Discussion in 'Civ4 Strategy Articles' started by Vol, Nov 22, 2005.
very interesting. i will tailor my strategy to this and see if it changes anything.
Well, you could design the city to grank Great Engineers and just use them to finish wonders flat, but is that undermining the versatility and purpose?
If you want to get mostly engineers or prophets you are better off spreading the great person points over your empire (since forges and temples only allow you 1 specialist)
If you want scientists, merchants or artists go for a focused city or two.
Great ones so far!
I'm not sure how you've calculated the odds for getting each type of great person, but I hope you're aware of the fact that the odds have nothing to do with the number of GPPs generated for each type, but only with the number of sources that generate GPPs for each type?
(It wasn't obvious from your post (at least to me) how you've calculated that, so I thought I'd mention it)
Actually, no, I had no idea. Number of sources, huh? That seems odd, and potentially expoitable. Does each specialist count as a source? Does it count as a source if it only contributed for one turn?
Don't think I'm going to try and answer those myself, but if anyone else wants to, I think we'd all be interested.
Yes, each specialist counts as one source, as does each wonder.
For that turn, yes. The odds are averaged over all turns though, so having the Pyramids (great engineer source) for 100 turns and adding a scientist for only one turn, you still have a 99% chance of getting a great engineer.
Well, it would have been potentially exploitable if it didn't work as you just described at the end of your post.
So for the purposes of determine what kind of Great Person it will be, all sources effectively contribute 1 point per turn towards that calculation. But for the purposes of determining when a Great Person appears, the full number of GPP per turn is used.
Correct. What I'm not sure about is whether the "memory" which sources of GPPs contributed on which turns gets wiped after producing a great person, or if the city will remember all sources that had ever produced GPPs. I'll investigate that, and maybe write a guide about all this.
Very good article. Thank you for all the work!
I think it's the number of points / total points needed decide what type of GP, say engineers, great wonders producing eng GPP make up 600 of 1000, then the probability will be 60%.
Or it might be the number of points/turn on the generating GP turn that determines the odds, a la, engineer sources generating 18/24 for the turn the GP is born, then it's 75%. Something like that.
I favour the first theory but it would take quite a lot of memory to keep track of all that... but it's highly doable and possible. Don't have CIV4 on this comp to test this
No it's not. I've tested it, see my summary about it here.
Thanks for the link Kylearan : it help
very good stuff.
anecdotally. however, i seem to have 4 or 5 cities producing gp's.
but i like gp's.
You also have to factor in smaller, or hemmed in cities that accumulate a lot of specialists.
and i try to spread out national wonders so as to accumulate more points in more cities.
Very useful guide that. It does beg the question as to if it's worth moving the capital to a high food area so that I can reap the mercantilism benefit. And if so, at what stage - you'd have to think ahead or you'd waste potential wonders.
Furthermore, I wonder how much the difficulty level will affect things? On Noble you can generally count on producing any wonder that you start ahead of anyone else, but as the difficulty rises, you have to rely on specialists only. Moreover, if you are able to go high production in one city and get most of the wonders, with few specialists and high food in another, for multiple specialists would that work on the graph? In the latter case you would presumably go for farms over cottages and ensure it wasn't your capital.
Higher difficulty levels are supposed to not get production bonuses on Wonders. So if you can keep up in tech, you have a shot at getting Wonders.
I didn't understand a digit of your math but I'll take your word on it, GP production has definitely been something I don't understand well enough to use to full advantage. Thank you.
Edit: This doesn't work in later patches.
How get Great Engineers at triple speed.
1. Build the Pyramids.
2. Wait 18 turns (+36 GPP, 18*100% Engineer)
3. Switch to Caste System.
4. Hire 10 scientists for 2 turns (+64GPP, 2*91% Scientist, 2*9% Engineer)
5. Final odds: 90% Engineer, 10% Scientist.
I've obviously been spreading my cities too far out. Dope.
Thanks for the very interesting post. I used to deal with optimization problems frequently and find these questions intriquing. While I need to think about this some more, my gut tells me your model is not properly constructed (or I don't understand your distribution metrics).
Items to think about:
In your model you treat GPP as a finite resource but it is not. It is an infinite resource simply dependent upon time. If the game were to run infinitely, then you would have infinite GPP. This simple issue can severely alter you dyanmics.
The finite resource in question is city-specific resources such as specialists and certain wonders.
Items that impact all cities equally are irrelevant to the question of distributione of the finite resource. These items impact the rate at which GP are produced but not the ratio of one city to another.
There is also a limit on specialists per city. You cannot convert every citizen to a specialist and keep a functioning city. So a city with high production values in other areas can afford to have more specialists than another city.
Some quick thoughts on how too look at this mathematically. This is not well thought out but is somewhat inuitive once put in this fashion.
How long does it take a single city to produce a GP? This is simply a function of time.
G = t(S+W)c
G = GPP points
t = turns
S = number of specialists
W = wonder gpp
c = civ wide modifier (like leader trait or civic trait)
This model describes how a city gets GPP over time.
Now the real goal here is to produce the most GP in the shortest amount of time.
Consider a 3 city civ. and its total civ-wide GPP production rate. If you have S number of specialists, it does not matter what city you put them in. Your overall civ GPP rate will be the same. The issue is that GP are not generated per civ but per-city.
Consider the formula above and 3 cities with only 3 specialists to distribute. If you put one S in each city, you will reach your first GP in ~33 turns since each city is producing +3 GPP per turn. The next GP will be in ~66 turns when the second city gets to 200 GPP. The next will be in ~99 turns. If you look at the rate of GP production; it is 3 GP in 99 turns For the first 3 GP, the rate would be 3GP in ~99 turns. Or a GP every 33 turns.
Now consider if you put all 3 specialists in one city. You now generate 9GPP per turn. The first GP comes in just ~11 moves; the second GP comes in ~33moves; the 3rd GP comes in ~66. The GP production is 3 GP in ~66 turns. or 1 GP per ~22 turns.
Since GP are produced per city, you should distribute your resource in a way to maximize GPP production in a single city.
The same holds true with wonders, stack them in a ctiy to maximize that cities rate.
One caveat is that once you get to a larger CIV, you have to consider sub-optimal distribution because if you can't pile all of your S in a single city, but the general tenet should hold true. Maxize one city at a time.
Just my nickle's worth. Look forward to comments, critiques.
I had a bash at a gp city with the Germans +100% GP rate. I got my capital up to 60gp points a turn. At one point i was making a GP each 10 turns from the capital.
Does reducing specialists in the highest GP for a few turns to allow other cities to make a great person make sense? What happens to your theory if you halve the GP rate of the number one city for 10 turns or so?
I did have to use the civic pacifism for 100% gp rate for this rate of 60. Each specialist was adding 9gp points. Although i needed a religion. Also the national wonder to add 100% but i wasnt able to build this in Berlin. Bad planning and late development of Berlin as GP city. It was interesting the plus 100% was done on the base rate not cumulative. Grrrr.
Begs the question what is better for a GP farm. Lots of irrigation to have loads of specialists or lots of wonders? GP rates over top 5 cities worked out at 60 24 12 12 12 spread over top 5 GP producing cities. I ended up with a lot of scientists. Wasnt ideal need to plan wonders better.
Ideally i would of prefered great artists or great engineers.
1. 60gppt -490
2. 24gppt - 342
3. 12Gppt - 640
4. 12Gppt - 613
5. 12Gppt - 522
(Great Person points turn) (1290ad on monarch standard map)
I would interested in any comments on these figures. Plus any advice on how to improve these numbers.
The only real downside to going for GP points was it didnt aid any strategy on my game.
Separate names with a comma.