when the goodie two shoes of Civ IV eventually get their time to shine.

 

[Thrasybulos]

Alright, kinda makes more sense that the read I posted.
I knew Peter would be bad (bad AI + worst start on the map), I didn't think he'd be 55% FTD bad.
A bit surprised at Qin's only 4 kills too.

About the Diplomatic Victory Condition, I've expressed myself on the subject a lot, but to recap my stance:

• I disable it in my runs because I'm interested in measuring up AI performance, and the AI is not programmed to use the UN (if it is, the code is buggy enough to be undistinguishable from random dice rolls).
  Basically, the UN just adds random noise to the measurements.
  • An AI won't call the vote when it can win (or delay it for ages, sometimes until too late)
  • It will call the vote when it can't win
  • It won't adapt its diplomacy to an attempt at winning
  • Votes and calls for resolutions are nonsensical (vote "No" to end a war it's losing badly, etc...)
• I call it the "troll winning condition" in reference to AI Survivor Picking Contest
  Even in games (like this one, apparently) when diplo is more prevalent, it's never the most likely winning condition
• The UN should definitely remain enabled for AI Survivor
  It may be bad (imo) for AI ranking purposes, but AI Survivor is first and foremost a show, and UN drama is an integral part of the show.

 

[Genghis the Barbarian]

A bit surprised at Qin's only 4 kills too.

Gandhi was a much, much tougher foe than expected, and often had more cities than Qin. Also, Qin was dead last in turn order which hurt a bit in claiming dogpile kills.

About the Diplomatic Victory Condition, I've expressed myself on the subject a lot, but to recap my stance:

• I disable it in my runs because I'm interested in measuring up AI performance, and the AI is not programmed to use the UN (if it is, the code is buggy enough to be undistinguishable from random dice rolls).
  Basically, the UN just adds random noise to the measurements.
  • An AI won't call the vote when it can win (or delay it for ages, sometimes until too late)
  • It will call the vote when it can't win
  • It won't adapt its diplomacy to an attempt at winning
  • Votes and calls for resolutions are nonsensical (vote "No" to end a war it's losing badly, etc...)
• I call it the "troll winning condition" in reference to AI Survivor Picking Contest
  Even in games (like this one, apparently) when diplo is more prevalent, it's never the most likely winning condition
• The UN should definitely remain enabled for AI Survivor
  It may be bad (imo) for AI ranking purposes, but AI Survivor is first and foremost a show, and UN drama is an integral part of the show.

All reasonable and I believe the UN and Diplo victory definitely belong in AI Survivor, another argument in favor is that it can save us 30-50 turns of uninteresting next turning in some scenarios

 

[Eauxps I. Fourgott]

I figured Mansa had to be the win with 11 defensive wars, but I would've bet on him being the 40% winner (I didn't get around to analyzing any of the other result lines - a lack of links to the original game materials certainly didn't help with that, by the way!) Definitely an unusual result.

Diplo should 100% remain enabled for AI Survivor because it's a core part of the game of Civ 4 (unlike the AP which was tacked on in an expansion) and more often than not makes things more interesting and/or entertaining when it does come into play.

 

[Genghis the Barbarian]

Speeding these up a bit as I want to finish S3 before 2025 at least.

Playoff 3 teaser! Recall in the actual game that JC and KK, two leaders with last names that have become synonymous with king, had partitioned the globe for themselves and only THEN turning on each other like the rabid dogs they are, Caesar ultimately winning the space race.

Here we have yet another one of these "Big Three" types of sets. The question is, for those of you who followed 2010s NBA, which group constituted the Miami Heat Big Three and which one constituted the Brooklyn Nets "big three"?

 

[Eauxps I. Fourgott]

I'll guess that this one played out largely along peaceweight lines. The big question is who's the odd one out? The fact that only one leader had a high kill total suggests that it's one of the two who did well in the actual championship... so I'll guess that JC flounders without the extra starting techs. That leaves Kublai to do much of the heavy lifting militarily, although he struggles to actually outpace Hannibal doing Financial things or Louis doing culture.

Hannibal is Financial, has good room to grow militarily, and doesn't border three warmongers, so I like his odds better than Louis's. Of the bottom three, JC seems pretty likely to be the one with the kills, while Rams probably squeezed out two wins and Burger King did not much of anything.

Final answer: Hannibal, Kublai Khan, Louis XIV, Julius Caesar, Ramesses II, Charlemagne

 

[Thrasybulos]

Hannibal borders both marked men, so that would indeed make him the de facto favourite. He has an exposed central position, so you'd expect things to go very wrong for him sometimes though.
That kinda fits the first entry.

Louis has a central position and borders none of the high peaceweights: he seems the least likely to do well among the low peaceweights.
Kublai's start I've already discussed: lots of room + no commerce makes for a dangerous proposition early game.
And Caesar's, according to Keler's data, wasn't that good either.

1. Hannibal
2. JC
3. Kublai
4. Louis
5. Ramesses
6. Charlie

(but honestly, 2 & 3 could be reversed).

 

[a pen-dragon]

Hanni - JC - Louis - Kublai - Rammy - Charly

Hanni is extremely close to Rammy, who has great land but is a clear PW outcast, along with Charly, and Charly. Going for Charly would probably be more difficult. Nonetheless this is a great advantage for him. Some FTD are on the cards due to a central start and Rammy having close horses. After the early game Financial carries him to win most of the games.

JC probably is the one going around the map killing people.

For the third good AI I am unsure. Louis has much better Land than KK, but is way more threatened. Basically Louis has to survive and beat Hanni, and KK has to beat Louis and then beat Hanni.

IMHO, the fourth having nine kills indcates that this is Kublai and Louis is third.

Rammy can probably get the odd culture victory, making him fifth.

Edit: spelling

 

[Genghis the Barbarian]

S3P3 is ready! @Eauxps I. Fourgott had an excellent read of the map (and the overall capabilities of these leaders), and this was a tough one that certainly surprised me a lot! Perhaps there is merit in the belief that ALL of the French leaders are quite underrated...

Also, thank you Eauxps, I foolishly completely forgot to link Sullla's writeups, which I will do in all the other posts. Whoops!

Anyways, here it is!

S3P3 Writeup

Overview

Although I expected the removal of Deity starting techs to nerf Julius Caesar, I was not anticipating that it would be to this extent. In these Alternate Histories, the Roman dictator was a complete buffoon, never ever coming close to even backdooring into the Championship game, let alone winning. What in Jupiter's name happened to one of AI Survivor’s OG titans, perhaps one of the most significant men in all of Western civilization? Well, Caesar was an odd man out in this field, despite his perfect peaceweight for this setup. Of his five rivals, four were major culture pumpers while the fifth, Hannibal, had the Financial trait and was no slouch at border expansion himself. All the Roman dictator could lean on was his Praetorians, but that is a segway to the second factor (the use of modern AI Survivor settings), which will be discussed in the individual leader section.

As it turned out, this ended up being another one of those “Big Three” Alternate Histories. The story of this map was that of Louis, Kublai Khan, and Hannibal being the alpha males who bullied and pansted the three beta males, that being Ramesses, the Burger King, and the aforementioned Caesar. Some stats to highlight this disparity:

1. The Big Three combined for 18 wins, with the two outliers (both Ramesses victories) coming when Louis and Kublai dragged each other out of the game with early warring.
2. The Big Three all survived the majority of these replays; the three stooges never exceeded a 35% survival rate.
3. 15/20 games saw two members of the Big Three occupy the top two spots. All except for one game (Game 4) saw at least one of the Big Three advance to the Championship Round.

In this crime syndicate, Louis was the big bad boss, the man whose ring everybody else had to kiss. Blessed with the best land, the easiest access to the lowest runt (Caesar), and the quickest path to victory, Louis lived up to his Sun King nickname, every game revolving around his orbit. If Louis was Don Eladio, Kublai Khan was Hector Salamanca. The Yuan emperor was an absolute brute on this map, netting himself a staggering 27 kills (albeit a lot of them were sniped from more deserving leaders), and taking advantage of his soft Egyptian target to his east. In game after game after game, KK smacked around Ramesses, encroached on his land, stole his lunch money, slept with his sister-wife, and then strung Ramesses upside down on top of a pole for the rest of the world to see. Following his dismemberment of the Egyptians, KK generally served as the mafia enforcer, although once in a while he could launch a coup of his own and establish himself as the boss. Meanwhile, even though Hannibal ostensibly lacked much in the way of quality land, his Gus Fring-like patience and ambition more than made up for it, as he was frequently able to break out of his initial funk and quietly leverage his economic skills into a strong performance. Moreover, he had easy access to two of the three wannabes of the map and was quite sound in his decision making, more so than his two mafia counterparts. There was an intriguing rock-paper-scissors dynamic amongst the Big Three. Although there were exceptions, as a general rule of thumb, Louis would culturally overwhelm the Carthaginians, Hannibal would out-tech the Mongolians, while Kublai would lay the wrath of Khan unto the French.

There were a few factors that separated the alphas from the betas. First was that in a world where space was limited, the Big Three excelled the most at the land-grabbing stage, whether it was from sheer settling, stretching their culture, or getting first dibs on barbarian cities. Conversely, the other leaders either got too hung up on mediocre wonders (Ramesses), got into war mode prematurely (Caesar, Burger King), had slow starts stemming from crappy land (JC, BK), or struggled at handling the barbarians (BK, Ram). Furthermore, the Big Three lacked any crippling flaws that the beta triumvirate had. Caesar’s avoidance of Culture as if it would bring the Ides of March combined with his tundra-infested land caused his Praetorians to arrive too late to be of any use; the Burger King’s awful starting techs, jungle-infested land, and religion focused strategy meshed poorly with his seafood start; Ramesses’ military incompetence was astoundingly apparent in these games, and his land, while rich and fertile, was rather hammer poor.

However, there was a boom-or-bust nature among the Big Three, with each member experiencing the entire gamut of possible results. Their victories could be easy romps or brutally hardfought wins, their 2nd place finishes could be near-wins or lucky backdoor games, and their eliminations could be embarrassing early exits or heartbreaking last-minute collapses. The Jekyll-and-Hyde nature of the Big Three was also apparent in the general map dynamics, as it felt like every game was either a laugher (Games 10, 14, 18) or a bloody slogfest (Games 13, 16, 20). Oftentimes, whichever leaders made it to the finish line depended on who was able to get on the right side of dogpiles. One thing remained constant: the relatively low number of wars. Despite this aggressive bunch of leaders, the war counter was quite low due to most wars being fought to the death, whether it was due to a war being a vicious dogpile, a protracted stalemate, or a much stronger leader Thanos snapping some poor sap out of the realm of existence. There was also frequent last minute vulturing of dying civs that artificially ticked up the war counter in many games.

An interesting tidbit was that although the Big Three clearly dominated the Winner and Runner Up categories, First To Die turned out to be much more unpredictable. Although the community favorites (Louis, Ramesses) were reasonable picks (especially Ramesses), every leader ended up First To Die at least once, and all save for Kublai Khan got that distinction multiple times. As it turned out, dogpiles were hard to predict, and a few games saw a leader collapse from a dominating position to be First To Die after getting ganged up on. Whichever victory condition took place was also a tossup, dependent on if the winning leader chose to leverage his tech advantage into world domination or into a more peaceful conclusion. Unsurprisingly, Louis leaned towards Cultural wins, Hannibal opted for Space, and Kublai for Domination. Unlike Playoff Two, where the UN was one of the defining facets of the map, these games never came close to a Diplomatic finish. The religions were dominated by Ramesses, Charlie, and (to a lesser extent) Louis, but religion was of little relevance in these games save for aiding pushes towards Cultural victories - border tensions were the primary drivers of conflict. Ramesses was by far the most successful at spreading his religion, but this did little to help him.

How Typical Was the Actual Game?

0/10. I had a hunch that the livestream game was an unlikely result. Both of the advancing leaders got insane goodie hut luck, with Julius Caesar popping Mysticism before morphing into Constantine, and Kublai Khan popping an early Writing to enhance his already formidable land-grabbing abilities. With Louis sandwiched between the two lucky bastards of the map, this was bad news for him and ultimately the rest of the world. However, looking at these results, I am not sure if Caesar has many prospects in this setup even with Deity starting techs.

Analyzing each leader’s individual performance:

Louis XIV of France
Offensive Wars: 33
Defensive Wars: 43
Survival Rate: 70%
Finishes: 10 Wins, 2 Runner Ups (54 Points)
Kills: 11
Overall Score: 65

Louis was the deserving winner of this set. He was the only member of the Big Three still capable of winning if things went sideways, if only because of his culture ticking-time bomb. Secondly, he was the most well-rounded of the Big Three, excelling at the landgrab stage, maintaining military supremacy, and keeping up in tech; Kublai and Hannibal were less consistent in these aspects. Louis did have a lot going for him here: as I had mentioned, he had the best land and the weakest neighbor. In many games, Louis’ victory was all but assured by Turn 100, as he would expand to a double-digit city count, pump out tons of culture to squeeze his competition, make the Roman lands fine additions to his own, and become the primary executor of dogpiles. It spoke volumes that his most dominant showings were not his Cultural finishes, but actually his three pre-T300 Domination wins. He had many pathways to victory: run over Caesar and snowball, run over Hannibal early, sit back and allow everyone else to fight while he cultured or teched away, or build up and run over one of his neighbors at a later stage of the game.

The biggest roadblock for Louis was his Southern neighbor, Kublai Khan. Every single game in which Louis faltered had one thing in common: the French found themselves mired in conflict with the Mongolians, whose single-minded dedication towards everything military made them formidable foes. Louis’ inability to deal with the Mongolian threat led to multiple poor showings, including three First To Die performances. The Sun King did run over Kublai multiple times, but this almost always came when he was technologically advanced enough to do so. Louis was lucky that despite his central position, he only had to deal with one dangerous neighbor – as the livestream showed, a stronger Caesar would not have been good news for the French. This set demonstrated what many already know about Louis: he is a quintessential boom-or-bust leader who can be his own worst enemy at times. At least he redeemed himself after his disappointing Opening Round Alternate Histories. For those who have not watched later seasons yet, I would keep Louis in the back of your mind – he is as capable as anybody at AI Survivor, warts and all.

Best Performance: His shockingly strong Game 10 victory easily takes the cake here.

Worst Performance: Going way, WAY too crazy with expansion in Game 18, provoking a 3v1 that led to his early demise.

Wang Kon Burger King Minority Shareholder Award:

All of Gaul is Divided Into Three Parts Award:

Kublai Khan of Mongolia
Offensive Wars: 53
Defensive Wars: 23
Survival Rate: 80%*
Finishes: 3 Wins, 7 Runner Ups (29 Points)
Kills: 27
Overall Score: 56

*This stat is EXTREMELY deceiving. When accounting for the games where Kublai Khan was on his way out when the game ended, his “true” survival rate plummets to 55%.

Kublai may have scored the 2nd most points in this set, but he was clearly the 3rd cog in the Big Three, more Chris Bosh than Lebron or D-Wade. With that said, it became pretty obvious why Kublai is one of the most formidable warmongers in Civ IV. His traits and the Mongolian civ nullified the two biggest weaknesses that most warmongers have (early culture and early economy), and he was excellent at expansion and warfare. Kublai was a kingmaker here: games swung depending on who Kublai chose to attack, and he was always a strong contender to make it to the Championship game due to the tasty snack to his east that was Ramesses. With all that said, there was a reason why nearly half of Kublai’s points came from kills. Being completely devoted to military matters had tradeoffs: Kublai was a subpar economic leader. There were a multitude of games where Kublai had conquered an entire hemisphere, only to falter because he had fallen too far behind in tech or because it was too late to stop Louis from getting three legendary cities. Kublai was also not the smartest with his war declarations at times. Too often did I see him launch cross map wars for little gain, wasting time, gaining little of note for his efforts. In particular, there were a couple of games where he had a chance to stop Louis from winning, only to sic his entire army against Charlie on the opposite corner of the map and then have to give away his conquests as they got culturally crushed.

Interestingly, killing Ramesses was the “safe” choice, one that made it likely that he would advance but extremely unlikely to outright win. Two of Kublai’s three wins saw him kill Louis early instead, as Louis’ land was just much better for snowballing while also giving him much more easily accessible secondary targets to keep up the momentum. The one game where he did not kill Louis yet still won (Game 20) was a major slog where everybody got stuck in wars, and Kublai was able to crawl to space by getting the most territory for himself (he somehow did not conquer anybody in that game). Kublai is deserving of his reputation as one of AI Survivor’s top leaders, but as these games showed, he lacks the economic heft to truly measure up to the cream of the crop.

Best Performance: His incredible Turn 267 victory in Game 18 takes the cake here, and this date undersold how amazing he played. Had Charlie not sniped a couple of cities from Kublai, he could have won before Turn 250.

Worst Performance: Somehow getting run over by Ramesses in Game 2.

Mongolia Delenda Est Award:

Hannibal of Carthage
Offensive Wars: 23
Defensive Wars: 38
Survival Rate: 65%
Finishes: 5 Wins, 7 Runner Ups (39 Points)
Kills: 11
Overall Score: 50

In many ways, Hannibal was the foil to Kublai Khan. While the Mongolians supplied the brawn to the Big Three mafia, Hannibal was the brains of the operation. Kublai was a far more consistent leader, while Hannibal had much greater variance in performance, either absolutely dominating or being a pathetic 5-6 city weakling without much in between (do not be fooled by his seven second place finishes – most of them were backdoor “everyone else died” finishes). Hannibal had three hurdles he had to overcome to succeed. First was his lack of cultural inclinations, which left his border cities struggling to control their first ring tiles considering who his neighbors were. Second was his jungle coated land, which, in the short term, left much to be desired. Especially painful was the fact that his capital was one tile away from the coast, preventing Hannibal from being able to play the water economy game that he has proven adept at in the past. Hannibal was also rather unlucky at times, as in quite a few games inopportune barbarian city spawns left him boxed in with six cities and completely irrelevant. Finally, due to having to deal with cultural border tensions from three sides, Hannibal was vulnerable to dogpiles that led to three games where he became First To Die.

In return, three factors kept the Carthaginians afloat. First was his Financial boosted double Ivory capital which kept his economy in pristine shape while giving him enough hammers for the landgrab (people underestimate Ivory capitals). Indeed, Hannibal’s best games saw him go absolutely crazy with settling, even claiming Northern territory between the Salad Man and the Burger King. Secondly, Hannibal built and utilized the Pyramids extremely well in most games, either adopting an early Representation to boost his economy or an early Police State to fuel an early conquest. Finally, Hannibal had two weak neighbors to pick on, and when a Financial leader doubles his empire size, the game is all but sealed for him. Hannibal’s performance in this set was much like many of his other AI Survivor games: it seems that he either has crazy games that garner him the “Chadnibal” moniker, or he has some stinkers that make everyone wonder if he is the most overrated leader in the game. In truth, all this balances out to make Hannibal an above average leader for AI Survivor purposes.

Best Performance: His crazy Game 14 teching performance, sparked by a perfectly executed conquest of Ramesses.

Worst Performance: All three of Hannibal’s First To Die eliminations followed the same pattern: he barely expanded, launched a failed war against one of his neighbors, and was quickly delended into the bottom of the Mediterranean sea.

Fingernails Scratching A Chalkboard Award:

Charlemagne of the Holy Roman Empire
Offensive Wars: 32
Defensive Wars: 28
Survival Rate: 35%
Finishes: 0 Wins, 4 Runner Ups (8 Points)
Kills: 9
Overall Score: 17

After the Big Three, there is a steep drop-off to the next tier of leaders, with the Burger King proving to be the least bad of the remaining leaders. Charlie is generally regarded as the worst leader in the game in human hands due to his bottom tier trait and starting tech pairing, and with Deity starting techs removed, this set certainly showed why. Charlie's early game was always too slow to ever contend for a victory, and his two hopes for a Championship spot was to 1) be lucky enough to sneak into the championship off of Caesar's or Hannibal's territory, or 2) for Kublai Khan on the opposite corner to take out everyone else.

Charlie is a decent leader at worst, but here, he was not able to execute his religious strategy due to his slow start and lack of rivers to spread his religion. Although he and Ramesses almost always founded two of the first three religions, Ramesses was far more effective at spreading his religion, as the Egyptians would often build Stonehenge for the Obelisks that give them extremely early Prophets, allowing for extremely early shrines and thus a virtual monopoly on the religious race. Too often, Charlie would find himself a diplomatic pariah, being the only practitioner of his religion alongside his relatively high peaceweight, and there were games where it seemed he would have been able to muster something when Kublai came crashing in from the other side of the world, leading to the Holy Roman collapse. Even in Charlie’s good games, he was far too behind in tech to amount to anything. Altogether, Charlie just did not have the tools to make much of his tenuous at best situation, and if “Second To Die” was an AI Survivor category, Charlie would have been a great pick.

Best Performance: The only game where Charlie had a shot to win was Game 13, where if Louis was not a culture-monger, the Burger King would have probably limped to a late Spaceship win.

Worst Performance: Allowing Hannibal to catch him without metals in Game 7, losing two cities on the first turn of the war.

Opera-Loving Aztec Empire Award:

Vandalized Headquarters Award:

Charlie advanced to the Championship round in this game.

Ramesses II of Egypt
Offensive Wars: 8
Defensive Wars: 36
Survival Rate: 25%
Finishes: 2 Wins, 0 Runner Ups (10 Points)
Kills: 1
Overall Score: 11

All one needs to do to diagnose what went wrong for the Pharoah here is to look at his war counters. Not only was Ramesses such a soft leader as his eight offensive wars suggested, but also his defensive war count was shockingly low considering his survival rate. That suggests one thing: when Ramesses was attacked, he usually melted quickly.

In most games, Ramesses’ death warrant was already signed by Turn 100 as his inability to outcompete his western neighbor Kublai Khan in the expansion phase would leave him without enough cities to defend himself from the eventual Mongolian invasion. His expansion was further hampered by constant barbarian harassment, worsened by his northern neighbor Hannibal often building The Great Wall to steer the barbarians towards Egypt. These issues were exacerbated by the Egyptian leader spending too much time on lesser wonders like Stonehenge and Temple of Artemis (a good reminder to human players that both of these wonders are classic noob traps), which were great for spreading his religion, but not for straightening out his other priorities. Ram’s issues with his Mongolian neighbor were understandable, but considering his War Chariots, his inability to handle the barbarians does not reflect well on him. Speaking of which, one issue I noticed from Ram was in general, he was far too attached to his unique unit at the wrong times. Despite being strength five, War Chariots make for terrible defenders, especially against Keshiks and Elephants.

War Chariots and Aqueducts will surely save me from War Elephants, amirite?

All Ram could muster was two somewhat flukey victories, which, as I had mentioned, stemmed from Louis and Kublai fighting each other into irrelevance in a pre-T100 war. After Game 4, it was not sunshine and rainbows for the Egyptian pharaoh. Of the games Ram did not win, he came close in Game 20 (and would have won had he turned up the culture slider), before finding out that going to Space on six cities generally is not a viable strategy, and there were a couple other games where he was having a decent game before he foolishly turned up the Culture slider too early, leading to his downfall. Ram is a great culture-monger, but for someone with a more neutral peaceweight, his diplomatic isolation felt enigmatic at times (especially when Charlie was able to get along with his rivals a bit better). I do wonder if he high rolled on his peaceweight in this set.

Best Performance: Killing Kublai early in Game 2, and then out-teching HANNIBAL to space.

Worst Performance: Tough to say. He got steamrolled a lot, perhaps his earliest eliminations in Games 5 and 6?

Real Poverty Point Award:

Tarquinius Superbus of Rome
Offensive Wars: 41
Defensive Wars: 22
Survival Rate: 15%
Finishes: 0 Wins, 0 Runner Ups (0 Points)
Kills: 3
Overall Score: 3
For all of Ramesses’ faults, at least he never reached the ineptitude of this guy. Clearly, Caligula’s horse was leading the Romans in this set.

To say this was an embarrassment would be an understatement. I saw Nero crash his economy, never use his Praetorians effectively, stop expanding after six cities (perhaps to plot war), ignore culture for way too long, and effectively serve as the game troll, launching fruitless wars at the worst possible time until he was swatted away like a barbarian. My hypothesis for Commodus’ perplexing performance: he is one of the leaders most hurt by the removal of Deity starting techs. The consensus among the best Civ IV players in the world is that Praetorians are an overrated unit, great for mowing down leaders on Monarch difficulty or for Hall of Fame speedruns, but a very awkward unit in normal high difficulty games. This is because even as an eight-strength unit (same as War Elephants for reference), military progress is virtually impossible without Catapults, which happen to be on the opposite side of the tech tree. Without Deity starting techs, Elagabalus was often beelining his unique unit without much in the way of useful development techs, smashing Praetorians against a Protective or Creative leader without Catapults, and subsequently becoming a non-factor. Moreover, with Deity starting techs, Honorius had six starting techs, while without, he was stuck with the Mining/Fishing combo, which in the hands of the AI is good in some situations but useless in others (because the AI are not good at maximizing the benefits of immediate access to Bronze Working).

That really is all I can say. Maximinus Thrax was a complete afterthought here, one whose only points came from random kill snipes. The only mitigating factor was his terrible commerce poor land. What a horrific performance from the Actual Game winner that really taints his dominant early season legacy.

Poisoned Bread and Circuses Award:

Conclusions

This was an interesting and unexpected set of games, and of all the Season Three sets I have run, this one was, in my opinion, the one most affected by the removal of Deity starting techs. I think Caesar might have fared better with Deity starting techs, while Louis would have been much weaker due to the faster starts of all the warmongers who surrounded him. There was definitely a luck factor in this set, as much depended on who was able to benefit the most from dogpiles. I would not be surprised if another set has quite a different result.

 

[Eauxps I. Fourgott]

Good grief Caesar Although as I've mentioned before, given the different conditions of these AHs, I think you assign too much weight to them when looking back on the season itself. JC wasn't playing without Deity starting techs during Season 3, so this doesn't taint his successful performance there nor prove it a fluke.

 

[Thrasybulos]

So, the misread here is that Louis was actually bordering one of the marked men!

 

[Genghis the Barbarian]

Merry Christmas everyone!

It's been a while - I have been extremely busy over the past 1.5 weeks or so, but I was finally able to get all the championship games done:

To recall: Stalin won in a stunning upset in a really tight game, while Justinian came oh so close to being AI Survivor's only leader with multiple rings as of now (post S8).

Now, a quick disclaimer: As may be obvious here, I arbitrarily decided to alter the scoring system so that the winner gets 10 points instead of the usual 5.

Why? Well, this is the Championship game after all, where winning means everything, and I felt that it was most important here that the final results actually reflect which leaders actually had the best chances of winning a title. Had I kept the usual scoring system, second and third place would have been switched, and it did not feel right for me to rank a leader with only two wins above a leader with six wins (and, frankly, quite a few near wins).

There were pretty much three tiers of two leaders: the winners, the spoilers, and the suckas. Who was where?

 

[Genghis the Barbarian]

Based on feedback, the inflated scoring did not sit too well with others, so I decided to also include a teaser with standard scoring:

 

[Eauxps I. Fourgott]

I'll also cast a vote against the inflated win scoring. The final scores in the normal version do reflect the flawed nature of the current scoring system with regard to the importance of kill credits. But massively inflating the value of wins isn't the way to fix that and this makes the results fall in line with historical charts (and also actually reflects more clearly that first place was not a ROFLstomp).

Having one leader with a very poor survival rate was to be expected, but a second is more surprising. Were the leaders' peaceweights locked to the same value across all twenty games for these replays?

 

[Genghis the Barbarian]

Having one leader with a very poor survival rate was to be expected, but a second is more surprising. Were the leaders' peaceweights locked to the same value across all twenty games for these replays?

The second to last place leader honestly had quite a few flaws and diplomatic issues that had little to do with peaceweight, that I will discuss in the writeup

 

[Eauxps I. Fourgott]

Uh, thanks for the hint?

But my question still stands. Was peace weight locked or variable in this set?

 

[Genghis the Barbarian]

Sorry, it was locked, I meant to state that in my response.

 

[Keler]

1.Justinian
2.Kublai Khan
3.Pacal
4.Stalin
5.Julius Caesar
6.Mansa Musa

so Stalin only won because he is next to a dead man?

 

[Eauxps I. Fourgott]

All right, let's do a proper analysis here! Been a while since I only had to do one of these at a time

Spoiler Analysis :

Let's start with the easy ones. After all the guesses for Thrasy's series, one pattern that's been thoroughly drilled into my head is that a relatively high win count with low kill count is almost always Pacal. Clearly there are a lot of games that go like the real one, where he gets killed before he can pull all the way ahead, but whenever the others leave him alone for long enough because he's a fellow villain, it's GG.

Meanwhile it's always GG for poor Mansa, and an only 35% First to Die rate is honestly pretty astonishingly low! It's obvious that he's one of the bottom two leaders (the real surprise is that somebody else had it close to as bad as him), so the real question is, was he able to actually win a game? This is Mansa we're dealing with, after all. However, fifth place has seven kills, and if Mansa was in position to get as many as seven kills, I think he'd have more than a single win as a result. That was my line of thinking at first... and then Atesh/Genghis/I-don't-know-what-you-actually-want-to-be-known-as dropped his little teaser that all but confirmed it. Mansa 0% survival confirmed.

So that leaves Stalin, Justin, Julius, and Kublai - three warmongers and a religious warmonger. One of them is a loser who almost never accomplishes anything significant and only managed one, probably very flukey win. The other three run the table and regularly run over the rest of the field, with one of them much better at actually winning than the others. But who fits into which role?

Ironically, even though Kublai was the loser in the real game, I think he's least likely to occupy that role here - he's neither toward the edge of the PW scale nor lacking in early culture. Justin and Julius both are at the top of the PW scale once Mansa's gone, though, and especially if one of them rolled high, that could result in a lot of diplomatic trouble down the road. Meanwhile Stalin and Julius are both weak leaders in the early-game who will especially be without culture for a long time early on, and next to AIs who won't be.

In the end, I settle on the combination of these two factors as the main reasons for fifth-place's poor performance: I think it's JC. We've established pretty well by this point that he's a pretty poor AI Survivor leader with the Deity starting techs removed; here he does start with two Fishing resources, but I don't think that's enough to swing the tide. Especially when he'll also be culturally crushed by both Justin and Kublai, and even more especially if he rolled high on PW, I don't think he can fight his way out of the corner. He repeatedly dies and is overall pretty embarrassing here. It turns out the real game was probably his last true chance at total glory. (As a side note, I think this game is especially heavily impacted by the starting techs change, and combined with ripple effects from earlier in the season is pretty useless at determining if Stalin was a likely champion or not.)

With Julius out of the way, then, we have three performances that, on paper are all quite similar, with near-equal survival and kill rates. The only real difference is that one leader is distinctly better at winning than the other two - and although it's not a creative choice, I have to guess that it's Justinian. He's just better at the competition, in general and especially economically, than either of those other turkeys, plus he borders JC who is usually dying at some point, probably fueling Justin's success in the process. Of course it's still not a crushing victory - this is a dangerous map and there's also Pacal on the other side to contend with - but he clearly performs the bets and proves his chops yet again.

That leaves Stalin and Kublai with, based on the results, no real difference between their performances. Maybe they were actually quite distinct in the replays themselves, but from the numbers alone they appear to have done more or less the same. To put an order to them, though, I'd bet on the one who has early culture to do a bit better overall.

All of which is a very long-winded way to say that I agree with Keler. (except for the comment about why Stalin won, because again this set is useless for analyzing that result due to its major nerf to both Stalin and Julius)

 

[Thrasybulos]

Keler and Eauxps's guess is the one which makes the most sense.

Now, those games tend to throw curveballs... so is there another possibility?

Last place has to be Mansa. I could see him winning a coupla games when everything breaks his way, but I can't see him not having the worst survival rate.
Stalin and Pacal border him. Stalin is bad but is more likely to take advantage.
Justinian and Pacal are the most likely to found the early religions, Mansa might make things even worse for himself by founding the 3rd.
Stalin and Caesar are the worst leaders in that field, Justinian is the only one who can't plot at Pleased.

Pacal tends to be weak militarily (at least until he gets the tech edge), so let's assume for contradiction's sake that he's the 20% survival rate.
7 kills and 6 wins could be Justinian then. Kublai, bordering Pacal, should do better than Caesar.

1. Stalin
2. Kublai
3. Justinian
4. JC
5. Pacal
6. Mansa

I still think the other guess much more likely, though...