Playoff Game Two 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. Playoff Game Two was a total runaway by Pacal that included a frighteningly speedy conquest of Bismarck as Mayan tanks rolled across the landscape. 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. 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.)
And I thought that Playoff Game One from this season had a predictable series of outcomes, sheesh! 
This game had the least variability out of any alternate histories that I've ever run with Pacal winning 19 out of 20 games and Cyrus coming in second 17 out of 20 games. Both of them survived 100% of the time and were never elimiated though Cyrus came close to dying a couple of times. We watched Pacal steamroll through the rest of the field in the Livestream game thanks to his ridiculous capital with triple seafood resources, a gold resource, four floodplains tiles, and a bunch of hills to mine and forests to chop. He was equally as dominant in the alternate histories and inevitably pulled out to an economic lead thanks to that starting position and his Financial trait. Pacal didn't expand particularly well in many of these games and it essentially never mattered because he would be a full era ahead of everyone else by Turn 200. Even getting pulled into a bunch of wars against Zara or less frequently Bismarck didn't seem to slow down Pacal much. He would typically stay on the defensive until he zoomed out in front of his opponents in technology and then ruthlessly annihilated the opposition. Pacal preferred to head for space rather than engage in combat which is why he "only" managed one kill per game. He easily could have achieved Domination in virtually every game had he been more inclined to attack.
The only other AI who was competitive on this map was Cyrus who drew another beautiful starting location in the southwest. The Persian capital had double seafood resources and a wet corn, then the second city location chosen by Cyrus in every game had a dry corn, a gold resource, and five floodplains tiles. Under virtually any other setup, Cyrus would have been the runaway winner of this game and indeed he typically came out of the landgrab phase with the most territory, largest population, and highest score. The problem for Cyrus was that he couldn't match a Financial leader who had an even better starting position and who was located at the furthest distance away on the map. The alternate histories revealed one important fact that determined the outcome of most games: Pacal and Cyrus almost *NEVER* fought one another. They have nearly-identical peace weights, they were situated on opposite sides of the map, and Cyrus frequently wound up adopting the state religion that Pacal was running. Furthermore, Cyrus is one of the leader that will not declared war at "Pleased" relations which prevented him from attacking so long as Pacal could reach as little as +1 on the diplomatic modifier. This meant that Cyrus couldn't attack Pacal at all in the vast majority of games.
Thus Pacal had an absolutely perfect setup: a dream capital paired with the only other strong AI on the map refusing to attack him. Of course, Pacal will declare war at "Pleased" relations and could choose to go after Cyrus if desired, but Pacal has low odds to plot war and almost always preferred to sit back and tech away into space. Most of the endgames in the alternate histories therefore turned out to be quite peaceful as both leaders researched away in a competition that Pacal would inevitably win. There were several games where Cyrus was the dominant AI and would have won easily if he had simply attacked Pacal, however his AI programming stopped him from taking action and allowed the Mayans to claim victory. Only in Game #18 was Cyrus able to take home the sole non-Pacal win thanks to an efficient snowball over Victoria, Shaka, and Bismarck while Pacal was dragged down with too many wars. I'd estimate that the true win percentage for Pacal was somewhere between 90-95 percent on this map and with slightly better luck he might have won all 20 matches, something that we haven't seen before in an alternate histories series. Heck, this is the first time anyone has made it to 19 out of 20 which is insanely impressive!
The other four leaders were irrelevant and didn't matter to the outcome of the game which was virtually hardcoded to procuce a Pacal/Cyrus finishing order. Shaka technically had the third place score simply because he engaged in nonstop warring and picked off more kills than the rest of this group. Surprisingly, he also mananged a survival rate around 50% which was far better than those other leaders. Shaka was never even remotely in contention to win, however, and he only had a single top two finish when Pacal was pulled into a lategame war against Cyrus in Game #20 and unintentionally backdoored the Zulus into the runner up spot. Shaka's economy was terrible as always and he was technologically outscaled by Pacal in every game leaving him with no chance to claim a victory.
The other three leaders were even worse if you can believe it. All of them were eliminated 80% of the time or higher and lacked even Shaka's random kills to score points under our AI Survivor system. Victoria managed the only other top two finish with a second place result in Game #6 where she actually won the straight up 1 vs 1 duel against Cyrus. Those two fought in almost every game and the Persians invariably prevailed in the vast majority of cases. Outside of that single success, Victoria failed to achieve anything else and largely served as roadkill for Cyrus in game after game. Bismarck was never a strong AI at any point in time as he seemed to struggle badly with his starting position. Of course, he was getting attacked constantly by Shaka from one side and pressured by Pacal's immense culture from the other side which was an unenviable location. I can't recall the Germans ever being atop the leaderboard and Bismarck was a repeated punching bag in the alternate histories. As for Zara, he liked to pick fights with Pacal and, well, you can imagine how that went on this map. Zara typically founded a different religion from Pacal and that served to drive conflict between the two of them, conflict which almost always backfired for the Ethiopians. There wasn't really any difference between Victoria and Bismarck and Zara on this map, they were all truly horrible performers who were relentlessly crushed by Pacal and Cyrus.
Now for a look at the individual leaders:
Pacal of the Mayans
Wars Declared: 30
Wars Declared Upon: 34
Survival Percentage: 100%
Finishes: 19 Firsts, 1 Second (97 points)
Kills: 20
Overall Score: 117 points
It was total economic domination in another playoff game from this season. Only a handful of the 20 games were close and most of them looked very much like the actual Playoff Game Two where Pacal annihilated the rest of the field. I mentioned this above but Pacal didn't win simply by getting lucky and avoiding wars; he was attacked about as frequently as everyone else in the game. Pacal was quite content to sign treaties even when winning wars so that he could go back to teching away in peace, the only reason why he didn't have more kills despite being utterly dominant. These alternate histories definitely showed Pacal's weaknesses as an AI leader, with him often stalling out expansion on six or seven cities so that he could build more wonders and found more religions. Those tendencies often get him into trouble on other maps but the Mayans were simply too strong in this scenario to be stopped.
Cyrus of Persia
Wars Declared: 32
Wars Declared Upon: 22
Survival Percentage: 100%
Finishes: 1 First, 17 Seconds (39 points)
Kills: 24
Overall Score: 63 points
Cyrus was extremely strong on this map and had the misfortune to run up against an even stronger competitor. He put his Imperialistic trait to good use and expanded very well across the western side of the map, sometimes getting as many as 15 cities from peaceful settling if Victoria or Shaka found themselves tied up in early warring. Cyrus clashed most frequently with Victoria and his strength in these games often hinged upon how quickly he could eliminate the English from the map. Cyrus won that clash in every match other than Victoria's sole runner up finish in Game #6 but the speed with which Persia achieved its victory dictated whether Cyrus could contend for a victory versus tagging along as a distant runner up. He came very close to winning Game #4 and Game #5 with narrow Spaceship losses to Pacal and lost to a Mayan Cultural victory by a mere 2 turns in Game #8 (although that was at an extremely late date after Pacal threw away his own certain Spaceship win). Then Pacal's Cultural victory landed 2 turns before Cyrus' own Spaceship victory again in Game #10 for another nailbiting finish. The inability of Cyrus to declare war at "Pleased" relations was the biggest thing stopping him from snowballing his position and winning multiple Domination victories. A major difference from the Livestream game was that Cyrus fought a lot less with Shaka than what we saw; they did fight the occasional war but most of the conflict for Cyrus was with Victoria. This was a great showing for Cyrus and he simply couldn't overcome Pacal's superior traits and starting position.
Shaka of the Zulus
Wars Declared: 60
Wars Declared Upon: 11
Survival Percentage: 45%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 11
Overall Score: 13 points
Shaka had exactly the AI personality that everyone would have expected, fighting an endless sequence of wars that resulted in a decent number of kills but little else of value. He was pretty good at pruning the deadwood of the other weak AI leaders in the field while being utterly useless against the top two AI personalities. He never defeated either Pacal or Cyrus in a war and never killed either of them (not that anyone did - they both had perfect 100% survival rates!) Shaka fought a ludicrous 60 offensive wars against only 11 defensive wars; he was the biggest factor in stirring the pot and driving the action of these games. He was eliminated about half the time and the alternate histories tended to become quite peaceful once he was removed from the scene. Shaka's economy tends to fare poorly in a normal game of AI Survivor and he found himself in an even worse position here against Pacal's unstoppable teching prowess. The Zulus would rampage around the map for the first 200 turns and then inevitably get rolled up against more advanced opposition as the game progressed. I'm honestly a bit surprised that he survived as often as he did, probably because Pacal and Cyrus both share the same low peace weight and Pacal had little interest in conquest. Shaka didn't have much success on this map but at least he made things interesting to watch.
Victoria of England
Wars Declared: 21
Wars Declared Upon: 37
Survival Percentage: 15%
Finishes: 0 Firsts, 1 Second (2 points)
Kills: 2
Overall Score: 4 points
Bismarck of Germany
Wars Declared: 17
Wars Declared Upon: 39
Survival Percentage: 20%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 2
Overall Score: 2 points
Zara Yaqob of Ethiopia
Wars Declared: 22
Wars Declared Upon: 39
Survival Percentage: 20%
Finishes: 0 Firsts, 0 Seconds (0 points)
Kills: 1
Overall Score: 1 point
The remaining three AI leaders were all terrible on this map and any difference in their scoring was purely due to small sample size bias. They were all eliminated at the same 80-85% rate and were never in contention to win a game at any point in time. Victoria had the only runner up finish in Game #6 where she was a massive beneficiary of an early 3 vs 1 dogpile against Zara which allowed her to double in size and then defeat Cyrus in their typical midgame showdown. Outside of that very unusual result, she was conquered by Cyrus in almost every match; you'll note the light blue borders of Persia occupying the English portion of the map in all of those endgame screenshots that I post at the top of this page. For his part, Bismarck was a total failure on this map and seemed to struggle with his starting position worse than I expected. It was by no means terrible, with several nearby food bonuses, but lacked much in the way of river tiles and was badly outdone by the offerings in the Mayan/Persian domains. Bismarck was a weak leader in all of the alternate histories and couldn't muster more than two random kills. There was no noticeable difference between him and Zara though, with the Ethiopian leader suffering the same rate of elimination and a near-identical ratio of offensive to defensive wars. Zara's demise tended to be more self-inflicted in most games as he repeatedly picked fights with Pacal (usually over religious divisions) which backfired horribly in game after game. The only real drama here was over who would be First to Die in each game which tended to be evenly split between Victoria, Bismarck, and Shaka. Zara had slightly lower odds to suffer the first exit which is the best thing that can be said about him.
I have to be honest, these weren't the most interesting alternate histories for me to play out. After watching Pacal take first place and Cyrus take second place a dozen times in a row, there wasn't exactly a lot of drama going into the next replay of the map. The biggest thing holding my interest was Pacal's quest to post a perfect 20 wins in 20 games which was finally snapped at the last minute when Cyrus won in Game #18. Otherwise, these games looked quite similar to what we watched on Livestream with the same two leaders occupying the top two spots again and again and again. There wasn't even much separation in how those results were achieved either: outside of the unusual Pacal Cultural victory on Turn 360 in Game #8, all of these games finished between Turn 277 and Turn 330. In fact, 17 of the 20 games finished between Turn 285 and Turn 317 - that's a narrow distribution of results! Every single game also landed between 7 and 11 wars, never anything higher or lower than those totals. This was an extremely predictable series of games where the dominoes kept falling over in the same order no matter how many times I ran the opening moves.
Thanks as always for reading, I hope you enjoyed this look back at Playoff Game Two!
