Civ4 AI Survivor Season 7: Conclusions

Season Seven was the season of upsets, even more so than Season Six had been, and as a result there were a ton of fresh faces advancing into the playoffs and then further on into the Championship. Hammurabi, Ramesses, Montezuma, and eventual overall winner Louis were all leaders who had never won a game previously before going on to great success in this season. In fact, hardly any Pool 1 or Pool 2 leaders were able to advance into the playoffs and then none of them managed to make the Championship which was a great contrast to what we had seen in past years. Season Six also had a bunch of upsets in the opening round only for the heavyweights to reestablish themselves in the playoffs; this time around, the upsets never stopped from beginning to end until we had crowned a new and deserving champion. I tallied up the numbers and the Pool 1 leaders coming into Season Seven scored 15 points, Pool 2 leaders scored 7 points, and Unseeded leaders scored 111 points in Season Seven. That's nothing short of incredible and speaks to what a year of upsets we had.

Thus a number of these leaders scored more points in Season Seven than in the first six seasons combined. Montezuma went from 1 point to 7 points with his extremely unlikely opening round victory, Ramesses went from 2 points to 10 points with his run to the Championship, Saladin went from 8 points to 19 points after somehow compiling 11 points across four games, and Louis was the biggest winner of all by going from 8 points to 25 points with a truly impressive triple victory performance. Almost every leader has won a game at some point by now in past AI Survivor history, and the Sitting Bull / Hatshepsut pair are the only ones below 5 points. Poor Hatshepsut remains the only one of the 52 leaders who hasn't scored a single point yet across seven seasons although the alternate histories showed that she was the runaway AI on her map and was fantastically unlucky to be defeated by Montezuma. Sooner or later she's going to break through; I can think of several different AI leaders who are clearly worse overall.

As always, the kills were heavily concentrated amongst the leaders who made the playoffs though not to the same degree as in past seasons. The 27 leaders that fell in the opening round managed 2 kills between them which is about typical for the group that fails to advance further. At the other end of the spectrum, the 6 leaders that made the Championship collected only 12 trophies between them and with none of them getting more than 4 kills. Darius was the incredibly unlikely winner of the Golden Spear after Mehmed failed to kill-steal his way to the prize in the final match. This simply wasn't a big year in terms of military performances, nothing like Season Three where Julius Caesar and Stalin both had 7 kills apiece with a bunch of Domination victories notched on their belts. I suspect this was due at least in part to so many Cultural victories hitting in the playoff round and Championship; only one of the final five matches went to the Domination ending.

The AI ranking list has also been updated to reflect an additional year of competition (the size of this is really starting to get out of hand but I left the image alone). As a quick reminder of how the ranking system works, leaders score 5 points for a first place finish, 2 points for a second place finish, and 1 point for each kill. This is a completely arbitrary system, of course, but it seems to do a decent job of recognizing AI performance across different games and different seasons. I included an additional column listing the points scored solely in Season Seven to help indicate which leaders were the biggest movers in this season. This was a season where our historical performers at the top of the table almost universally faltered and allowed room for some newcomers to start pushing their way up the table. Kublai Khan claimed an opening round victory and Pacal had a very misleading 5 points from a Runner Up finish and three kill-steals before he was crushed in his playoff match. Otherwise the leaders at the top achieved basically nothing though, a huge contrast to how well Mansa Musa and Huayna Capac and Pacal all did back in Season Six. We'll see if they can rebound back in the next season.

Thanks to those unexpected victories in the opening round, there are fewer leaders than ever down at the 1 point or 2 point threshhold. Hatshepsut and Genghis Khan are the only leaders who have never managed a first or second place finish whereas Sitting Bull and Hatshepsut again are the only leaders who haven't scored a single kill. We know from the alternate histories that Hatshepsut was really unlucky in this regard and should have gotten off the zero mark this year. We've now run enough games that most of the mediocre leaders fall into the "one good game" category. This includes one-off showings from leaders like Bismarck and Isabella and Lincoln; odds are that this is where Frederick and Sitting Bull and the like will find themselves in future seasons. These leaders can be effective given the right starting position or a uniquely favorable diplomatic situation but aren't good enough to consistently put up big numbers like the seeded leaders. (I had Churchill in this category before but he's scored enough Runner Up finishes by now that he seems to have a knack for hanging around.)

This was a season of opening round upsets and as a result the points were a bit more evenly distributed when compared to past years. In Season Four, the leaders in the bottom half of the list accounted for 7 points while the leaders in the top half totaled a whopping 128 points. (The same ratio for this season was 95 points against 38 points - still lopsided but nowhere near the same degree.) Obviously as we continue to run more seasons of AI Survivor the results will get less stratified in this manner since the effects of any one season get reduced. Big single game performances from the aforementioned Hammurabi, Ramesses, and Montezuma helped inject some additional points into the bottom parts of the leaderboard. Special mention also should go to Napoleon who scored ANOTHER 3 kills and has NINE kills in total without ever finishing first in a game; the only other comparable leader is Genghis Khan with 5 kills thanks to a similar AI personality. It's honestly amazing that this guy could score that many kills, more than 80% of the other leaders in the field, without having a first place finish somewhere along the way.

After seven seasons, the average score for an AI leader is now 17.72 points. I've noticed that the average goes up by about 2.50 points each season which makes perfect sense: there are 13 * 7 = 91 finish points available and we typically see about 42 eliminations. 91 + 42 = 133 and divide that by 52 leaders for roughly 2.56 additional points per leader per season. Obviously these points are not evenly distributed and instead tend to cluster at the top end of the leaderboard, which is why the average is 17.72 points but the median is only 15 points. Amusingly Churchill and Asoka are currently the median leaders with Churchill remaining exactly in the middle of the table two years in a row. (You might think that Churchill should have moved up higher for making the Championship again but two Runner Up finishes and zero kills only equates to four points which largely held him in the same spot.) The median leader has about 5 kills at this point which helps show the exceptional status of the top of the group. The median has been going up by about one kill per season in the last few years; the top of the table didn't move so Huayna Capac and Julius Caesar remain deadlocked for the total kill lead at 15, with both of them having zero kills in Season Seven.

Another year's worth of data has allowed us to update the Trait rankings for Season Seven. We've done enough seasons at this point that a single additional year of competition doesn't swing the ratings around too noticeably. The traits that were strong in past seasons largely graded out as being strong once again. The one major change in the table this year was Imperialistic having a poor showing, causing it to drop back from near-parity with the Financial and Creative traits. There seems to be clear separation at this point between the top traits of Financial and Creative from the rest of the field after Creative had a great year and Financial, well, did its usual thing. There's no question that Financial is one of the best traits for AI Survivor purposes as the extra commerce from Financial definitely helps a lot and the other expansion-oriented traits of Creative / Imperialistic are helpful to the AI leaders for obvious reasons. They let the AI claim more land on the map and then snowball off of having more total production queues. Most of our top-performing leaders have at least one of these three traits at their disposal.

There was another clump of six different traits in the middle of the pack that graded out as having roughly similar scores. Industrious, Expansive, Spiritual, Organized, Aggressive, and Protective were all roughly comparable to one another without much differentiation between them. I wrote the exact same thing about these six traits at the end of Season Six and their continued ranking in the middle of the group suggests that they are all roughly comparable to one another. It doesn't appear that these traits are particularly great or particularly terrible for AI Survivor purposes. The Protective trait is still hanging solidly in this rankings (although it did slightly worse in Season Seven) so it appears to be pretty average in AI hands, not the "worst trait in the game" status that Protective has when played by humans. We've seen that the extra Protective promotions and cheap walls/castles genuinely do seem to help out when the AI leaders inevitably get into stupid wars. At the very least, it's hard to argue that Protective is noticeably worse than the other traits under these cirumstances.

The Industrious trait continues to feel overvalued due to the presence of Huayna Capac, Stalin, and after this season Louis, with all of the other leaders having poor seasons but the trait grading out as average instead of terrible due to the three of them. Stalin increasingly looks like an early season fraud (with alternate histories showing that he was really lucky in Season Three) but Louis emerged this year as a real competitor in his own right. I think that this is one of the weakest traits for AI Survivor purposes that only looks decent because the best leader in the game (our boy Huayna) happens to have the trait. Stalin continues to return to earth and it was only the brilliance of Huayna Capac and now Louis that keeps holding up the value of the trait. I continue to think this is a bad trait for AI Survivor that happens to have some excellent leaders propping up its value.

Charismatic remained in next-to-last status this season, and that was a bit of a surprise because it seems to be helpful for the AIs in terms of managing happiness. Even Churchill making another run to the Championship game wasn't enough to bump the performance of Charismatic and this is increasingly looking like one of the weakest options for AI Survivor purposes. There was a clear gap between the glut of traits in the middle of the pack and then the Charismatic/Philosophical duo at the bottom. Speaking of the Philosophical trait, it once again graded out as the worst in the competition for the fourth year in a row. It's been a precipitous fall from the Season One version of these rankings which had Philosophical as the best trait overall. That was under the old Finish Ranking system that heavily valued silly stuff like not getting eliminated in the Wildcard game so it's not a direct comparison, but it's still noteworthy how the trait has collapsed in value over the following seasons. Philosophical is a poor trait for the AI leaders because they inevitably run so many specialists that they always get lots of Great People regardless. Plus the AI has no idea how to leverage individual Great People for timed Golden Ages, and of course we turn off Tech Trading to remove the value of lightbulbing followed by tech brokering. The Philosophical trait also happens to be paired with lots of pacifistic leaders with high peace weights, and as we've seen repeatedly, those are some of the worst-performing AI personalities under our settings. Philosophical is a solid trait overall in Civ4 but it clearly struggles under the games that we run.

We brought back the Fantasy contest for AI Survivor for Season Seven and once again it proved to be a huge success. After rolling the maps for the season and then drawing the 52 leaders onto those maps, we held an auction draft where each of the eight contestants had 200 gold to spend however they chose; I was the moderator for the auction this year after participating (and winning!) back in Season Six. Our fantasy contestants could invest in a handful of favorites or spread out with lots of smaller bids on less popular leaders. Fantasy scoring was identical to our normal AI Survivor setup with 5 points for first place, 2 points for second place, and 1 point per kill, along with an extra 2 points for having a First to Die leader to attract bidding on leaders who seemed likely to make a quick exit. This was enormously entertaining and I had a blast tracking the results of the fantasy competition throughout the season... even if it turned out that there wasn't much drama thanks to Amicalola stomping the field. Amica had the top two scoring leaders in Louis and Darius along with excellent second tier bids on Mehmed and De Gaulle to run away from strong fantasy teams fielded by cutHrAXys and Slashin'. Amicalola will have a chance to come back and defend the fantasy crown next season; we'll see if there are seven new faces willing to try their luck at the format next time around!

There will also be a more in-depth report looking at the fantasy competition once the alternate histories for Season Seven are complete. That will give us a better indication of who made excellent bids in the auction and who simply got lucky with their picks. Expect to see the results of this analysis in a few months when the alternate histories have been finished.

We crowned Louis as the overall winner at the end of Season Seven and it was a very deserving victory indeed. Louis won all three of his matches in Season Seven going away, racing out to score leads and then turning on the cultural slider to wrap things up at an early date. He never gave anyone a chance to head for Spaceship or Domination, instead taking his Cultural victory on Turn 318, Turn 281, and Turn 272. If anything, he was actually getting faster as the competition continued despite the opposition nominally getting more difficult. Louis had never scored a single first place finish before Season Seven which meant that these results came a bit out of nowhere. The new community view of the French king is that he's basically Huayna Capac without the Financial trait: similar peace weight, similar wonder-building preferences, similar affinity for Cultural victories, etc. The historical data suggests that Huayna Capac is definitely the better of the two leaders, however even a budget HC turns out to be a pretty darn good leader in this competition. Louis had some favorable starting positions in Season Seven and he never looked back, smashing the competition to establish himself as the clear winner for this season.

Anyway, I'd like to give thanks here once again to everyone here who took part in Season Seven of AI Survivor. Even if it was nothing more than spending two minutes to submit a prediction, the competition wouldn't have been the same without all of your contributions. Turnout in the picking contest was noticeably up from Season Six, with about 250 predictions on average for each game, basically back to Season Five levels at the height of the COVID pandemic and essentially the best we've ever seen. Season Seven held seemingly the entire audience from the previous year and added new faces as well, averaging well above 200 viewers watching along on the Livestream. We peaked at 305 live viewers in a match at the beginning of June which I believe is the highest I've ever had outside of IPL casting for League of Legends. A special shout-out to the CivFanatics contingent who added their own contributions with weekly forum threads and playoff betting odds to spice things up further!

I continue to be amazed at how many people are interested in watching the computer play against itself. This event simply would not be what it is without the lively banter in the chat taking place alongside the action. I am surprised again and again at the brilliant insights that the viewers come up with in their predictions, along with the hilarious and inane running commentary that makes it such a pleasure to put these games together. Over the upcoming months we'll take some time to look at alternate histories for other games from Season Seven and think about any additional fixes that we might add for the future. With any luck, we'll be back again next year for Season Eight - thanks again everyone!