The mechanic of harvesting is quite brilliant, I think. Especially with the adjacency and placement restrictions we have, it would be infuriating if
literally the perfect campus site was blocked by a sheep resource on a hill. Incan fans from Civ5 will remember the rage of being denied your +4 terrace farms. Injustice most high!
The question with harvesting, though, is whether or not it's balanced with the yield of leaving the tile intact. Well, this is easy to compute, if we know a few things:
- What is the value of harvesting at any given point in the game?
- What is the value of 1 production at any given point in the game?
- How long do games last?
We know #1, it's an equation in the files.
Harvesting is worth either base 20 food/prod (woods/jungle/marsh) or 25 food/production (a bonus resource.) This scales up linearly as you progress through the tech or civic tree, to a maximum of 10x at the end.
So, in other words, HarvestValue=20*(1+9*max[techprogress%, civicprogress%])
For #2, we can look at production costs over the course of the game. If you look at districts, I believe the base district cost (before applying any discounts for Uniques or how common they are in your empire) follows an identical formula as harvests- it scales up by 10x over the game. Unit costs do the same thing- mounted units cost between 65 (chariot) and 660 (Modern armor); infantry range from 40 (warrior) to 660 (Mech inf.) So they also have a roughly 10x scaling. What this means is 1 hammer at turn one is the same as 10 at the end of the game.
On point 3, the game might have been balanced around some number like 500 turns, but in reality, I think it's pretty reasonable to guess 300 for a common case. Yes, you can finish faster depending on map, etc, but I'm being generous to the case of more average players.
Combining #2 and #3, we can infer a production cost "inflation rate" or "interest rate" of ~0.77%/turn.
Anyways, let's focus on the resources. If we borrow some math from the world of finance, we can actually compare to see what kind of deal we're getting. Specifically, we can treat the resource's bonus yield as an annuity payment, and the harvest yield as the lump sum value. Obviously, as the number of turns left in the game dwindles, harvesting becomes a better and better deal. What does that tell us about how resources themselves should be improving over time- what yield would a resource need to be worth the same as harvesting it? (Again, this value will skyrocket at the end, because as there are less and less turns left, the equalizing tile yield eventually becomes the harvest yield by the last turn)
I'm making a hand wave assumption that we progress at a consistent rate throughout the tech and civics tree. This is obviously not quite true, but this is a crude analysis.
Okay, let's have a look:
Interestingly, around turn 100, is when having +1 yield from that bonus resource starts to become insufficient. For forests and jungles and marshes, this crossing point happens around turn ~125, or sometime in the middle ages.
But, we also can recall that while nothing makes resources yield more, some techs boost the improvements on them.
The bar chart shows what a resource would need to give as a tile yield to be equal to the harvest value. Note that there are a few points that matter: when it crosses whole numbers like 1, 2 and 3. 1 is the base resource (most resources give +1.) 2 is when the improvement is applied, and 3+ would be when the tech boost to those improvements come in. Note that, in assuming all these resources are improved, the builder charge to improve it will ~roughly~ cancel out the builder charge to harvest. That's why I'm ignoring that messy math here.
"Turns" is really "progress through tech tree," but since the tech tree progress determines both harvest yield and tile yield via improvement boosts, it doesn't matter if we compress or expand certain parts of the x axis to account for real game progress.
Anyways, let's look at the pasture. The pasture gets a nice +1. So if you build a pasture before the bar chart hits 2.0, you're coming out ahead. Exploration gives +1 food to all pastures, so exploration (which comes about halfway through the game) boosts you to +3 for that resource tile. The equivalent harvest value is only 1.5 yield per turn- going strong! Globalization also gives +1 production, but this comes so late it doesn't matter. Plantations have it the worst, by the way. One gold is worth 0.5 production or food, in the math of civ6. So they start with effectively +1, and then get extra food at scientific method, which comes pretty late.
Things that you can build a farm or mine on act differently- since farms and mines get the boosts regardless, you should think hard about harvesting some of those resources. Especially if you're expanding into new turf and don't have the improvements down yet.
The caveat of all this is that most of the things you want to harvest for (wonders) are not about absolute value, but rather positional. In a race for e.g. Big Ben, it doesn't matter how good your production will be after I finish it. So this is where chopping is actually a huge advantage- can be worth much more than simple computing the lump sum value of that +1 yield per turn.
Lastly, when accounting for that premium of being able to beat someone to something by getting yield up front in a harvest, the game is actually pretty balanced over most of it. Features should almost always be removed after a point, but resources have surprising staying power. However, in R&F, while they threw some nice buffs at improvements over resources, they also add our Lord and Savior, Sir Magnus Choppington. You can imagine that outright doubling the value of chops just wildly skews things, which is true. It's quite a powerful thumb on the scale.
If they really wanted to fix this, I would siggest adding some inexpensive city center buildings from civ5: the stable, stoneworks, forge, mint, etc., which added a boost to relevant bonus resources in the city. This would help give some extra sway to bonus resources so you don't end up with such barren maps by the end.
