[BTS] AI Autoplay Ranking- testing with fair maps

[Keler]

ROUND 46 (no show:Elizabeth,Gandhi,Mao Zedong,Saladin)
agg ai

non agg ai

 

[Keler]

ROUND 47 (no show: Catherine,Isabella,Shaka,Wang Kon)
agg ai

non agg ai

 

[Keler]

ROUND 48 (no show: Alexander,Hammurabi,Qin Shi Huang,Roosevelt)
agg ai

non agg ai

 

[Keler]

ROUND 49 (no show:Augustus Caesar,Montezuma,Suleiman,Zara Yaqob)
agg ai

non agg ai

 

[Keler]

ROUND 50 (no show: Gilgamesh,Louis XIV,Ramesses,Washington)
agg ai

non agg ai

 

[Keler]

ROUND 51 (no show: Asoka,Brennus,Pericles,Victoria)
agg ai

non agg ai

 

[Keler]

ROUND 52 (no show: Boudica,Charlemagne,Pacal,Sitting Bull)
agg ai

non agg ai

 

[Keler]

SUMMARY 1 for Round(1-52)
Result for this settings are:
Noble difficulty
Option=GAMEOPTION_AGGRESSIVE_AI
Option=GAMEOPTION_NO_TECH_TRADING
Option=GAMEOPTION_NO_TECH_BROKERING
Option=GAMEOPTION_NO_VASSAL_STATES
Option=GAMEOPTION_NO_EVENTS

Hello, as I am done with 52 rounds I think it is time to make a summary for now. This is the table. It is in the Summary page of excel file names civ4 commodoremap.

There has ben 52x8=416 Games so far where every leader appeared in 48 games.
Out of 416, there are 120 Culture, 114 Domination, 101 Space, 42 Diplomatic, 39 Time victories. While only 1/101 space game ended before turn 400, 50/120 culture games enden before turn 400. In the excel file, you can find more about this stuff in Count Victories page.

Average ranking point is above 78, and average kill per leader is above 22. And every leader should have won 8 games averagely. 52x8/52=8 .

Darius seemed to have the fastest tech rate ever, probably followed by Pacal according to my personal observation. Hatshepsut impressed me only when it comes to culture victories. Most of the early culture victories happen before they tech replacable parts and physics for some reason...

I don't know what matters for you personally, Power ranking or Victories only.
It is obvious that the sample is a bit small right now to come to final conclusion. A leader all of a sudden wins 3 games streak and jumps 15 leaders above in the power ranking.


And it is very funny that I randomly make these games by excel rand function that generates numbers between 0 to 1 and sort them fro big to small and put them 6 by 6 into groups and still Round14Game8 and Round40Game8 were almost same games by coincidence. Maybe there are more coincidences. Funny

Alright hope you enjoyed this. Hopefully I will continue with round 53.

 

[Louis the XIV]

Wow! Great work! I gave up my AI Auto play ranking because of 3 reasons.
It's not actually fair to a lot of the civs. (especially European civs) in the Earth map.
Mali, Mongolia, or America always ended up winning.
I realized that modding is a much better time use instead of uploading videos on youtube.
But anyway I can't believe you did 52 rounds!

 

[Lennier]

SUMMARY 1 for Round(1-52)
And it is very funny that I randomly make these games by excel rand function that generates numbers between 0 to 1 and sort them fro big to small and put them 6 by 6 into groups and still Round14Game8 and Round40Game8 were almost same games by coincidence. Maybe there are more coincidences. Funny

I assume that you're making sure the Excel rand function isn't using the same seed every time you run it.

 

[Keler]

Wow! Great work! I gave up my AI Auto play ranking because of 3 reasons.
It's not actually fair to a lot of the civs. (especially European civs) in the Earth map.
Mali, Mongolia, or America always ended up winning.
I realized that modding is a much better time use instead of uploading videos on youtube.
But anyway I can't believe you did 52 rounds!

Thank you. I upload replays on youtube so get rid of them from my computer

I assume that you're making sure the Excel rand function isn't using the same seed every time you run it.

I don't understand what you mean by seed, I just copy leader names list and generate random numbers by using this function. It generates different every time?


ROUND 53 (no show: Frederick,Gandhi,Pacal,Shaka)
agressive AI

Non agressive AI

 

[Keler]

ROUND 54 (no show: Asoka,Catherine,Ragnar,Washington)
aggressive AI on

aggressive AI off

 

[Keler]

ROUND 55(no show: De Gaulle,Isabella,Mao Zedong,Wang Kon)
agg ai

non agg ai

ok I think I am going to rerun the past 3 rounds same games with aggressive AI option on, non Agressive AI may reduce stupid suicide wars but those suicide wars usually happen in a way AIs know that they would be declared war on anyway,mostly. Beside, this is just ridiculous. No aggressive AI just favors undeserving leaders, let me turn that culture slide %100 and hope no one attacks me, and nobody does. Especially when these AIs tech electricty industrialsm etc. they will almost never ever declare war on each other before sending that space ship. Which won't happen because someone with tiny empire with culture slide %100 on will win the game and even that they are annoyed they will do absolutely nothing about it before sending that spaceship. This is so bad overall. No domination or less elimination does not bother me, I am annoyed by the fact that this just puts cultural leaders who can tech rifling a great position to win their games almost every single time.
And I also tried tech trading on which makes things even worse, it just makes all AI same somehow, beside my autoplaying AI keeps trading techs and keeps up with other noble AIs with just one ice city, and the map is way too small to never meet until the United Nations... So I am just going to keep doing more rounds with absolutely same settings and who knows after round 130 I can give a try to Sulla's round map used in first 3 survivor championship games.
In conclusion, I am not happy with aggressive AI off ranking.

 

[Kjotleik]

ok I think I am going to rerun the past 3 rounds same games with aggressive AI option on... In conclusion, the last 3 rounds with no aggresive AI on results won't be taken into the ranking.

I understand what you mean. Mixing non-AGG and AGG games may make the tables a bit weird. Keeping the setting the same for the whole campaign makes sense.

However, at the same time I would be interested in how much of a difference this setting actually makes. If you have the time, it would peak my interest to see one table (52 games) with AGG on and one with non-AGG. The comparison should be an interesting read. And it would be educational if we could draw some conclusions from it. Based on more than just three games.

Hannibal, HC and Justinian seems to be getting ahead of the rest a bit (after 52 games). Ragnar is down on 31st (I'm Scandinavian, so I'm a bit biased on his behalf you know...) and the current CIV4 AI Survivor champion - Mehmed II - is starting to climb the table a bit. Monty is still last, but Boudica, Tokugawa and Lincoln better watch out. They are within reach of Monty now.

Good luck with the rest of the games.

Have you considered your YouTube playlist, by the way? Whenever I try to watch it, I always get presented with the latest videos as the first videos in the playlist. Shouldn't it be the other way around? Just a thought...

 

[Louis the XIV]

I hope Louis does better in the next couple of games. Could you tell me what type of victory did Louis win most of the time? I'm guessing cultural or space race but I'm not sure.

 

[Lennier]

[QUOTE="Keler, post: 15930664, member: 271092]I don't understand what you mean by seed, I just copy leader names list and generate random numbers by using this function. It generates different every time?[/QUOTE]Computer random number generators aren’t truly random. They generate a sequence of numbers. One inserts a “seed” number into the function and a number pops out (which is generally used for the seed for the next call of the random number generator function.) One of the Civ IV game options is “new seed upon reload.” That option when enabled will allow different things to happen in the game if you reload. Otherwise, only changing human player decisions will change the game.

I’m not sure of the exact mechanics of the Excel RAND function to know if that was the cause of your similar runs.

 

[Keler]

I understand what you mean. Mixing non-AGG and AGG games may make the tables a bit weird. Keeping the setting the same for the whole campaign makes sense.

However, at the same time I would be interested in how much of a difference this setting actually makes. If you have the time, it would peak my interest to see one table (52 games) with AGG on and one with non-AGG. The comparison should be an interesting read. And it would be educational if we could draw some conclusions from it. Based on more than just three games.
.

I decided to restart from Round1 to all the way here with Aggressive AI off for alternative ranking to see difference. So it will take a month or something for me to keep going with Round57. Wish I had kept those worldbuilder files anyway

I hope Louis does better in the next couple of games. Could you tell me what type of victory did Louis win most of the time? I'm guessing cultural or space race but I'm not sure.

see the table above,or you can download the excel file, louis won 10 games out of 48 and 7 of them were culture

One of the Civ IV game options is “new seed upon reload.” That option when enabled will allow different things to happen in the game if you reload. Otherwise, only changing human player decisions will change the game.

I’m not sure of the exact mechanics of the Excel RAND function to know if that was the cause of your similar runs.

Thnaks for the info but I can't track this, I use same alphabetcal order of leaders everytime,maybe using the function twice can help, sorting them from biggest to lowest again and again

ROUND 56 (no show: Augustus Caesar,Boudica,Kublai Khan,Qin Shi Huang)
agg ai

non agg ai

 

[Lennier]

Lennier said:
One of the Civ IV game options is “new seed upon reload.” That option when enabled will allow different things to happen in the game if you reload. Otherwise, only changing human player decisions will change the game.

I’m not sure of the exact mechanics of the Excel RAND function to know if that was the cause of your similar runs.
Thnaks for the info but I can't track this, I use same alphabetcal order of leaders everytime,maybe using the function twice can help, sorting them from biggest to lowest again and again

After digging through the Excel help file, I discovered that you can't set a seed for the Excel RAND function. The RAND function uses a Mersenne Twister algorithm, but I'm not sure how it seeds the function. I would suspect that it takes a seed from the fractions of a second in the time. If you really want control over the seed value, you'll have to go with the add-in Analysis Toolpack-VBA. I think we've gone far enough down this rabbit hole.

 

[Keler]

ROUND 57 (no show: Genghis Khan,Julius Caesar,Napoleon,Stalin)
agg ai

non agg ai

 

[Keler]

ROUND 58 (no show: Cyrus,Hammurabi,Roosevelt,Willem van Oranje)
agg ai

non agg ai