Pioneer Expansion Profitable? - A Real Example

[Stalker0]

I'm lost here. Why would a puppet need to produce anything to break even?

NP let me explain. So from my previous work, building the colony effectively increased my overall science/culture by 3.16% (3.2 rounded). So my question is, how much would it take for a puppet to increase by that same amount. Now the puppet gets an advantage in that its existence doesn't increase your overall science needs (aka it has no science "cost"), so I wanted to know how much science it had to make, even with the "discount".

The results show that even though the puppet doesn't have any "costs", its output is so much less than the colony city that the colony still ultimately generates more science/culture yields.

 

[bonniepbilly]

I haven’t read all the comments, but based on the OP it seems that you are missing a kind of time discounting on the yields. One science at turn 350 is worth much less than one science at turn 350, because it delayed all your techs in between. It should make the outcome less profitable.

 

[Stalker0]

I haven’t read all the comments, but based on the OP it seems that you are missing a kind of time discounting on the yields. One science at turn 350 is worth much less than one science at turn 350, because it delayed all your techs in between. It should make the outcome less profitable.

Quite right, the biggest question with that is....what time discount should we use? I have debated this myself several times in other analysis but never come up with a really great answer. I have used a simple linear reduction of 1/400 (aka yields are worth maximum value on turn 1, and basically 0 on Turn 400), but I don't think that's a great model, though it does provide a good first approximation.

Adding a discount of 1-1/400*Turn Number, changes the %increase from 3.2 to 3.63%.

Effectively what we are seeing is that in the last stages of the game, the rest of the civ is starting to work more specialists and more GS are getting created. So the colony is getting "overshadowed" as time goes on, but that is blunted a bit by the time discount, so the colony is a bit more productive with that factored in.

 

[bonniepbilly]

How about taking it the other way, and calculating an internal rate of return? For that I guess you’d need to convert all yields first, that may be tricky as well, for example 1 science = 1 culture = 2 prod = 4 gold (purely illustrative!).
And then we can compare this IRR with what we get on producing/investing in buildings in established cities.

 

[Stalker0]

How about taking it the other way, and calculating an internal rate of return? For that I guess you’d need to convert all yields first, that may be tricky as well, for example 1 science = 1 culture = 2 prod = 4 gold (purely illustrative!).
And then we can compare this IRR with what we get on producing/investing in buildings in established cities.

So this example does assume 1 science = 1 culture. When I calculate the return, culture is negative for the colony, and so I directly subtract from the overall positive to get the final return.

Production wise: At the end of the day the colony does not "cost" any production. Any production it had it gained itself, it did not siphon that from elsewhere. I do convert production to science for the colony using the science process in the later turns, as that is when the production starts to be useful to the civ as a whole, as opposed to just helping the colony get where it needs to be.

Gold: Gold is an interesting question, and there are several opinions. Some people think gold is practically free and always plentiful in the later game, other people think its more valuable. It also depends on map settings....for example with tech trades on gold is actually quite valuable....lately I've been changing my opinions on gold to rank it more valuable than I have historically.

Just to take a cut, lets go with the 4 gold = 1 science. That is a common yield ratio we have used on these forums. So factoring that in (with the same time discount as before)....ah, its like 25 science in total, the needle isn't really moved. What's interesting is with the time discount gold is basically a wash. The investment cost of the colony becomes much higher with the discount, so you really aren't getting any effective gold.

 

[bonniepbilly]

For the production I had in mind the cost of the pioneer/colonist. But it might be not significant, now that it was significantly.
Attached is what I meant by calculating an IRR. I probably misused your file, but this is the idea.

 

[Stalker0]

For the production I had in mind the cost of the pioneer/colonist. But it might be not significant, now that it was significantly.
Attached is what I meant by calculating an IRR. I probably misused your file, but this is the idea.

Yes I see what you mean. Only thing missing was to factor in the Turn multiplier (not every datapoint is a single turn, some points are 2 or even 5 turns worth). So I adjusted the net yield you provided by that and then put in the time discount (I assume time discounts work normally in IRR calculations?). That gives me an IRR of 35%.

This seems like a very simple and reasonable way to calculate a comparison (hehe almost like its a standard method or something . Like you said, if I can compare this to a few normal cities I could get a good comparison. My issue is on the data collection. I've gotten some logging help to get the civilization yields every turn recorded, but I don't have anything that records the instant yields per city, and so far I haven't found a coder who can help me with that. So the only way to do it is the painstaking manual approach I used before....and I'm loathe to do that at the moment, it was an exhausting effort last time

 

[bonniepbilly]

It's too generous if you multiply by the number of turns. I added the missing turns instead. I also put a placeholder for the cost of the pioneer.

 

[Stalker0]

It's too generous if you multiply by the number of turns. I added the missing turns instead. I also put a placeholder for the cost of the pioneer.

Ah good to know...the frequency of the data effects the IRR function, so a simple turn product is not sufficient. I'm getting a little statistics lesson today, thank you.