Sawmills

[Nelphine]

I'm inclined to say dark elves are too strong at the moment but it's difficult to say.

However, what are you specifically worried about in terms of housing benefitting these races too much?

[Nelphine]

Seravy, to your question: adding lairs should be done. This balances extra starting gold the AI has. The original reason for this was due to sawmill cost, but if sawmills and associated change, that strictly won't impact the humans or ai noticesbly more (unless you need to reteach the AI what to build first).

Therefore, the sawmill and associated change can be done independently of lairs.

I'm not convinced that the housing change is a good one, but I am convinced that I could certainly be wrong, and that current housing isn't great.

So I'd say add lairs. 

I would say add sawmill and associated. You've indicated the amount of work (including reteaching AI what to build early) is reasonable, and you are convinced that someone who chooses housing, or doesn't, is not significantly worse off than the other - and equally that someone who buys the sawmill first, or doesn't, is not significantly worse off than the other. That means the change won't massively hurt people who are used to sawmills, nor be a problem for new people. Therefore over all it seems to be solid to implement.

[Settemio]

I'm inclined to say dark elves are too strong at the moment but it's difficult to say.

However, what are you specifically worried about in terms of housing benefitting these races too much?

I just mean that 1 population of Barbarians is not worth as much as 1 population of Dark Elves, in that the Elves also provide 1 power. So a buff to Housing benefits them more than it does a Barbarian. However, on second thought, maybe the fact that racial growth is part of the new formula means new Housing is actually less effective for Dark Elves, or that it balances out?

[Nelphine]

My own conclusion is that's part of the point of playing the race. For instance, dark elves you have to spend a retort pick on, which is a major cost. They also get slower growth and worse unrest tables than some people. In return they get better population, all buildings, and a particular set of better units.

I think the housing change should affect everyone equally; if a given race benefits too much, that should be balanced by the other properties of the race, not by changing the housing formula.

[Seravy]

True, ROI of housing is better for races that produce power of research per population so this is a small buff to those races. (or maybe even not so small, if you find a pop 25 place to settle)
Still, raising the productivity of 1 population from 2.25 to 3.25 only makes it worth about 2-3 population higher than usual. It's about the same as having a Builder's Hall.
(In case of Halflings, even more, as they also produce extra food. So halflings will be super good at housing. So good that maybe the AI needs to learn to do it for halfling cities.)

I'm a bit unsure on how the AI should handle it, it basically has two routes to take. Either buy marketplace for gold, then do housing and switch to building a sawmill around pop 4-5, or buy sawmill first and don't do housing. The former is better for long term economy but the latter will give the AI more military presence (gold won't go back into the cities as quickly, while production has immediate effects). It would probably be best if they always did sawmill, unless they play halflings.

I'll finish my ongoing Gnoll game today and start working on implementing these tomorrow.

[Nelphine]

Ok I still don't get it. Why would you ever buy a market place, if you can buy a sawmill, and build a marketplace? Heck, depending on exactly what you provide to the city as free resources, you can just build a marketplace, and still end up with more gold.

[Seravy]

Well, the marketplace produces 8 gold a turn which buys 4 hammers. The sawmill costs 4 gold but produces 6 hammers, so it also produces 4 hammers. The sawmill is cheaper so if you need hammers (local production), you buy the sawmill. If you need gold (production elsewhere, or money for heroes, mercenaries, maintenance, whatever) then you buy the marketplace.

If you have the money, you want to spend it as soon as possible (each turn of delay means you miss out on the resource the building would produce, and gain nothing at all for doing so). Meaning you want to spend it prior to housing - if you already have it by then. However during housing, the hammers are lost. So you don't want to spend it on anything that produces hammers. So you buy the marketplace, accumulate the gold, and after housing you can use it to buy something locally instead. The cost difference is 12 (80 for market, 68 for sawmill) and sawmill "produces" 6 hammers a turn that goes to waste. So if you do housing for 3 or more turns, buying the sawmill is strictly worse than the marketplace.

Is that clear?

This so far is for the human. The human could also opt to buy any other eocnomy building before doing the housing, as long as that building does not produce hammers. Wizard's guild, amplifying tower, anything you can think of, just no hammers.



However, the same also applies to the AI. The AI could look at the available gold and say "I'll most likely be able to buy this marketplace immediately".  In which case, queuing and buying the marketplace, letting it produce the gold, and do housing meanwhile would result in the highest ROI. (buying library and magic market before housing would work the same way. Or anything else but I won't teach the AI to instant buy more expensive and less essential buildings so let's keep it to those three.)
However, while the AI is doing housing, it is effectively spending turns on building economy. Which means it's not doing military. The current AI does "sawmill (for gold if able) -> Smithy -> swordsmen garrison if needed -> Fighter or equivalent"  meaning they spend on military starting turn 2 of the city. They do that because they have to - not doing military means the human player can march in and have the city for free, without a real battle. Yes, 4 swordsmen, even halberdiers aren't that great but having to overcome them contributes to the game being a game, and does mean the human needs to also have some sort of military and cannot get away with just a magic spirit as their invading army.

So while the housing is more economical, if you need troops, and the AI needs troops early, then you have to drop the housing, buy the sawmill, and start producing anyway. Which actually makes this change quite bad for the AI - they won't be affected but the human can, if they avoid early aggression, choose the much better ROI housing. However since peaceful strategies are underpowered, this should not be an issue - except it is because being human, you can just use the capital and summoning power for your early military needs.

So we need to decide which do we want for the AI :
-Sawmill first (buy), military first, as before, in which case the AI falls behind in economy because the human can do better (and fails to capitalize on its economy bonus that has snowballing potential)
-Marketplace first (buy), housing, then build as normal (sawmill, military)
-Marketplace, Library, Magic Market first (buy), housing, then build as normal (sawmill, military)

The second gives the AI unspend gold, so they might pump out more units in their capital or more developed cities, but the new settlement will not get a garrison through that. Housing would last... it was 10 turns for the human. The population growth bonus the AI has is the lowest in the categories, so about 8-10 turns depending on difficulty. After these turns, the produced citizens give the AI extra hammers to work with. Assuming the AI buys the sawmill immediately now, they have about 3 extra citizens worth of hammers (again ignoring gold as it won't be used for garrisoning the city locally) - if that's 1 farmer and 2 workers, then 4-8 hammers depending on difficulty.
So we lose 11 turns of progress with base amount (9-18 hammers, again difficulty) hammers but instead get 4-8 extra for each turn after that. Meaning we are difficulty*(99-4*turns) hammers behind, in other words we catch up in 22 turns. So on turn 33 we are equal in our military progress as the other option, but have an extra marketplace, and an extra 3 population. (we did spend on the marketplace but it was also producing so it already paid for itself and even got us extra money!)
So where does this 33 turns put us?
Well... the AI has 9 hammers per turn (*difficulty). It goes smithy (80), 3 swordsmen (3*25), Fighter (250).

So on normal difficulty, it's turn 9 for smithy, turn 17 for the three swordsmen, turn 45 for Fighter's.
If going economy first, it's turn 10 for stopping housing, turn 16 for smithy, turn 22 for the three swordsmen, and turn 41 for Fighter.


...okay, so for Normal difficulty, it's a 5 turn delay on the swordsmen (who cares), and 4 turns faster for the FG plus the other benefits. The AI clearly needs to go economy first...

For higher difficulty, assuming double hammers, it's half as many turns for all, but housing which is probably about 2 turns less.
So it becomes turn 5, 9 and 22 with sawmill and turn 8, 11, 23 for Fighter's. That's worse, but only by 1-2 turns and you get the economic benefit.

...so that means the AI need to be reprogrammed to do
housing to pop 4 (or if we want to make it that smart, pick depending on max pop and race) ->sawmill for gold -> the usual things.
Plus it needs to learn to interrupt housing if it can instantly buy a marketplace (and maybe library, magic market) and let them do so then continue housing.

...which means starting gold becomes super relevant for the AI suddenly, as it will be able to invest it into major snowballing, (or power/research if we add the library/magic market to the strategy).

[Nelphine]

Ugh. Ok followup problem.

Housing isn't fun.

Stealing cities from the ai where it hadn't built anything? Feels like the AI isn't playing.

Building housing as the human? You have to check every turn (or have some other method of keeping track) if your cities have reached a particular population, that isn't written anywhere in game. (This is particularly important later in the game when production can take 5-15 minutes and you forget that anything has gained any population, let alone which one gained population, especially if you have multiple cities with the same name, which is fairly common). Yes you can check the turn events advisor every turn, but unlike normal production it doesn't jump in your face to tell you.

[Seravy]

Well, umm I can agree with that but it's already part of the game. At the very least with my formula housing is only worth it early, and it doesn't make a significant difference if you turn it off +/-1 or 2 population earlier or later (ROI is about the same as the thing you'd build anyway). Even if you don't build it at all, it's still not that bad unless it was a very large max pop location.
We were doing it anyway while waiting for having enough gold to buy the sawmill, so it's not new.
Removing it entirely is an option but it isn't really an option I like.
We can reduce the desirability by reducing the bonus the Builder's Hall provides. (In fact, do we even need the Builder's Hall to do that? It just means an additional turn of wait time and basically a 50% penalty if you don't have 40 gold. It's like a mandatory first building again except for less gold. And a racial one so races who don't get it are massively penalized.)

We might want to find a new role for the Builder's Hall but I'm pretty sure it should not boost housing in the new system. Any ideas? It should be something that you want to get sometimes, but doesn't really matter in the big picture.

Stealing cities from the ai where it hadn't built anything? Feels like the AI isn't playing.

As calculated above this isn't a thing. The AI will get to the fighter's guild units faster or at the same time even if it uses the new, more economic build order. They will be, in worst case, 5 turns slower on finishing their 4th swordsmen but those aren't really that effective at defending anything anyway.
The city will already have a sawmill, marketplace, magic market and library, assuming the AI had the gold to buy those, otherwise it'll only have the sawmill just like it does now.

So what changes?
First, you won't get to steal the sawmill as quickly - instead of at pop 1, it'll show up after 8-10 turns of housing. But, if the AI had money to invest, you get to steal a marketplace, library or magic market, or all three, instead (as they would buy those before housing). So overall, there will be more things to steal before the AI has the basic 4 swordsmen (or better) up.

(not sure if that's your problem or the lack of troops. Or you assumed the AI will have fewer buildings, which it won't unless you play on like, Normal where they won't have gold to buy them early.)

(we might want to make buying library or magic market before housing personality based, as those don't snowball as much as gold. Or maybe just the library.)

[Seravy]

Question, what name should "basic production", those 2 hammers, appear as in the detailed city view? I rather not add it under "workers" or "terrain" because it would be confusing.

Also an idea :
New outposts start with Smithy already built. Smithy produces the two basic hammers. Smithy has a higher (125-150) cost.

Pros
-Conquest becomes less desirable. If the smithy breaks, you end up with a hamlet that's worse than a newly founded one (need 200-300 gold to repair smithy)
-AI can get those basic swordsmen faster
-Swordsmen tier unit might get to see more use (you can build them asap if you wish, no need for a cost 80 building)

Cons
-Spearmen even less desirable if swordsmen available immediately
-AI will be several turns more behind if using the "economy first" strategy because the time to build up is less, so less chance to "catch up" and break even with the housing delay.