Game Six Alternate Histories Online 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 Six saw Darius explode out from one of the best starts in AI Survivor History, easily making it to Space before Turn 300. 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 we have since used the same general format. This particular set of alternate histories were run by Myth - many thanks for spending so much time on this task! Myth posted the resulting data from the alternate histories and then discusses some of the findings below 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 originally noted, 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.)
Myth: Welcome to the "Game of Incompetence."
Every single leader in this game had some of the dumbest moments I have ever seen while running alternate histories, and they happened in almost every game. Darius threw away a contender for the best start of AI Survivor History in multiple games by not expanding or declaring EXTREMELY early wars (see spreadsheet for examples) on Brennus. Julius Caesar, the big pre-game favorite, was also one of the worst expanders on this map - and the dude literally has Imperialistic as a trait! Mehmed did absolutely nothing in many of these games, just sitting there and teching poorly (not pleased-locked, just utterly useless). Tokugawa had one of the worst starts in AI Survivor History, consistently 500+ points behind anyone else without even getting crippled by war. However, he also didn't help himself, declaring suicidally stupid wars against much, much more powerful neighbors. Brennus and Saladin declared too many wars they couldn't win as well, but Saladin at least had the excuse of getting dogpiled in the majority of these games. Finally, Willem's incompetence was the most familiar, not researching Rifling in typical Willem fashion and throwing away multiple games because of it.
All of that being said, we have a dichotomy in terms of our top two leaders. Darius's performance was the most condemning of all 7 on this map. Yes, he was the top AI on this map (though barely by the most extreme of margins). But look at that capital!
There is literally zero reason this dude shouldn't have won at least 70% of these games. Persepolis had double food and double gold, with room for at least 10-12 peacefully settled cities that mostly consisted of very green land (after Iron Working). But there were plenty of games he should have won but just didn't because he was unable to kill an opponent with tanks when said opponent had rifles or because he threw away promising starts by not settling land. The 30% I gave him as leeway were admittedly due to things outside of his control, including 2 or 3 games where Mehmed walked across the entire map and turned Darius's 1v1s with his neighbors into 2v1s, dooming the Persian leader. I saw some chatter about thinking Darius is actually a decent AI after his performance this season, and watching these games completely disproved that notion in my eyes.
He also gets docked points in this specific game due to his civilization. Immortals are pretty good units for the human, allowing early rushes and absolute impunity from barbarian archers and axes. However, there is one MAJOR drawback of Immortals for the AI: they allow Darius (and Cyrus) to plot war with just them. The way it is coded is that for an AI to plot war, they need 2 units that have specific unit flags. Chariots usually only count as 1 (mounted), so a leader can't declare with just chariots. However, Immortals are one of two early unique units that break this rule, counting as TWO units in this regard (mounted & anti-archer). This resulted in Darius being able to plot war as soon as he researched Animal Husbandry & The Wheel due to the horses at his capital, so there were multiple times that he torpedoed his own game by stopping his expansion in order to declare on Brennus before Turn 60 (in Game 7 he declared on Turn 40!) and allowing Willem to easily settle land or take barb cities that he should never have had access to. It was like a hyper-copper-dec, ruining his own game by focusing too much on unit production instead of actual expansion.
Speaking of Willem, he was probably the most competent of these leaders, even if he lost first by 2 points. His start wasn't all that special (definitely above average, especially his second city, though way more constricted than Darius's), but he consistently outperformed it, often ballooning to one of the top 2 scores by Turn 150. It wasn't just peacefully, either, declaring almost double as many wars as were declared upon him and winning most of them. The Willem who died first in the Actual Game turned out to be hugely misrepresentative of how the rest of these games went: 0 further first to dies, basically tied for top dog with Darius. Like aforementioned, he definitely had some issues, mainly dealing with his refusal to research Rifling tech, and he threw away some victories by attacking leaders with power ratings that dwarfed his, but overall he was the most impressive to me. If you were to run 100 games on this map, both of these leaders would have a shot of barely edging out the other, as they both had a lot of consistency (for vastly different reasons), but they would definitely still be extremely close.
Then we have our three "Tier 2" leaders. Brennus was the most feast or famine: although he was the only leader to win outside of the top 2, he also wound up dying more than any other Tier 2 leader. He was extremely strong in the first two games, leading me to think I had completely missed him as a strong AI on this map, but he then kinda cratered after that (outside of the VERY weird Game 12). Mehmed was the most consistent of these three, boasting a 95% survival rate and the highest number of second places (though JC was right up there). He only once was actually a contender to win the game, however. Julius Caesar was somewhere in the middle: he got dogpiled a lot more than I expected, and came very close to winning a couple of games (he had the land percentage for domination in Game 13 but not the population because he gifted a city to Darius..). His 5 second places were all quite strong, no distant backdoor podiums for Caesar. Ultimately, however, these three had one main issue in common: the inability to keep up in technology with Willem and Darius. No matter which of these two won, the others were so far behind they would be committing suicide to attack (which they did, many times).
And finally, our two rocks, at most providing a stubbed toe to the rest of these leaders. Saladin and Tokugawa were handed rough, rough draws on this map. Saladin had decent land, but he was constantly dogpiled by everyone around him. He and Brennus founded the first two religions almost always, and they automatically were enemies. Willem declared on him most of the time, Caesar often piled in, and Mehmed sometimes decided to take a swing as well. This led to the inevitable 60% first to die rate, something I was not surprised to see. Tokugawa, on the other hand, was handed one of the absolute worst starting positions in AI Survivor, only beaten out to the title by the true duds such as Hannibal's in Season Six, Gilgamesh's and his own in Season 1 (before the starts were more normalized), and Sitting Bull's this season. His capital was fairly good, but the rest of the land was not, and he was hemmed in by Mehmed to his north and Caesar to his east.
Overall, this game had some of the most boring games I have ever watched during alternate histories, but also some of the most entertaining. If you have the time, check out the individual game notes on the spreadsheet linked at the top. Games 1, 2, 12, and 17 especially are games where I was riveted, and I really wish one of those four wound up being the Actual Game.
Now for a look at the individual leaders:
Darius of Persia
Wars Declared: 25
Wars Declared Upon: 33
Survival Percentage: 70%
Finishes: 9 Firsts, 3 Seconds (51 points)
Overall Score: 66 points
The King of Incompetence himself, it only makes sense that he winds up being the top AI for the Game of Incompetence. Darius has consistently been one of the most disappointing leaders in AI Survivor, losing way too many games that he should have been better in. After an insanely strong Season Seven, it looked like the community was starting to feel a bit better about him. This game proves that no, we should not suddenly jump on the Darius bandwagon. Like aforementioned, the fact that he wasn't a 100-pointer on this map is absolutely damning for Darius's abilities as an AI. It's not like he had wins stolen from him, either, as he wasn't nearly the top AI in any of the games he didn't come in first. He did have a couple close calls, being about 15-20 turns from space (including transit time) in a couple of Willem's victories (which IS impressive; it's hard to out-tech a cultural victory, but his start carried him to that point). However, Willem was still vastly stronger in those losses.
Willem of the Dutch
Wars Declared: 57
Wars Declared Upon: 37
Survival Percentage: 70%
Finishes: 9 Firsts, 2 Seconds (49 points)
Overall Score: 64 points
While Darius was the undisputed tech leader, Willem was the undisputed culture leader. Willem's 9 victories were almost all cultural (except for 2 domination wins and a diplomatic win, all on track to be culture victories themselves). Willem was definitely far more impressive than Darius, only barely losing the top AI slot to a leader with a vastly stronger start. For most of the last game, it looked like Willem would come second (until Darius declared on him), and Willem would have come in first overall (as Darius would've lost a kill and Willem would have gained a second). These two were pretty similar in terms of how they made it to the podium, and both having almost half of the victories is something I would predict to continue with any more iterations. Both Willem and Darius had some streaks, with the former winning 7/8 times from Games 4 to 11 and Darius winning 6/8 times from 13 to 20, but these leaders were still almost identical in performance. I personally think Willem might parse out very, very slightly ahead of Darius after watching these games: think maybe 5 victories ahead after 100 games. But that difference is still so miniscule as to not tell a lot.
Brennus of the Celts
Wars Declared: 33
Wars Declared Upon: 52
Survival Percentage: 35%
Finishes: 2 Firsts, 3 Seconds (16 points)
Overall Score: 26 points
Brennus wins the not-so-prestigious award of "being able to win a game or two, I guess." Outside of his two firsts (and Game 2), he was pretty bad, but the first of the two was quite impressive. Obviously a major outlier, just like Game 12, but at least he was able to cross the finish line! It was quite funny seeing him win by culture in Game 1 due to capturing holy cities from Willem and Saladin and then popping into the cultural gameplan. But he really wasn't impressive other than these games (if you can call a turn 400+ victory impressive…), and often found himself on the wrong side of Willem and Darius. His path to victory consisted of dogpiling Saladin and getting the most spoils, which he was fairly good at doing. He just could not keep up in military tech, leading to mid- to late-game collapses. I would probably rate our next two leaders above him in terms of actual strength and consistency.
Mehmed of the Ottomans
Wars Declared: 42
Wars Declared Upon: 11
Survival Percentage: 95%
Finishes: 0 Firsts, 6 Seconds (12 points)
Overall Score: 24 points
95% survival is definitely pretty consistent! Outside of the incredible survival rate, our Ottoman sultan was disappointing and boring, often just sitting in his corner and having an abysmal economy until it was far too late to do anything. Mehmed is known for aggression, and he had a game (G7) where he took part in ZERO wars. His only elimination was pretty funny - Caesar walked up with Praets before he had metals, taking his entire territory without losing much in return. Very fluky, but funny nonetheless. But it says a lot that getting destroyed in this fashion was the most entertaining thing he did in all 20 matches. It almost felt like he wanted to lose these games, and his 6 second places were almost entirely very distant, getting lucky enough to be the furthest away from Darius and Willem.
Julius Caesar of Rome
Wars Declared: 36
Wars Declared Upon: 33
Survival Percentage: 55%
Finishes: 0 Firsts, 5 Seconds (10 points)
Overall Score: 21 points
Unlike Mehmed, the majority of Julius Caesar's second places were anything but distant seconds. He had trouble getting there, but when he took out a neighbor and got rolling, he was extremely scary. He was far and away the second-strongest leader in 4 of those 5 games, and probably should have won Game 13 if not for fluky AI programming as mentioned before. His start was pretty bad, and his biggest flaw was a strange inability to expand well. This was compounded by a lack of culture in most games, and this caused him to be weaker than he should have been. He also got dogpiled more often than you would expect, many times declaring on Mehmed, Saladin, or Willem, then getting attacked from behind by Toku (as in the Actual Game). His strongest games all stemmed from him attacking Toku directly, leading to a 1v1 (or 2v1 when Mehmed also declared on Toku) and a strong Rome. He was definitely the most impressive of these three, in my opinion.
Saladin of Arabia
Wars Declared: 12
Wars Declared Upon: 43
Survival Percentage: 25%
Finishes: 0 Firsts, 1 Second (2 points)
Overall Score: 4 points
The wars declared upon stat really tells it all for poor Saladin: he was a global pariah, hated by every single one of his neighbors in game after game. I think a lot of people expected this: he was peace weight 4 surrounded by 0-2 peace weight leaders (or JC and Willem, both of whom can plot war at pleased), and was likely to be the main target of Brennus due to the religious divide. There wasn't a lot he could do in most of these games: he was dogpiled to oblivion. There was also the incredible game in which he was safe from everyone on the map, then signed a defensive pact with Toku, dying by the stray bullet of Darius declaring on Japan. Poor Saladin couldn't catch a single break, and I felt bad for the guy.
Tokugawa of Japan
Wars Declared: 29
Wars Declared Upon: 20
Survival Percentage: 15%
Finishes: 0 Firsts, 0 Seconds (0 points)
Overall Score: 2 points
Tokugawa was handed a very bad start, and he is not at all good enough to make it out of it. His tech rate was abysmal, and in the few games he survived he never once made it to muskets, let alone rifles. I don't have a lot to say here, Toku isn't a great AI (he's mid-tier at best), and he had by far the worst spot on the map. Making it all the way to the Playoffs was an incredible stroke of luck for our poor isolationist.
The Actual Game turns out to be typical, atypical, and everywhere in between. Darius winning wasn't atypical, and his win often looked very similar to those in these alternates. Mehmed coming second also wasn't atypical, happening more often than everyone else (and in similar fashion - sitting there doing nothing for 200 turns and getting tangled up with Toku). However, Willem dying was unlikely, especially first, and he should have gone much further this season (even if it was just to the Wildcard!). Saladin and Toku also should not have gone to the Wildcard, and then they both made it out of there to the playoffs! Who knows what the Championship would have looked like if it was Willem/JC instead of Darius/Mehmed/Saladin (obviously assuming a lot about the Playoffs here), which was not at all the most unlikely of outcomes. It being basically a 50/50 as to who won the game is very unusual, but like I said before, these two were so consistent that I do think this is what would hold up under further testing - Willem might be 5-10% higher than Darius (or vice versa), but they're probably gonna be extremely close regardless. The community as a whole whiffed hard on this game: JC had almost 50% of votes to win and Willem was the favorite for First to Die. This map was also HUGE; domination was extremely difficult to achieve, as it required getting the majority of the spoils from at least 4 rivals, so 50% of the votes being for domination was a whiff as well.
Now for fantasy. Darius carried a lot of Amica's fantasy team, scoring a total of 16 points, only beaten by the Champion of Season Seven, Louis (also on Amica's team). Like I said, however, this was basically a coin flip. If Willem wins instead, he gets Darius's playoff spot AND a very, very good championship draw. Willem could have scored at least 15 points at the end of it, which would have had Slashin's team much, much better off. Of course, if we sat here and talked about what should have been, Amica would have had Suleiman AND Elizabeth advance for extra points, plus a ~50% chance of Darius still winning, and Slashin' would have lost a lot of points from Hannibal and Hammurabi. So, I think that puts Amica in first, anyway. Regardless, Slashin' had a genuinely good bid (better than Amica's 57 for Darius!!!), seeing something the rest of us didn't, and it's sad that it didn't work out for him; slightly more luck here would have seen him challenge for that first place overall. Amica also scores for Mehmed most often, even if he just went to the Wildcard. Sunhuge loses very, very hard: Saladin's overall silver medal run seems to have been destined to be stopped here, and thankfully for him, Saladin managed to make it to the Wildcard. Sure, he should have gotten 2 points here, but he wound up getting 11 from Saladin - definitely a downgrade. Finally, Jmie's bid on Caesar did wind up being a miss - 0 points on JC seems to be fairly typical, though 25% odds to advance isn't the worst (not really worth the 65 gold, though).
Thanks for reading - I hope you enjoyed this surprising look into Game Six!