3.6 AI Only Game Statistics 700 game update (350 Emperor / 350 Warlord) + vote for next config

L. Vern · Jun 26, 2023

Includes the 263 from the previous post plus an additional 87, as well as 350 warlord games. Results are overall quite different on warlord difficulty compared to emperor; results with no AI bonuses might be insightful. I've also included a Poll this time to decide on the settings for the next batch of games to play, can vote for multiple options if you're interested in multiple (d4 = difficulty 4)

My quick thoughts:

pretty interesting how without the instant yield bonuses France performs better than Japan, could be largely how the civ scales with instant yields behind it
Austria pretty good on both, although interestingly more culture focused on Warlord
Warmonger civs in general seem to be performing way better on Warlord (Songhai, Iroquois, Greece, Rome in particular doing quite well) which could explain a good portion of the larger share of time victories, although in general civs are having a hard time reaching a win condition (maybe also cause they're getting rekt by warmongers)

Spoiler Victory Types :

Spoiler Civilizations (Graph) :

Spoiler Civilizations (Graph with 95% Confidence Intervals*) :

Spoiler Civilizations (Table) :

Emperor:

	Civ	Games	Victories	Winrate	Culture Victories	Diplomatic Victories	Science Victories	Time Victories	Domination Victories	Average Victory Score	Average Score Rank	Turn 100 Average Score	Turn 100 Average Rank	Turn 200 Average Score	Turn 200 Average Rank	Turn 300 Average Score	Turn 300 Average Rank	Turn 400 Average Score	Turn 400 Average Rank	Turn 500 Average Score	Turn 500 Average Rank
1	Austria	66	18	0.27	2	16	0	0	0	1941.92	4.14	382.14	4.06	952.94	4.15	1445.08	4.32	1913.12	4.35	791.8	6
2	Arabia	61	16	0.26	16	0	0	0	0	1324.92	4.56	366.54	4.92	908.67	4.7	1279.36	4.81	1112.29	5.71	446.4	6.4
3	Japan	58	15	0.26	5	0	4	0	6	2736.69	3.64	402.19	3.52	1131.9	3.6	2088.29	3.41	2389.18	4.11	2956	4
4	The Ottomans	66	17	0.26	4	5	6	0	2	2448.03	3.82	404.41	3.71	1024.08	3.74	1733.91	3.7	2678.97	3.5	1159.5	6.25
5	Germany	69	17	0.25	7	8	0	0	2	1587.04	4.64	359.3	5.19	885.59	4.52	1369.22	4.72	1354.06	5.39	nan	nan
6	Greece	79	19	0.24	3	13	2	0	1	1921.97	4.54	391.68	3.78	953.87	4.09	1439.91	4.75	2013.66	4.53	1419.6	5.3
7	Poland	53	10	0.19	9	0	1	0	0	1412.15	4.79	390.38	3.91	849.32	4.74	1284.06	4.81	1342.1	5.21	918.5	5.75
8	China	68	12	0.18	10	1	1	0	0	1727.24	4.31	380.22	4.13	925.12	4.32	1501.6	4.22	1661.43	4.51	2316.4	3.8
9	Rome	63	11	0.17	0	1	0	4	6	2867.7	3.62	360.83	5.06	984.57	4.11	1726.68	3.81	2470.76	3.7	4050	3.29
10	India	63	11	0.17	8	1	2	0	0	1585.1	4.59	376.06	4.38	938.75	4.27	1390.46	4.4	1279.45	5.55	770.67	5
11	Korea	86	15	0.17	6	0	9	0	0	1522.21	4.71	367.67	4.76	812.41	5.13	1229.49	4.98	1373.59	5.3	0	7.5
12	Brazil	64	11	0.17	7	0	4	0	0	1751.59	4.31	379.27	4.36	886.34	4.47	1441.75	4.27	1713.05	4.76	535	6.33
13	Babylon	61	9	0.15	4	0	5	0	0	1591.15	4.49	401.69	3.54	906.59	4.77	1317.21	4.85	1268.1	5.52	929.33	5
14	The Inca	62	9	0.15	2	0	2	0	5	2050.52	4.45	400.56	3.6	977.44	4.26	1583.77	4.45	1836.88	4.96	839.25	6.5
15	Morocco	76	11	0.14	3	7	1	0	0	1534.7	4.74	344.61	5.72	804.84	5.13	1224.45	5.04	1479.43	5.17	1099.33	6
16	Indonesia	73	10	0.14	2	0	7	1	0	2080.48	3.99	417.45	3.14	1077.71	3.34	1649.41	3.73	2163.19	4.33	2563.67	5
17	Assyria	66	9	0.14	2	0	7	0	0	2018.17	4.35	409.32	3.38	1000.09	3.83	1548.61	4.24	2101.08	4.22	1150	5.67
18	Egypt	54	7	0.13	7	0	0	0	0	1122.69	5.13	360.87	4.93	832.43	4.85	1125.61	5.33	950.06	5.94	0	7
19	Ethiopia	63	8	0.13	7	0	1	0	0	1559.46	4.49	397.49	3.84	903.87	4.48	1321	4.71	1565.07	5.03	859.5	7
20	The Zulus	73	9	0.12	0	1	0	1	7	2267.7	4.51	349.53	5.37	849.03	4.85	1433.95	4.71	2551.07	4	4406	3.67
21	Siam	62	7	0.11	0	6	1	0	0	1210.35	5.32	380.4	4.02	747.26	5.61	1050.43	5.68	1019.88	5.94	nan	nan
22	Byzantium	65	7	0.11	3	2	2	0	0	1522.57	4.68	363.14	5.03	894.95	4.63	1329.46	4.67	1723.39	4.27	1807.33	4.33
23	The Iroquois	56	6	0.11	1	2	0	2	1	2808.64	3.41	381.89	4.27	1045.39	3.71	1831.8	3.5	2801.23	3.31	4599	3
24	Sweden	69	7	0.1	0	1	0	3	3	2177.29	4.28	345.88	5.77	834.06	5.19	1400.01	4.81	2045.57	4.66	2703.78	3.67
25	The Aztecs	53	5	0.09	1	1	0	0	3	2875.42	3.75	384.47	4	1014.91	3.72	1576.06	3.98	2807.61	3.9	2826	3
26	Songhai	85	8	0.09	0	1	1	2	4	2225.64	4.13	366.48	4.81	953.39	4.44	1545.34	4.43	2177.71	3.89	3044.73	3.27
27	America	77	7	0.09	2	3	0	2	0	2110.69	4.09	392.04	3.99	925.45	4.32	1397.36	4.61	2181.12	3.81	4068.8	2.2
28	Portugal	56	5	0.09	1	2	2	0	0	1445.54	4.96	353.11	5.23	805.07	5.16	1249.12	5.11	1518.88	5.18	1172	5.5
29	England	57	5	0.09	0	2	0	0	3	2150.11	3.86	353.63	5.14	916.58	4.47	1478.81	4.16	2057.7	4.58	3825	2
30	Russia	71	6	0.08	1	0	5	0	0	1882.96	4.32	391.46	3.87	853.3	4.76	1352.38	4.7	2093.5	4.32	844	6
31	The Celts	50	4	0.08	1	0	2	0	1	1394.36	5.28	408.08	3.26	886.82	4.56	1216.73	5.1	1448.75	5.59	992.75	6.5
32	The Netherlands	65	5	0.08	3	2	0	0	0	1502.28	4.66	351.55	5.46	885.51	4.85	1301.31	4.71	1418.12	5.45	1082.67	6
33	Persia	55	4	0.07	2	1	0	1	0	2034.84	4.22	366.75	4.71	901.16	4.55	1446.56	4.22	2177.62	3.97	2280.4	4.8
34	The Maya	72	5	0.07	2	1	2	0	0	1362.81	5.06	392.94	3.82	909.76	4.5	1160.31	5.25	1288.55	5.74	0	8.5
35	The Shoshone	82	5	0.06	1	3	0	1	0	1545.52	4.56	366.13	4.59	882.67	4.66	1301.56	4.81	1660.12	4.6	2440.75	4.88
36	Denmark	52	3	0.06	0	0	0	1	2	1827.94	4.6	326.42	6.08	842.77	4.9	1310.41	4.96	1931.71	4.61	3042.83	3.17
37	Mongolia	64	3	0.05	0	1	0	2	0	1315	5.06	343.12	5.61	759.28	5.31	1087.19	5.41	1659.11	4.97	2516.29	4.43
38	Carthage	65	3	0.05	1	1	1	0	0	1342.74	4.71	416.77	3.37	964.17	4.02	1334.7	4.69	1572.97	4.53	1056.5	5.25
39	Polynesia	66	3	0.05	2	0	1	0	0	1531.2	4.7	378.44	4.58	934.71	4.55	1402.85	4.68	1524.6	5.03	1323.4	5
40	Spain	66	3	0.05	0	1	0	1	1	1384.62	5.24	340.42	5.94	777.36	5.59	1162.85	5.21	1566.53	5.16	3313	3.75
41	Venice	60	2	0.03	1	1	0	0	0	1395.15	4.72	413.53	3.25	982.38	3.78	1200.45	4.92	1332.25	5.32	1296.5	6
42	The Huns	60	2	0.03	1	0	0	0	1	1311.98	5.13	352.5	5.12	885.07	4.88	1147.76	5.22	1671.26	4.84	1208.43	5.43
43	France	68	1	0.01	0	0	0	1	0	1272.85	5.03	340.04	6.03	821.79	5.32	1209.78	4.96	1268.24	5.32	1654	5.75

Warlord:

	Civ	Games	Victories	Winrate	Culture Victories	Diplomatic Victories	Science Victories	Time Victories	Domination Victories	Average Victory Score	Average Score Rank	Turn 100 Average Score	Turn 100 Average Rank	Turn 200 Average Score	Turn 200 Average Rank	Turn 300 Average Score	Turn 300 Average Rank	Turn 400 Average Score	Turn 400 Average Rank	Turn 500 Average Score	Turn 500 Average Rank
1	Austria	61	18	0.3	11	3	0	4	0	1739.82	4.16	286.25	3.23	651.2	4.25	1012.03	4.49	1451.6	4.43	1630.78	4.43
2	The Netherlands	70	19	0.27	12	4	0	1	2	1951.27	3.81	268.81	3.93	653.83	4.11	1094.76	4.13	1600.81	3.99	1719.46	4.25
3	China	59	16	0.27	11	0	0	5	0	1843.56	3.93	292.58	2.54	680.63	3.64	1141.12	4.03	1536.79	4.28	1990.85	4.08
4	Songhai	67	17	0.25	0	0	0	12	5	1918.61	4.09	237.15	5.61	627.87	4.39	1057.35	4.08	1428.19	4.37	2071.3	4.19
5	The Iroquois	66	16	0.24	0	0	0	13	3	2489.73	3.05	250.95	4.68	701.08	3.52	1241.58	3.42	1962.1	3.3	2584.67	3.15
6	India	68	16	0.24	13	0	0	3	0	1649.91	4.1	274.29	3.81	655.54	4.07	1124.72	3.75	1579.9	3.89	1425.28	5.07
7	The Aztecs	55	11	0.2	0	0	0	6	5	1882.42	4.29	242.49	5.33	678.18	4.11	1072.8	4.47	1441.4	4.58	1905.44	4.22
8	Greece	76	15	0.2	4	2	0	8	1	1684.78	4.55	278.84	3.16	672.93	3.71	990.37	4.53	1319.91	4.89	1992.83	4.61
9	Poland	72	14	0.19	13	0	0	1	0	1547.15	4.11	276.94	3.47	665.31	3.72	1067.51	4.01	1483.21	4.22	1443.72	5.08
10	Carthage	68	13	0.19	4	0	0	6	3	1727.43	4.29	289.37	2.93	701.79	3.51	1069.91	4.26	1457.82	4.5	1576.89	4.48
11	Siam	59	11	0.19	7	2	0	2	0	1759.59	4.1	283.27	3.36	628.44	4.51	1009.03	4.47	1486.72	4.47	1580.24	4.33
12	Germany	71	13	0.18	5	3	0	5	0	1553.94	4.35	271.01	3.89	617.87	4.34	974.18	4.48	1274.47	4.79	1590.86	4.44
13	Indonesia	66	12	0.18	6	0	0	5	1	1895.74	4.11	266.26	4.03	676.62	3.56	1123.27	3.97	1564.52	4.25	2206	3.88
14	Rome	58	10	0.17	0	0	0	8	2	1632.52	4.43	219.31	6.59	583.38	5.12	941.64	4.84	1331.91	4.68	1718.39	4.58
15	Denmark	80	12	0.15	0	0	0	9	3	1718.99	4.21	229.92	5.72	596.34	4.75	1059.1	4.13	1405.83	4.48	1916.26	3.93
16	Brazil	62	9	0.15	4	0	0	5	0	1925.63	3.6	258.13	4.71	624.81	4.39	1167.89	3.84	1637.39	3.78	2146.29	3.71
17	America	65	9	0.14	1	0	0	7	1	1213.29	5.23	241.72	5.52	531.09	5.68	783.98	5.62	1061.87	5.42	1364.88	5.44
18	The Inca	64	8	0.12	1	0	0	7	0	1761.62	4.2	257.25	4.41	675.7	3.66	1124.11	4.05	1554.11	4	2262	3.77
19	Morocco	70	8	0.11	8	0	0	0	0	1412.56	4.57	251.31	5.06	571.63	5.4	990.31	4.73	1294.58	4.61	1201.25	5.28
20	France	71	8	0.11	3	0	0	5	0	1245.31	4.97	230.77	6.01	512.24	5.99	773.27	5.67	1082.25	5.4	1445.32	4.89
21	Sweden	72	8	0.11	0	0	0	5	3	1370.65	4.92	230.92	6.15	539.68	5.71	802.9	5.58	1073.85	5.3	1438.41	5.21
22	The Ottomans	46	5	0.11	0	1	0	3	1	1919.24	3.8	268.78	4.04	680	3.57	1174.91	3.62	1622	3.73	1966.23	4.05
23	Arabia	60	6	0.1	6	0	0	0	0	1264.92	4.9	265.93	4.53	669.07	4.2	1049.05	4.3	1334.62	4.69	1222.4	5.32
24	Persia	74	7	0.09	0	0	0	4	3	1510.66	4.66	240.14	5.34	563.68	5.28	868.03	5.19	1203.87	5.2	1486.46	5
25	Korea	57	5	0.09	4	0	0	1	0	1049.44	5.11	260.7	4.67	557.49	5.37	856.44	5.18	1054.29	5.13	949.88	5.68
26	England	70	6	0.09	0	0	0	4	2	1819.66	4	240.14	5.36	607.06	4.79	1024.41	4.51	1546.84	4.19	2265.31	3.54
27	Egypt	50	4	0.08	4	0	0	0	0	1413.26	4.48	267.6	3.98	620.36	4.5	1025.41	4.55	1260.76	4.83	1116.5	5.27
28	Babylon	76	6	0.08	4	0	1	1	0	1344.57	4.49	273.25	3.42	654.26	4.11	1022.59	4.43	1284.46	4.53	1068.36	5.58
29	Venice	80	6	0.08	4	1	0	1	0	1596.64	4.1	297.2	2.81	754.26	2.69	1167.04	3.5	1430.58	4.22	1444.56	4.56
30	Assyria	57	4	0.07	0	0	0	3	1	1321.25	4.84	267.56	4.09	682.46	3.46	1049.68	4.23	1293.35	4.69	1421.81	5
31	The Shoshone	79	5	0.06	2	0	0	2	1	1537.84	4.54	260.1	4.35	619.75	4.53	983.86	4.71	1310.47	4.84	1396.21	5.12
32	The Huns	65	4	0.06	0	0	0	4	0	1036.31	5.12	242.09	5.34	534.43	5.15	791.02	5.36	992.9	5.41	1341.19	5.46
33	The Maya	82	5	0.06	2	0	0	3	0	1092.99	5.11	267.7	4.17	630.78	3.95	913.42	4.72	1061.06	5.19	1332.89	5.24
34	Portugal	69	4	0.06	0	3	0	1	0	1521.41	4.43	257.26	4.41	564.26	5.2	963.06	4.8	1318.88	4.5	1377.75	4.86
35	Ethiopia	57	3	0.05	1	0	0	2	0	1261.37	4.88	267.37	4.14	607.26	4.56	907.35	4.82	1161.02	5	1334.72	4.9
36	Japan	60	3	0.05	1	0	0	2	0	1275.75	5.1	260.72	4.43	630.22	4.9	992.68	4.87	1284.43	5.05	1484.6	5.57
37	Polynesia	62	3	0.05	0	0	0	3	0	1345.53	4.68	254.53	4.77	598.97	4.89	978.76	4.52	1244.11	4.82	1562.85	5.04
38	The Zulus	68	3	0.04	0	0	0	2	1	1265.88	4.84	233.88	5.93	526.68	5.71	805.1	5.46	1143.59	4.89	1414.55	5.03
39	Russia	52	2	0.04	0	0	0	2	0	1341.87	4.6	274.08	3.62	628.29	4.35	913.94	4.63	1172.38	4.98	1728.74	4.32
40	Mongolia	57	2	0.04	0	0	0	2	0	1153.96	5.14	223.51	6.53	510.75	5.91	788.86	5.57	1038.91	5.38	1310.28	5.44
41	Byzantium	58	2	0.03	0	1	0	1	0	974.47	5.24	257.03	4.62	573.38	5.26	842.45	5.29	1058.38	5.27	949.25	5.83
42	The Celts	71	2	0.03	1	0	0	1	0	972.28	5.32	271.66	3.77	600.79	4.41	935.21	4.82	997.57	5.38	1063.17	5.41
43	Spain	50	0	0	0	0	0	0	0	910.1	5.2	241.2	5.4	540.02	5.72	785.94	5.59	917.58	5.58	947.29	6.1

Spoiler Policy Branch Stats :

Spoiler Technology Research Times :

L. Vern · Jun 26, 2023

Spoiler Emperor Policy Branch Adoption Flows :

L. Vern · Jun 26, 2023

Spoiler Warlord Policy Branch Adoption Flows :

azum4roll · Jun 27, 2023

Definitely play without Time Victory next time. You can still get the "winner" on turn 500 from the data, but it won't stop the "actual" victory from being generated.

CppMaster · Jun 27, 2023

azum4roll said:
Definitely play without Time Victory next time. You can still get the "winner" on turn 500 from the data, but it won't stop the "actual" victory from being generated.

Yet the "actual" victory doesn't matter if it happens after the time victory. In the time victory, the greatest empire in general wins (Although the formula for score may not be perfect).

Stalker0 · Jun 27, 2023

Ultimately I can’t accept results where SV is nonexistent, that just throws everything out of whack. I look at my own recent game as India, I won SV on turn 454, and I was in the middle of the pack score wise, but I still won. Hey in an AI setting that win would have gone to someone else.

I think you have to remove time victory to see the “real results”.

der_weihnachtsmann · Jun 27, 2023

Although it speaks for itself that the AI half of the time doesnt manage to win before the time runs out which in itself sounds kinda bad.

azum4roll · Jun 27, 2023

CppMaster said:
Yet the "actual" victory doesn't matter if it happens after the time victory. In the time victory, the greatest empire in general wins (Although the formula for score may not be perfect).

Well IMO Time Victory shouldn't be a thing at all. The world doesn't end in year 2050 and decide a winner based on arbitrary scores. Now that the game doesn't start breaking on turn 501 anymore, we don't need a time limit.

Also don't play on Continents+, you'll just see Siam at the rock bottom.

N.Core · Jun 27, 2023

Maybe still count Time Victory but disable Time Victory so then the game won't end.
Account civs that are reached the highest score at turn 500, but let them still gain other VC types.

Stalker0 · Jun 27, 2023

So I thought about this more, and more I'm convinced that warlord is just a bad difficulty to do this on.

Now it does seem fairest at first glance, but what the results to me show is.... warlord is not a competitive difficulty. The AI isn't winning as much as it just plays until the time runs out. and that is perfectly fine for a player that wants a more casual experience, nothing wrong at all. But from a balance standpoint....it doesn't matter. It frankly doesn't matter how balanced or imbalanced the civs are on warlord, because warlord isn't a competitive difficulty. Its there to kick back and have fun, you don't need really good balance at that point.

Likewise if we remove time victory, sure we will see other VCs come out. but at the same time competitive players aren't playing games till turn 550 or 600 on standard speeds....so that assessment is somewhat flawed as well.

Perhaps its better to try King, the AIs get a few bonuses but not as many, and if they are actually winning games in the 500 turn window, than that maybe the most accurate representation we can hope to get.

ma_kuh · Jun 27, 2023

Warlord results can still be helpful in showing what the "AI gap" looks like though, compared to a human. They can't win SV in a timely fashion for example, so however humans are managing to do it isn't how the AI is playing. Maybe something to look at in terms of helping cleanup scoring on the AI logic side.

At the very least it shows that the AI bonuses are important to the AI's ability to win given the current constraints (research time being one that has been tweaked semi-recently), and so mechanics that interact favorably with AI bonuses (Japan's on-level-up ticks being a prime example right now) are going to have outsized impact for higher difficulties.

L. Vern · Jun 27, 2023

Stalker0 said:
So I thought about this more, and more I'm convinced that warlord is just a bad difficulty to do this on.

Now it does seem fairest at first glance, but what the results to me show is.... warlord is not a competitive difficulty. The AI isn't winning as much as it just plays until the time runs out. and that is perfectly fine for a player that wants a more casual experience, nothing wrong at all. But from a balance standpoint....it doesn't matter. It frankly doesn't matter how balanced or imbalanced the civs are on warlord, because warlord isn't a competitive difficulty. Its there to kick back and have fun, you don't need really good balance at that point.

Likewise if we remove time victory, sure we will see other VCs come out. but at the same time competitive players aren't playing games till turn 550 or 600 on standard speeds....so that assessment is somewhat flawed as well.

Perhaps its better to try King, the AIs get a few bonuses but not as many, and if they are actually winning games in the 500 turn window, than that maybe the most accurate representation we can hope to get.

I agree with this assessment. I ran a ton of warlords games as it was by and large the most requested scenario that people wanted to see, and what we've seen is at this difficulty is civs don't really have the oomph to score consistent wins. Which as you've said is absolutely great if you're new or casual and don't want the AI trying too hard to win, but probably not so great for gauging performance in situations where players want a challenge.

Probably still some useful information to be gleaned from this though, like "Nerf Austria"

CppMaster · Jun 27, 2023

azum4roll said:
The world doesn't end in year 2050 and decide a winner based on arbitrary scores.

What kind of argument is that?

The world also doesn't end when a country launch a rocket or is very popular. It's just a game and those are just winning conditions.
Time victory is similar to science victory. It happens when you prevent all other victories from competitors and you have the greatest empire. This is similar to a victory in Humankind or Dune. It feels much more natural than those other arbitrary victory conditions that you have to accumulate enough tourism, launch a rocket or have enough votes... With the time victory, whoever has the greatest empire wins.

der_weihnachtsmann · Jun 27, 2023

Also if we balance around Emperor where the performance is skewed a lot by handicaps, we dont generate any meaningful data on how a civ would perform handled by the player, who obviously doesnt have handicaps above chieftain.

Edit: And balancing around AI performance with difficulty handicaps also means the balance is strictly dependent on them which also means that the balance is very different with another difficulty with different handicaps.

The ideal balance situation for me would be that AI is balanced around Warlord and difficulty handicaps minimally impact the distribution.

CppMaster · Jun 27, 2023

Stalker0 said:
It frankly doesn't matter how balanced or imbalanced the civs are on warlord, because warlord isn't a competitive difficulty.

L. Vern said:
Which as you've said is absolutely great if you're new or casual and don't want the AI trying too hard to win, but probably not so great for gauging performance in situations where players want a challenge.

That's also a good point, but with that in mind, the balance would be the most important on deity difficulty, because the most competitive players play on it. On the other hand, deity is the most different difficulty from other difficulties and balance around it could break earlier difficulties.

pineappledan · Jun 27, 2023

I was excited to see what the AI's, independent of the handicap bonuses, could show us, but ultimately I think this leaves me a bit cold. I don't think the Warlord tests tell us much other than that the AI needs the handicaps in order to win.
At warlord, the AI just has to give a human player a run for their money, but the human is still heavily favored to win. These win %s reflect that. It's not a competitive game mode.
This also brings me back to my older comment: The AI doesn't need to win, it just needs to knock a human out. At Warlord, I think we are seeing the lower end of that reality.
Other observations:

With time victory gobbling up most of the % wins, all other wins recede in roughly the same order as their speed. The average victory speed for DomV, DiploV and CV all moved back roughly 80 turns. The average SV on emperor was already within 80 turns of 500, so it is just gone. The fastest victory types (domV and CV) are much less affected than the slower victory types (DiploV and SV). I interpret this as the handicaps affecting all victory types fairly evenly. If that weren't the case we would see victory times move a lot further for some victories than others.
The exception to this victory drought is the absolute top tier civs. There are very few time victories in the top 3. These are all civs that can win an actual victory in the time given.
Nerf Austria.
As @L. Vern pointed out, military civs do a lot better in general without the AI bonuses. This makes sense because conquest/wide gives a lot of score, and if no one is winning any other way, the person that simply has the most conquest will have the most score most of the time. We aren't seeing military civs do better, we're mostly just seeing how the game biases score, which doesn't have much bearing on who is actually closest to winning.
- One exception is domination/Science civs. Assyria and Russia both got their SV's deleted by Warlord. Russia was already at the bottom though
- Another exception is Domination civs that augment XP and levels. Once again Assyria dropped a lot. So did Zulu, and so did Japan. All civs are getting free XP and more % XP from combat, and it looks like taking that handicap away hit those 3 civs hard. Hard enough for them all to go down while the rest of the Domination civs went up.
- Mongols, Huns, Russia, and Spain all stayed at the bottom on both difficulties. The only conclusion is that they're just bad, even when the settings favor them. Their badness is a strong signal that cuts through the noise.

CppMaster · Jun 27, 2023

pineappledan said:
The average victory speed for DomV, DiploV and CV all moved back roughly 80 turns.

Hmm, yeah, maybe the time victory turn limit should be dependent on chosen difficulty to accommodate AI bonuses. Like maybe 600 turns for Warlord?

der_weihnachtsmann · Jun 27, 2023

der_weihnachtsmann said:
balancing around AI performance with difficulty handicaps also means the balance is strictly dependent on them which also means that the balance is very different with another difficulty with different handicaps.

An argument can be made that balancing around AI performance is more important than around players, because you handicap yourself through difficulty anyway and runaway AIs hurt your experience much more that way.

N.Core · Jun 27, 2023

der_weihnachtsmann said:
Also if we balance around Emperor where the performance is skewed a lot by handicaps, we dont generate any meaningful data on how a civ would perform handled by the player, who obviously doesnt have handicaps above chieftain.

Edit: And balancing around AI performance with difficulty handicaps also means the balance is strictly dependent on them which also means that the balance is very different with another difficulty with different handicaps.

The ideal balance situation for me would be that AI is balanced around Warlord and difficulty handicaps minimally impact the distribution.

The ideal balance is that AI is balanced around having no bonuses at all aside from their own kit. Let them play with their raw kit properly, and then we can adjust the bonus later to balance the difficulty handicaps.

If we balance around Emperor, then the game will only be balanced around Emperor. There will be lopsided results on other difficulties.
We have to balance while accounting for all existing difficulties.

azum4roll · Jun 27, 2023

I don't see how you can balance around Warlord besides lowering all costs/inflating yields more so they can win before turn 500.

3.6 AI Only Game Statistics 700 game update (350 Emperor / 350 Warlord) + vote for next config

Settings for 3.7 games

D4 - Communitas

D4 - Communitas - 4UC

D4 - Continents+

D4 - Continents+ 4UC

D5 - Communitas

D5 - Communitas - 4UC

D5 - Continents+

D5 - Continents+ 4UC

D6 - Communitas

D6 - Communitas - 4UC

D6 - Continents+

D6 - Continents+ 4UC

Warlord

Warlord

Warlord

Lost the game

Deity

Baller Magnus

Chieftain

Lost the game

1st Class of Neoa

Baller Magnus

King

Warlord

Deity

Chieftain

Deity

Deity

Deity

Chieftain

1st Class of Neoa

Lost the game

Similar threads