South African/Boer civ

[civ_king]

To just quickly return to the suggested civ:
The proposed UU would be a replacement of the machine gun but with the current idea it does neither fit with the tech requirements nor the resources of the normal machine gun.
Would it be possible to do something like the Indonesian UU? To take the unit and make it accessible much earlier, actually make something totally new out of it? Or would that be off limits?
Of course, if implemented, the Boers would need a horse close to their core area that spawns with them.

By the way, I would be more than willing to compose a list of Great People, city renaming (actually just translating the Dutch names into Afrikaans) and so on for that new civ.

Tamils also have an earlier UU.

If is is to replace Machine Gun then it should have 14-15 and +50% defense in exchange for the ability to attack.

Boers brought their horse from abroad and bred them on a small scale level there.

TL;DR

Might as well add Canada and Australia at this point.

Definitely not Canada as that makes it too easy for US.

 

[banestrider]

I think all decolonization civs should be able to be enabled or disabled with the exemption of Mexico because it is there to be challenging to USA. That includes Argentina, Brazil, South Africa, Canada, Australia. That would put to bed the argument that they will just slow down the game because people with less powerful computers (me) can disable them and people with more powerful computers can play with them if they choose.

 

[Leoreth]

It's already a rule that optional civs will spawn regardless of your settings if they are your neighbor.

 

[ales_]

BTW who is considered who's neighbors?

 

[Leoreth]

A bunch of different definitions depending on context. Here I used civs that would normally expand into the region taken by the optional civ, i.e.:

Tamils: India and Mughals
Tibet: China
Moors: Spain and Portugal
Poland: HRE and Russia
Congo: none

 

[civ_king]

A bunch of different definitions depending on context. Here I used civs that would normally expand into the region taken by the optional civ, i.e.:

Tamils: India and Mughals
Tibet: China
Moors: Spain and Portugal
Poland: HRE and Russia
Congo: none

Tamils should be there for Indonesia.

Congo should be there for Portugal and Brazil, both need slaves especially the latter.

 

[Leoreth]

I think the Tamils are included for Indonesia, I was writing this from memory.

 

[VSarius]

I like the idea of additional ocean tiles to prevent Dark Ages settling of Africa by Euros. Sounds like a simple and clean solution to a simple problem.

Inre to the AI;
If the AI could be sufficiently upgraded to actually be a threat in the mod, UHVs would have to be toned down, but it would make for a more interesting game. The problem is of course that I doubt that can be done, even in base BTS the K-Mod AI can still be gamed and out-witted, and base BTS is significantly less complex. As such, until there is a time that the AI can be made not-********, balancing by taking them into account is folly.

 

[Leoreth]

To have a really non-******** AI, you would need actual AI with learning capabilities, not just the "let's apply some more or less arbitrary weightings to a limited set of possible behaviors and choose the best" approach most games use.

 

[iOnlySignIn]

^ In such a game as RFC (where every player is inherently different with different goals) it's much easier to give AIs free cheats like Firaxis does. As long as those cheats appear realistic to the player they are perfectly fine IMO.

The issue right now with DoC is that those free cheats are not quite good enough. The Greek, Roman, and Mongolian AIs still fail rather miserably even with their free conquerors, for example.

In case of Greeks and Romans they need some more Culture boosts in their conquered cities (to represent Hellenizatin) instead of being stuck with 1 tile cities. Barbarians in the middle east also need to be toned down in case the Human player is not there (Human China building the Great Wall early really makes AIs' lives miserable.)

In case of Mongolia (especially on Normal speed) they need more pre-existing Roads across Eurasia to actually see Baghdad before they collapse from instability.

 

[VSarius]

To have a really non-******** AI, you would need actual AI with learning capabilities, not just the "let's apply some more or less arbitrary weightings to a limited set of possible behaviors and choose the best" approach most games use.

Well, for one you would have to write the AI in a more...multi-layered process.
A Grand Strategic, Strategic, and Tactical AI. The Grand Strategic would be the easiest, model it's overarching goals around the optimal plays that the human has discovered, so to roll with the example, the England AI would found London/Edinburgh/Dublin/Durban at the start, and then plan to win the game via it's UHV, making it's strategic decisions around that.

The strategic AI would be focused around tech paths, and responding to the development of the game. Here...yeah I can't see a way to preprogram an AI to respond properly. The tactical would be responsible for turn-based moves, optimizing city improvements/tile improvements, and troop movement. This is where the AI is least competent, it simply has no idea how to wage war or improve it's territory, the only way it is even remotely competitive with the human is with the astronomical bonuses it acquires on Deity.

In regards to improvements that could be made without touching the AI. They kind of feel like sledgehammer moves and I'd need to play around more on Emperor to really say what they should be without making the game unwinnable for the human, as is, the Babylon UHV for example is very very close as is.

 

[Leoreth]

Well, in general it's very easy to say "the AI should do X when Y" without ever having looked at what the AI code of a given game actually looks like.

 

[civ_king]

Well, for one you would have to write the AI in a more...multi-layered process.
A Grand Strategic, Strategic, and Tactical AI. The Grand Strategic would be the easiest, model it's overarching goals around the optimal plays that the human has discovered, so to roll with the example, the England AI would found London/Edinburgh/Dublin/Durban at the start, and then plan to win the game via it's UHV, making it's strategic decisions around that.

The strategic AI would be focused around tech paths, and responding to the development of the game. Here...yeah I can't see a way to preprogram an AI to respond properly. The tactical would be responsible for turn-based moves, optimizing city improvements/tile improvements, and troop movement. This is where the AI is least competent, it simply has no idea how to wage war or improve it's territory, the only way it is even remotely competitive with the human is with the astronomical bonuses it acquires on Deity.

In regards to improvements that could be made without touching the AI. They kind of feel like sledgehammer moves and I'd need to play around more on Emperor to really say what they should be without making the game unwinnable for the human, as is, the Babylon UHV for example is very very close as is.

If Leoreth were writing a game from scratch this would be implementable, but he isn't and while Civ IV is highly moddable in general the AI isn't able to be changed to be be all that great as artificial intelligence isn't all that sophisticated right now. The AI playing historically and the AI playing as a challenge are two very diffferent things.

 

[VSarius]

Well, in general it's very easy to say "the AI should do X when Y" without ever having looked at what the AI code of a given game actually looks like.

It's more a post about 'What-ifs' then actual 'Do this now!'. I'm quite well aware that such sweeping changes aren't possible, would be swell if they were.
I feel like there just isn't an elegant solution to AI incompetence, you either have to shower them in bonuses or they are buffoons that are no different from large independents.

 

[Leoreth]

I think you've missed my point. You haven't proposed anything so far because qualitative statements are easy to make, but meaningless.

 

[VSarius]

Like I said, I need to play a lot more games on Emperor before actually suggesting something that has a basis behind it. I don't think it's possible to make the AI behave both historically accurate, and be challenging. Increased aggression, higher unitprobs, favoring cottages as tile improvements, slashed tech costs, and faster expansion via an extra worker+settler are the types of sledgehammer moves that should be avoided. Say if Japan were or Korea were able to field larger armies and likely to attack China, even wipe it out. I'll get back to you.

 

[Lone Wolf]

Increased aggression, higher unitprobs and favoring cottages (and also low techtradeknownpercent and high tech trade treschold) are not sledgehammer moves at all, since it actually affects the way the AI plays without giving it any bonuses. Slashed tech costs and extra settlers, however, are sledgehammers which should be used sparingly.

 

[Leoreth]

You can play 500 games on Emperor without learning anything about how to actually improve the AI. You're confusing goals with solutions.

 

[VSarius]

Alright, the reason I said I'd need more Emperor games played is because suggesting changes that would invalidate UHVs is something that should be avoided. Or at least to get a grasp of what would need to be changed to still keep them achievable. As I see it the goal is to make the AI a valid and credible threat, while keeping UHVs possible.
I think a step in the right direction would be disabling stability as a mechanic for the AIs (or at least severely decreasing it, in the realm of War Weariness-levels of decrease) as well as the increased tech costs for larger empires. As it is, the AI can expand to a large territory. The Ottomans, pre-nerf Mongols, certain europeans, and rarely Prussians have demonstrated they can expand, but then all that happens is they either stagnate in technological irrelevance, or collapse. Basically any AI that does 'succeed' instead fails. Couple that with the other changes, and they may actually pose a threat.

 

[Lone Wolf]

Basically any AI that does 'succeed' instead fails.

That's indeed a problem in RFC and its spinoffs, but I think that stability in general needs to be a bit softer anyway. See also the relatively successful Prussia and Russia in my Mughals game - as an example of what can happen with stability softened (and how a strong Prussia turned out to be beneficial to me by allowing me to sic her on the threatening Ottomans). Also, see American success-collapse in that very game (the AI doesn't really understand stability, the USA would have probably survived had they spammed courthouses in all their cities).

Also, if you want to slightly improve the AI performance, set AI unit train percent to 95/90 at monarch/emperor, instead of 110/100. Since Leo prevented the plague from killing its build-up units, however, the effect will be relatively modest.