Quadruple Holy City

[johnny_rico]

So I'm playing a game with Joao II. It's going poorly but I'm the first to optics and having fun with carracks. I come across Victoria on another continent and notice York is the Holy City for Confucianism, Taoism, Christianity, and Islam. She's got a total of five cities on that continent with a handful of smaller cities on a chain of islands to her west.

What gives? How can one city become the holy city for four different religions. Again, the city is York, her second city. The other three on the continent are much smaller than York, as are the island cities.

Just haven't seen that before and I'm not aware of the algorithm that determines the holy city when founding a religion.

Settings on the game are monarch/standard/normal/medium and small if it matters.

 

[MrCynical]

It's very rare to see a quadruple holy city, particularly one with the four later religions. I can't remember the exact algorithm either, but it tends to favour new cities with low culture - and hence it'll tend to be biased against existing holy cities. It's also strongly biased against the capital.

The odds of getting a quad holy city with those four are extremely low. At least with the early religions there's a higher chance simply due to there being fewer cities.

 

[VoiceOfUnreason]

Some of the answer is here. I really need to double back and flesh out the rest of it.

Edit: culture has nothing to do with it

 

[johnny_rico]

thanks for the replies. So it's very rare period. I'll have to double check on the four religions but I believe I am accurate.

 

[DaveMcW]

The "holy city score" is:

+10 for any city
+1 per population
+1 to +10 random factor
... divided by (1 + the number of religions present)
... divided by 8 if it's the capital

If Vikki spread her religions around equally, none of the cities would have "low religion" to push them up in the scoring. The biggest factor would be population, which means York wins each time.

 

[Flamer123]

i managed to found three religion in the same holy city using exactly what DaveMcW said - 2 rules:
1) the number of religions in ur holy city should always be equal or smaller to all the other cities in ur empire (besides ur capital)
2) ur holy city should always have the highest pop right before u found a new religion - that means u use the whip a lot the turn before founding (again, besides ur capital)

if u have a low number of cities (like victoria in ur game) it's a lot easier. I have'nt seen the computer intentionally aim for that.

 

[Flamer123]

oh, by the way, i did some experiments a while back on that formula with "new seed on reload". Frankly, i noticed a different formula. The addition of a random number from +1 to +10 would mean a non-deterministic holy city. However, in my game, i specifically had 2 cities - pop 6 and pop 7, each with the same number of religions, and the pop 7 won 10 out of 10 reloads. i reloaded, whipped the pop 7 to pop 5, pressed end turn, and the holy city was founded on the pop 6... highly highly deterministic


that means that once u keep those 2 rules i've written earlier u'll deterministically get a holy city in the city u want.

 

[Percy]

It probably warrants extra research, but i'd like to point out that 10 tries is not enough to say it definitely is not random. Your example has me thinking that your conclusion is right, but still, randomness is something... well, random =)

 

[VoiceOfUnreason]

that means that once u keep those 2 rules i've written earlier u'll deterministically get a holy city in the city u want.

Given the number of times I've looked at the implementation, I'd happily bet a nickel that the results you observed would not be repeatable in another lab.

 

[Flamer123]

would you rather i try it 10,000 times and publish an article to Nature/PRL about it?

The probability of pop 7 winning over pop 6 is 59.5% [1- (1+...+8 + 9/2)/100], so the probability of it happening 10 out of 10 is 0.6%. i've read the cpp code and the addition of a random number is specifically written there, so i think there might be a problem with the implementation of "new seed on reload" on it.

go ahead and try it, a different lab also requires a different computer, tell me what u find.

 

[VoiceOfUnreason]

I know what I'll find. But let me offer you a better test. Use prophets to lightbulb all 7 religions, with missionaries on hand to convert the city that doesn't catch the religion. If all seven religions go to the same place, then I'll look into it.

(Like you, I consider your random seed suspect).

 

[kniteowl]

Funny I had the same thing happen to me in Vanilla Civ with Victoria except I only founded 3 fo the late religions but they all ended up in the same city, I had like 7 Cities. Although it kind of makes sense becuase that city had many desert tiles and only 1 food source and copper so it was generally relatively small throughout the game.

 

[Defiance]

You should be happy.. thats just begging to be taken

Start popping out those priests boy!

 

[weregamer]

(Like you, I consider your random seed suspect).

Since starting to play again with BtS, I've gotten the impression that "new seed on reload" may be broken. I've not done a lot of reloading - if I really can't stand something enough to reload for it, hitting Ctrl+W and fixing it is both faster and guaranteed to work. But the times I've tried it, I seemed to get the exact same results after the laborious reload time.

For that matter, I've noticed that goody hut results seem to be bunched for the same unit on the same turn. That is, if you move a scout 1 mp onto a hut and get, say "map", then go into worldbuilder and add a second hut right next to it, then hit that second hut with the other movement point, you seem to get "map" again well over 50% of the time. But if you wait until the next turn the results seem to have a more normal distribution. This is also purely anacdotal given only a dozen or so tries, but still interesting.

I presume Civ4 is using a normal linear-congruential PRNG, so this ought not to happen. Is it re-seeding the PRNG on some regular basis, like with the turn number or time of day or something?

 

[VoiceOfUnreason]

I presume Civ4 is using a normal linear-congruential PRNG, so this ought not to happen. Is it re-seeding the PRNG on some regular basis, like with the turn number or time of day or something?

It's the bog standard LCG - iterating through the numbers using unsigned longs, but masking away the low order bits when casting back to the short to use everywhere else.

 

[Flamer123]

I'm mistaken.

i just tried what u said, founding religions with great prophets (last time i did with tech btw), built 3 cities with worldbuilder (one of them capital), one city had pop 6 and the capital and another had pop 5.

though it started out 7-1 for the pop 6, and all the 7 were one after the other upon reload (at which time i suspected new seed reload), it changed later on and i stopped when i got to 12-7 for pop 6. 12/19 is around 60%, but definetly not deterministic, so i'm eating my hat now