Mixing GPPs and loving it

[oyzar]

MestreLion where did you get this info? cause i learned the same as percy through my games... Say build the great wall, then start running a scientist or a preist and it is very easy to see that it works like that...

 

[Percy]

Hmmm, i've ran a couple of tests in WB, and i must admit i'm confused. It doesn't seem to work quite exactly like i said, and it doesn't exactly like MestreLion said. I looked at the article from the War Academy, and tested its last paragraph, and it doesn't work that way either (for those who read it, it says i should get 67/33 instead of 50/50, but i do get 50/50...).

Can someone look through the XML and check to see if they changed something with BtS ?

 

[KMadCandy]

i never knew the power of golden ages pre-BtS. partly because i've done like 3% space victory games. i used to totally ignore taj mahal. GAs i'd use for culture bombs to end city revolts and get tiles to work if i had a war in my game.

but with MoM, man those things rock! and in my first BtS game i got a quest where one reward option was that if i had the Statue of Zeus (which i did) i could choose a golden age, so i did. so now i get an artist, okie dokie, he'll take a nap for a golden age. i'm getting addicted to golden ages...

 

[Percy]

Okay, the results of my further testings is that it seems it is indeed the POINTS that count, now, and not the sources.

Among the tests i ran, here are those that i find the most interesting:
I ran 10 turns with only Stonehenge (+2 GProphet points per turn) and had 100% chance to generate a GP at the end of the 10 turns, obviously. Then, i added 2 Scientists for another 10 turns, and had this probability at the end: 40% GP and 60% GS.
If sources counted as i said, it would have been 50/50 (20 turns of 1 GP source, and 10 turns of 2 GS sources). If it counted as MestreLion said, it would have been 66/33.
But it was 40/60, and that corresponds to 40 GProphet points (20 x 2) to 60 GScientist points (10 x 6).

I restarted and popped Stonehenge, the Oracle, and Angkor Wat in a city, along a Scientist, for one turn. If sources counted, it would have been 75% GP and 25% GS. But it was 66/33, which corresponds to 6 GP points (2 + 2 + 2) to 3 GS points (the specialist).

Since it contradicts an article that is in the War Academy section of this website, i assume it has been changed in a patch at some point.

Please feel free to correct me if i made obvious mistakes or missed something. Just do it nicely =P

 

[VoiceOfUnreason]

Hmm, it looks like the game mechanic changed....

In Vanilla, and also in Warlords, it was common for people to get confused by the GP probability system. The guilty code was this fragment, from CvCity.processSpecialist

changeGreatPeopleUnitRate(eGreatPeopleUnit, iChange);
​

Where iChange was the number of specialists (and there was similar code for buildings)

In BtS, the same calls to changeGreatPeopleUnitRate now look like

changeGreatPeopleUnitRate(eGreatPeopleUnit, GC.getSpecialistInfo(eSpecialist).getGreatPeopleRateChange() * iChange);
​

In other words, the probabilities are now point based, rather than turn based.

So in Percy's scenario (10 turns of Oracle, 10 turns of Oracle + 2x Scientist)

Warlords Answer:
20 turn-sources of prophet points, 20 turn-sources of Scientist points. Therefore 50% chance of getting a scientist.

Beyond the Sword:
40 Prophet points, 60 Scientist points, therefore 60% chance of getting a scientist.

This checks with the world builder experiment that I just ran in BtS.

The idea that the rates on individual turns are somehow averaged together is complete bunk. I think it comes from Kylearan's original article - but he was experimenting without the benefit of the SDK, and got it wrong.

It's still the case that your GP points lose their flavor on the bedpost over night. That is, the city erases all memory of where the overflow originated when it spawns a GP. So the outcome probability is driven entirely by how your points were distributed in this GP cycle.

 

[Percy]

Thanks for the confirmation, VoU.

 

[popejubal]

1. Don't trade if you can help it, unless it's some crappy tech and you are about to lose monopoly power over it anyway. Else you accelerate the pace of teching for EVERYBODY. Snaaty pointed out rightly that Deity level means fast AI teching no matter what and that it's more beneficial to try to slow down the AI than try to out-tech it. Instead, infiltrate and extort.

...unless you can trade one tech with a few neighbors at the same time.

If I trade Tech X with Liz, Catherine, Gandhi, Mansa, Freddie and I sell the Tech for a couple hundred gold each to two other civs, I've accelarated myself quite a bit. Sure, everyone else is accelerated a little too, but it's well worth it if I can get enough of a comparative advantage.

I now have an extra 5 techs and several hundred gold extra to either upgrade units or to finance deficit research. Everyone else just has one extra tech.

Obviously, that's not going to be possible in most cases. At higher levels, though, if you can get a relatively expensive tech that the AI doesn't go for quickly, you can often trade with 3 or even 4 civs at the same time. I wouldn't give away Horseback Riding to an immediate neighbor who has both Elephants and Construction, but you can get some very good deals with the computer as a whole if you trade with several AI civs simultaneously. It seems to me that this is why Lightbulbing was so successful in Warlords.

It's not that a Lightbulb gave you 1 free tech. It's that a lightbulb gave everyone a free tech while it gave you 4-6 free techs and some extra cash.


Sometimes it's even worthwhile to trade a tech at a deep discount to one of your neighbors if it is to your long term strategic advantage. I'll often trade Code of Laws at a discount to my immediate targ^H^H^H^Hneighbors. I want them to build Courthouses in their cities. It makes those cities just that much more attractive for when I conquor them later. I'll also consider gifting Bronze or Iron Working to a neighbor who has that resource in his lands but doesn't know it yet if they are currently at war with another neighbor when I want that 3rd civ taken down a notch. (if I can't gift the metal resource directly)

 

[axident]

Thanks Percy and VoU (VoU your explanation of leaders' propensity for war and backstabbbing was very helpful!)... and sorry if I ruffled anybody's feathers but I was ticked off by this latest theory-without-an-XML-explanation. This makes it 3 out of 3 times this week where someone called me out on a game mechanic, insisted that it was some other way, but was wrong in the end. I can't look at XML worth a crap, but I have a pretty good feel for the game as a player and GrL seemed to help more for GS generation than some were giving it credit for.

 

[axident]

...unless you can trade one tech with a few neighbors at the same time.

If I trade Tech X with Liz, Catherine, Gandhi, Mansa, Freddie and I sell the Tech for a couple hundred gold each to two other civs, I've accelarated myself quite a bit. Sure, everyone else is accelerated a little too, but it's well worth it if I can get enough of a comparative advantage.

I now have an extra 5 techs and several hundred gold extra to either upgrade units or to finance deficit research. Everyone else just has one extra tech.

Obviously, that's not going to be possible in most cases. At higher levels, though, if you can get a relatively expensive tech that the AI doesn't go for quickly, you can often trade with 3 or even 4 civs at the same time. I wouldn't give away Horseback Riding to an immediate neighbor who has both Elephants and Construction, but you can get some very good deals with the computer as a whole if you trade with several AI civs simultaneously. It seems to me that this is why Lightbulbing was so successful in Warlords.

It's not that a Lightbulb gave you 1 free tech. It's that a lightbulb gave everyone a free tech while it gave you 4-6 free techs and some extra cash.


Sometimes it's even worthwhile to trade a tech at a deep discount to one of your neighbors if it is to your long term strategic advantage. I'll often trade Code of Laws at a discount to my immediate targ^H^H^H^Hneighbors. I want them to build Courthouses in their cities. It makes those cities just that much more attractive for when I conquor them later. I'll also consider gifting Bronze or Iron Working to a neighbor who has that resource in his lands but doesn't know it yet if they are currently at war with another neighbor when I want that 3rd civ taken down a notch. (if I can't gift the metal resource directly)

Oh I agree, but it doesn't happen as often in BtS as AIs seem to have more variation in their priorities and tech rates. Maybe it's just me, but it seems to me that it was more common in WL to be able to pull off 3 or 4-for-1 trades or better. Now I'm stuck with 2-for-1s many times, which I don't think is worth it if it means that those 2 will use that tech to trade with everyone else for their own backfill. AIs don't always have lots of gold lying around, either. I also play Standard sized maps with 7 civs, Fractal or Big and Small the most often these days, so it's not like in Pangaea-style maps where you increase the odds of pulling off a multiple trade. So if a very nice trading opportunity like the one you described opens up, sure, but else I'm a total scrooge and mainly self-research, extort, and steal tech.

As for trading at a discount.. eh. Things change, what happens if you you get DoW by someone else and now you just gave a bunch of other civs CoL for almost nothing? I'll sell techs for big money sometimes if it's a non-monopoly tech, but that's about it. And sometimes I will trade some nice stuff to someone just to get them to DoW on our mutual nemesis.

 

[MestreLion]

MestreLion where did you get this info? cause i learned the same as percy through my games...

And heres my answer:

The idea that the rates on individual turns are somehow averaged together is complete bunk. I think it comes from Kylearan's original article - but he was experimenting without the benefit of the SDK, and got it wrong.

Ive read this article an year ago, and stopped playing Civ in October (shame, yes, i know). So i wasnt aware that many new things were found about it in the SDK.

Thanks a lot VoU for sheding light in this subject. I guess both concepts (turn average and source points) are now obsolete. So now the common sense is the one thats correct? Odds are calculated based on total points generated, period?

 

[VoiceOfUnreason]

Thanks a lot VoU for sheding light in this subject. I guess both concepts (turn average and source points) are now obsolete. So now the common sense is the one thats correct? Odds are calculated based on total points generated, period?

Yes, for Beyond the Sword. Warlords and Vanilla still use the original source based formula.

 

[MestreLion]

Yes, for Beyond the Sword. Warlords and Vanilla still use the original source based formula.

What about the "turn normalization" concept on Vanilla ? 3 turns with 1 source are worth more or the same as 1 turn with 3 sources ?

 

[VoiceOfUnreason]

What about the "turn normalization" concept on Vanilla ? 3 turns with 1 source are worth more or the same as 1 turn with 3 sources ?

As far as I know, that hasn't changed since I posted last week

In Vanilla (and Warlords), three turns with 1 source has precisely the same weight as one turn with 3 sources.