Help with Yield Discount Rate (Math)

[Stalker0]

So adding in a minimum forced compensator, here's some additional charts. Notice how crazy fast the yields have to climb in the last bit of turns. I also just showed Turn 1- 350, which is a smoother curve.

Spoiler :

 

[Gazebo]

Well it probably just triggers whenever an instant yield activates, so it won't have that issue.
I fixed it in my code with a kinda jank workaround, it seems Gamevent.PlayerDoneTurn doesn't work?
@Gazebo just wondering how you would enable it, is the code on the github yet?

I’ll push the code tonight. But it works correctly.
G

 

[CrazyG]

Trying that formula out, I get.

1 Science (Turn 1) = 2.1 (Turn 50) = 6.9 (Turn 100) = 20 (Turn 150) = 52 (Turn 200) = 120 (Turn 250) = 249 (Turn 300) = 458 (Turn 350)


Another note, this formula fails the smell test around Turn 323. At that point, the discount is still not high enough. Effectively what this is saying is that civ yields at this point do not grow fast enough to compensate for the ever shrinking rate of return before the end game....which throws the discount off. In other words.....the 458 science on Turn 350 is 100% objectively worse than any of the numbers previous, there is no subjective feel required.... if someone offered you 249 onTurn 300 or 458 science on Turn 350 you should always take the first deal.

I think it's worth just accepting that the values get weird at the very end (the very beginning too).

Otherwise this equation works fine, but it doesn't mean that the yields are even to each other.

 

[CrazyG]

Trying that formula out, I get.

1 Science (Turn 1) = 2.1 (Turn 50) = 6.9 (Turn 100) = 20 (Turn 150) = 52 (Turn 200) = 120 (Turn 250) = 249 (Turn 300) = 458 (Turn 350)

Okay so here's a practical example.

I'm choosing a follower belief on turn 100. Two new beliefs exist, A and B.
A gives me 7 forever, starting now.
B gives 20 starting on turn 150.

I think I'd usually take B. However if you change to , I think I'd take A. So in conclusion these values are pretty reasonable, I'd consider most options, except the turn 300+ stuff.

 

[Stalker0]

I think I'd usually take B. However if you change to , I think I'd take A. So in conclusion these values are pretty reasonable, I'd consider most options, except the turn 300+ stuff.

Also keep in mind when we were are done this will NOT be the curve for culture. Each yields needs its own discount curve. I will graph the culture data soonish and see what it shows.

 

[azum4roll]

Anyone took the science corporation? It should affect yields a lot.

 

[InkAxis]

So with instant yield logging, are there any updates on this?

 

[Stalker0]

So with instant yield logging, are there any updates on this?

I'm actually going to run a few AI only games today with the new logging

 

[Stalker0]

So I could use some opinions. I thought it would make sense to use the same AIs in repeated games, effectively our "model AIs", ones that we think give a good representation of the various types of playstyles, while also not having bonuses that can really skew results. So here is the list I'm going to run at the moment, but would love some thoughts if there are better candidates here.

Arabia - The stereotypical Tall play
Byzantium - Strong Religious Play, with Medieval Warring
Carthage - Sea/Wide/Money
Germany - Diplo Focused/Trade/Late game bloomer (I pondered Austria or Siam but I think they are both more swingy than Germany depending on their starts).
Mongolia - Early game Domination Type
Morocco - Wide/Culture/Trade Focused
Russia - Wide/Science/Good late game war
Zulus - Gunpowder Domination Type

 

[Stalker0]

hmm, so the first new AI run has thrown me a wrinkle. Carthage pulled off a domination victory. So previously I was only recorded the yields of civs that were solid throughout the game....as there was little point in basing our numbers on "bad performing civs" as we would expect the human to be at least be competitive in the game.

But in this scenario, only carthage made it through, all other civs were crushed beneath her boot. So should I just preserve her numbers and continue on with other scenarios? Hmm....not sure on that one.

 

[CrazyG]

hmm, so the first new AI run has thrown me a wrinkle. Carthage pulled off a domination victory. So previously I was only recorded the yields of civs that were solid throughout the game....as there was little point in basing our numbers on "bad performing civs" as we would expect the human to be at least be competitive in the game.

But in this scenario, only carthage made it through, all other civs were crushed beneath her boot. So should I just preserve her numbers and continue on with other scenarios? Hmm....not sure on that one.

I'd ask an applicable question, like 'How important is Carthage's late game science?'

I imagine that once she has 7 capitals, everything is pretty unimportant, even less important than your discount rate would suggest. So I'd probably drop the late game data, the first 100 or 200 turns might be useful though.

 

[Stalker0]

Unfortunately these firetuner runs have not been going well. I'm getting a lot of soft CTDs, and frankly the firetuner runs are MUCH slower than my previous runs... though obviously with the benefit of not requiring any interaction from my part.

If I could just get the crashes to stop I could run them at night while I'm asleep, but so far its not been working well.

 

[Gazebo]

Unfortunately these firetuner runs have not been going well. I'm getting a lot of soft CTDs, and frankly the firetuner runs are MUCH slower than my previous runs... though obviously with the benefit of not requiring any interaction from my part.

If I could just get the crashes to stop I could run them at night while I'm asleep, but so far its not been working well.

Mine never ever crashes unless there's a clear bug. Try running it in strategic mode?

 

[Stalker0]

Mine never ever crashes unless there's a clear bug. Try running it in strategic mode?

I was I might try running it overnight with nothing else running just to see if it runs any better.

 

[azum4roll]

It's fine on my end too, but I run it in several sessions (restarting the game in-between) while I'm awake, so there shouldn't be as much memory leak. Should be fine even after turn 500, assuming that bug has been fixed.

I've never seen an AI pulling off Domination, by the way. AI refusing to make peace is probably helping that?