Login

Sullla · April 28th, 2018, 14:18

I did finish the Alternate Histories #2 report that I mentioned working on a few weeks ago. There's some interesting data here about what happened when I played out the same map repeatedly with a collection of terrible AI leaders.

haphazard1 · April 28th, 2018, 17:42

Thanks for posting this, Sullla! Very interesting results, seeing the same trends repeat in most of the games.

It is remarkable how poorly a leader like Willem performs, despite having very strong traits. The importance of the less obvious AI leadeer characteristics, like military unit emphasis and research focus, becomes apparent in these matches.

I do wonder about Hannibal's performance on this map, and how much impact his starting position and lack of room to expand had compared to his traits. That his results were the most variable adds to my curiosity about how the same set of leaders would perform on a different map.

Fluffball

(April 28th, 2018, 17:42)haphazard1 Wrote: Thanks for posting this, Sullla! Very interesting results, seeing the same trends repeat in most of the games.

It is remarkable how poorly a leader like Willem performs, despite having very strong traits. The importance of the less obvious AI leadeer characteristics, like military unit emphasis and research focus, becomes apparent in these matches.

I do wonder about Hannibal's performance on this map, and how much impact his starting position and lack of room to expand had compared to his traits. That his results were the most variable adds to my curiosity about how the same set of leaders would perform on a different map.

IMO the map plays a way bigger role in successes than the leader (edit: leader's preferences, not counting peace weight which is kind of "part of the map"). There are some freaks like Mansa* that can be the tech leader with 4 cities, but give Sitting Bull 25 cities and he'll win the game. Put a super cramped Monty next to SB who has tons of room to expand and SB probably wins that. (Of course no one would be surprised if he somehow managed to lose. rolleye

) Without re-looking at the maps, I'm guessing that's the reason idiots like De Gaulle did well. And also why the "average" civs that are just sort of solid tend to have their moments of glory. Stalin, Kublai, Qin, and many other forgettable leaders... all pretty well rounded and just waiting for their chance. Even De Gaulle has an obvious path to victory with his personality.

The map variability and people's skill at reading it adds most of the excitement to the streams, so it wouldn't make sense to do a perfect circle Finals map for every game, but that would certainly help things like Hannibal getting screwed by the map. And even then you get the peaceweight screwing over certain AIs. Fredrick can easily win a culture victory if he's sitting behind a friendly and helpful Kublai. Mansa had basically no chance in the finals.

*Minor note, He should really be referred to as Musa, not Mansa. Mansa was his title. Same with how Sullla always refers to Ghengis as Temuijin I suppose.

haphazard1 · April 29th, 2018, 11:01

The map is certainly a big factor. It would be very interesting (but a lot of work) to take a game such as the one Sullla just ran the alternate histories for, and shuffle the start using World Builder to exchange the positions of the AIs. Put Shaka in Hannibal's position, Toku in Isabella's, etc., and then run the game. How much would change?

Some things would still be similar -- Frederick having big peace weight differences from most of the field, and Isabella founding a religion and hating everyone who does not adopt it, for example. But which AIs are next to the warmongers, and which have more space, would change some of the trends.

It is not really practical to control for all possible variables -- there are just too many of them in Civ.

Hmmmm, wild idea. Map with half a dozen instances of the same leader, and see what happens? Is that even possible to set up?

Lord Parkin

Thanks for doing the extra report Sullla, very interesting stuff.

Definitely good evidence for the case that - while there can be occasional outlier results - over multiple games the same "weak" civs will always tend to get weeded out of the competition, while the same "strong" civs will tend to remain. Of course, sometimes the brackets happen by chance such that there's an extra-lightweight game (weak leaders breeze through on easy mode) or an extra-tough one (inevitably dooming a few otherwise tough contenders to early elimination). And occasionally the layout of the map might favor or hinder certain civs significantly more than average - though reading the map is a large part of the fun, so I wouldn't suggest making everything a donut.

Are you planning on running Season 4 this year? Pretty keen to see how the predictions improve (or not!) next time around based on the existing data.

haphazard1 Wrote:The map is certainly a big factor. It would be very interesting (but a lot of work) to take a game such as the one Sullla just ran the alternate histories for, and shuffle the start using World Builder to exchange the positions of the AIs. Put Shaka in Hannibal's position, Toku in Isabella's, etc., and then run the game. How much would change?

Permuting the starts in some of the test games would certainly be interesting to see. I suspect victory probability is somewhat correlated to starting location. Enough perhaps to reliably switch the victor between two similarly strong contenders over a statistically significant sample of games. Probably not enough though to switch the victor between a strong contender and a weak leader. Unless we're talking about an exceptionally poor start.

haphazard1 Wrote:Hmmmm, wild idea. Map with half a dozen instances of the same leader, and see what happens? Is that even possible to set up?

Sure it'd be easy enough to set up. Might be tricky to deal with the notifications though!

"Gandhi has declared war against Gandhi - Gandhi has joined the war against Gandhi - Gandhi has made peace with Gandhi" eek

I'm guessing you'd want to rename them. :P

yuris125 · May 2nd, 2018, 04:26

Leader names can be edited in the Worldbuilder file, that's not a problem

Frankly, the only annoying part would be pressing Enter a billion times, especially if you want to run multiple simulations. If there's a way to make the game run without asking the user to press Enter every turn, I couldn't find it, but it would make survivor style games so much easier.....

rho21 · May 2nd, 2018, 04:42

(May 2nd, 2018, 03:21)Lord Parkin Wrote: "Gandhi has declared war against Gandhi - Gandhi has joined the war against Gandhi - Gandhi has made peace with Gandhi"

You missed "Gandhi has nuked Gandhi". smile

LKendter · May 3rd, 2018, 11:16

(May 2nd, 2018, 04:42)rho21 Wrote:
(May 2nd, 2018, 03:21)Lord Parkin Wrote: "Gandhi has declared war against Gandhi - Gandhi has joined the war against Gandhi - Gandhi has made peace with Gandhi"

You missed "Gandhi has nuked Gandhi".

What about Gandhi voted for Gandhi in AP elections?

haphazard1

(May 3rd, 2018, 11:16)LKendter Wrote: What about Gandhi voted for Gandhi in AP elections?

Gandhi voted for Gandhi, defeating Gandhi, while Gandhi abstained. lol

It would get confusing.

ArtDeco · May 5th, 2018, 07:48

Hey all,

I also tested some alternate plays, care if I share here?

Season3 Game 5

Season3 Game 6

Basically same things happen over and over but with small surprises that change the results. Julius Caesar can't stand Brennus and pretty much he made everything for Elizabeth to win unlike what we had in survivor, fortunately. We should take a look how fair the starting positions are, aparently there are very big differences, I blame tropical climate. Speaking of which tundra and rocky are left out right now, maybe Sulla makes tundra for season 4 this year parallel to Russia world cup? :crazyeye:

I don't think Willem is weak, and Tokugawa is just lucky in that starting place. To alternate even more we should swap starting positions but I rather see more and more seasons with different maps to ever come to a conclusion.

Navigation

Extra Menu

About us