@Wodan:
Your "definition" of GP farm presumes that a GP Farm must have the National Epic and thus is limited to one. That is like restricting your definition of a military city to one which has the Heroic Epic, which I think you'd agree is in error.
Not quite. You're calling a "GP Farm" the same as any old "specialist city", which is not what I would do, myself. What defines a GP Farm is the National Epic, because that's what leverages the resources and provides the most bonus.
Yes, you can have multiple cities making units. That does not mean that each of those cities is your "Military Academy city" or whatever you want to call it. Plain and simple, if you have a limited number of some mechanics (which you do), namely settled generals and the Heroic Epic, etc, then there's only one city which will provide the most bang for the buck.
You can have a second city make units, but those units will take longer (no HE), and will have less XP (no settled generals).
Your interest rate example assumes the ability to assign infinite number of specialists (as well as a zero-sum situation) which does not apply when dealing with specialists
You can assign more specialists than you can possibly feed, even if every tile in your city is a food resource. In real terms, this is equivalent, so YES it does apply.
You also assume zero risk for both the 5% and the 3.5% accounts but usually a difference in rate implies a difference in risk. In the GP farm world this risk is that you will get an undesirable great person next.
Up above in this thread I clearly said that giving a "mixed bag" of GP was part and parcel of the strategy. Some people call this playing the "GP lottery".
I do maintain that there is no such thing as an "undesirable GP," particularly with this strategy. As long as you don't tie the success of your whole game into getting one and only one type of GP (such as doing a multiple religion Shrine strategy), then this isn't an issue.
I am unsure how you get the 5-6 golden ages number, though I presume you are speaking in hyperbole. Even with random great people getting the necessary 4-5 unique great persons is quite difficult; more-so if you are using wonders for a portion of the GPP since there is generally a distinct bent toward the kinds of GP your wonders will be generating.
You are building as many wonders as possible, and thus you do not have a "bent". You already have a mixed type of wonders. And, you can (and should) change your specialists to bias your GP farm toward the type of GP you want at that particular moment.
I would even go as far as to ignore the National Epic if I were to try and maintain specialist GP farms since there is probably no single farm that I want to produce more GP than the others, especially at the risk of getting a great artist.
So you by choice are running multiple bank accounts at 3.5% interest, and voluntarily saying you choose not to have a 5% bank account. That's your choice; all I'm pointing out is that you could be getting 5% instead of only 3.5%.
Let's look at an example with real numbers. Say you have 3 cities. Each of them is running 4 of a different type of specialist. Plus, each of them has built 3 wonders providing that type of GPP. So, each city is generating 4*3 + 3*2 = 18 GPP/turn.
Say you have made 5 GP so far. City 1 will make a GP at 600, city 2 at 700, and city 3 at 800. Right? If all started at the same point (which isn't going to be true, but will give a basis for comparison), then the following are the # of turns for each city to make a GP:
City 1 = 600/18 = 33 turns
City 2 = 700/18 = 39 turns
City 3 = 800/18 = 44 turns
If you had put all that in the same city plus NE, that city would be making 108 GPP/turn. So, following are when the GP would arrive:
1st GP = 600/108 = 6 turns (6 turns)
2nd GP = 700/108 = 6 turns (12 turns)
3rd GP = 800/108 = 7 turns (19 turns)
And, if it waits until turn 44 (like the first example), it could make 3 additional, for a total of 6 GP. Not only that, but you get the GP much earlier (the first one on turn 6 instead of turn 33), which is huge benefit.
So, it's a matter of choosing between 3 GP where you get to pick the type of GP, versus 6 GP where you get a mixed bag.
In addition, the first example also has diminished specialization... i.e., you have to devote 3 cities to doing this. In the second example, those other 2 cities can be doing whatever you want, rather than running merchants, or engineers or whatever, if that is not ideal for that city. In other words, if the only reason that city would be running merchants would be to generate a Great Merchant, then that's a negative of the first example.
(The above both ignore Philo and Phi bonuses, which would be the same in either case.)
The most important parts of creative are the fast first border pop and the huge amount of early hammers it saves you (no monument, half-price library).
