So, are these mods so necessery? is the game really out balance?

The AI will never fully understand anything, not unless there is a major breakthrough in technology that allows for faster processing power (which lets us get more powerful AI).

I would still prefer an AI that doesn't play to win but merely plays to become powerful. Like, an AI that plays to win but doesn't know how to is really a waste of space, am I correct?
 
I would still prefer an AI that doesn't play to win but merely plays to become powerful. Like, an AI that plays to win but doesn't know how to is really a waste of space, am I correct?

Basically. The current AI doesn't generally play TOWARDS any goal at all, but the practical effect of its semi-randomness is to move it towards a higher effectiveness. The closest it comes to long-term planning are the "Grand Strategy" AI flags, which basically translate as "which victory condition are you working towards?", and the Economic Strategy flags for the critical short-term stuff ("needs workers", "has enough units"). This sort of setup can be expanded with very little modification to develop more coherent long-term strategies for the AI.

This is why I mentioned working within the AI's limitations previously; you don't need the DLL modified if the best strategies are the ones the AI already knows how to do. Consider ICS; humans figured out that by playing a very "artificial" playstyle, they could consistently beat the AI even on the highest difficulties. Certain modders (see also: Thal) figured out that tweaking the base unhappiness, increasing the minimum distance between cities, and lowering the number of specialist slots in a Library would, together, remove ICS as a viable strategy for the players, without harming the AI's normal strategy. The devs, obviously, agreed since they implemented those XML fixes into a subsequent patch without any apparent change to the underlying code.

Given that, no one can seriously claim that balance can't be improved without a DLL change, since this has already happened. So why is it so hard to believe that there might be more changes like this that can improve the balance even further?
 
yes yes I know... but for laymens terms, its fine.

(I could get into how it's actually insulting to even call it an AI, but that's a different tangent)

;) No problem. Spatzimaus used the same "AI doesn't understand anything" argument against me in a long ago thread. Then I used it against him when he made the error of putting "AI" together with "understands" in the same sentence. Anyone modding or playing civ5 understands this stuff. But it's hard to resist using this line.

And yeah, there are some things that can be fixed without the dll, but to really make the game better, the dll is needed.

To make the existing AI really function better, yes. However, to "add" AI (for example, forcing the AI to do a particular beeline that Deity players have identified) is just as easy to code with Lua as C++. It's difficult programming, no doubt. But if no one is interested in doing this programming in Lua (where it is possible) then I have little confidence that someone is going to do it in C++ when the dll is out.

Perhaps you should try some of these balance mods before declaring that they can't really make the game better?

It would be nice if some of the Deity players (MadDjinn and others that post a lot in the strategy forums) tried out some of the more developed "balance" mods. These are the folks that would be the very best testers -- zeroing in on "metagaming" tricks (or "exploits" or whatever you want to call them) where the human can do it but the AI doesn't compete. There are many examples where the AI really just doesn't even play. In Civ4, you could be hugely behind and catch up by tech brokering. In Civ5, the human does a lot of things that just aren't even in the AI repertoire. I'm not saying the AI does it badly. I mean that the AI just doesn't do it at all. An example in Civ5 is selling resources for cash. I don't know if that is absolutely necessary for Deity victory, but it is certainly one of the tools that a high level human player has that the AI doesn't. Frankly, I don't really think Firaxis will ever "fix" this because it is not really a problem for 99% of players. However, a good modder will want to do one of two things: 1) program some AI logic so the AI can play this game too (I believe this specific example could be done in Lua), or 2) do some game rule changes or re-balancing to effectively remove this as a useful strategy.
 
However, a good modder will want to do one of two things: 1) program some AI logic so the AI can play this game too (I believe this specific example could be done in Lua), or 2) do some game rule changes or re-balancing to effectively remove this as a useful strategy.

Right. The path you choose with these will depend on whether you think the strategy in question SHOULD be part of the "right" way to play the game, but it's not hard to level the playing field in these ways without DLL access. ICS was an excellent example of where a few simple XML changes can fix the problem, but there are plenty of other examples of imbalances where the human does things that the AI won't or can't.

Here's one I discovered recently: the AI won't hook up his water-based Oil deposits. Ever. (This might have been broken in a recent patch, but it's definitely a problem now in vanilla.) He'll trigger the "needs a water tile improvement" strategy, he'll make a work boat for the resource, and he might even move the work boat to the Oil tile, but he just won't trigger the action that creates an Offshore Platform. Since only ~1/3rd of the oil in the world will be water-based, this won't drastically skew game balance, but it's something that clearly needs to be fixed. (This also applies to any other water-based strategics you add; my Alpha Centauri mod adds two strategic resources that can spawn in water, one of which ONLY spawns in water, so I've paid more attention to this issue than most folks.)
It's obviously a bug. No one's going to pretend that the AI shouldn't want to hook up that oil. So you can either level the playing field by removing a human's ability to hook up his own water-based Oil, you can force the AI to improve his Oil through some sort of override (forcing the nearest work boat to move towards the deposit and sacrifice itself), or you can change the game such that work boats aren't needed to improve the Oil in the first place.
 
And the point that we're making is that this isn't necessarily true, if changes are made in a way that is sensitive to the limitations of the AI.

good thing the person I responded to actually read/quoted the next thing I said.

Actually it does. Try it. For example, in vanilla Civ Maritime city states are ridiculously powerful, but the AI doesn't always realize that. So the human can get a huge advantage by investing in them. In a mod where Maritime city states are less powerful, investing in MarCSs is not a no-brainer decision, the playing field is more level and gameplay is much improved. The human is "forced" to act more like the AI (who doesn't always grab the MarCSs) because now it isn't always a good strategy to devote your gold to them.

Actually, you just proved my point. You suggest that AIs don't know why Maritime CSs are strong, and advocate ruining them for the human player instead of fixing them. They aren't *that* powerful, but I see AIs making them allies a lot more than making militaristic civs allies. That's from experience, not from code, so take it for what it's worth. So that tells me that the AI might still utilize CSs somewhat. So, instead of forcing the human player to act like the pathetic AI, maybe we do some work on the AI instead? just a thought.

Many of the balance mods (VEM for example) remove the ridiculous AI happiness and food bonuses that exist on the higher difficulty levels.
The game remains hard on high difficulty levels because many of the tricks and semi-exploits that let the player win are weakened too.

Perhaps you should try some of these balance mods before declaring that they can't really make the game better?

Here's the problem: The AI is given the 'ridiculous' AI happiness and food bonuses for a reason. (really though, the mass gold ones are worse) It's to cover over issues with the 'AI' programming. SO making changes to that annoyance, without fixing the real issues, means you're just shifting the problem somewhere else, not balancing it. Just messing with the human players use of the real game does not a 'balance' mod make.

Maybe it's just my understanding of what 'balance' means vs some peoples opinion of just forcing the player to act badly to ensure the AI isn't at much of a disadvantage that's at issue. Random thought, that.

Oh, and you should check my blog for fresh ideas. I actually think there's stuff that needs balance and there's appropriate ways to do it. I also expect that the AI needs a major overhaul to actually provide competition to the player. Then again, MP is a a better choice for balance questions.

Basically. The current AI doesn't generally play TOWARDS any goal at all, but the practical effect of its semi-randomness is to move it towards a higher effectiveness. The closest it comes to long-term planning are the "Grand Strategy" AI flags, which basically translate as "which victory condition are you working towards?", and the Economic Strategy flags for the critical short-term stuff ("needs workers", "has enough units"). This sort of setup can be expanded with very little modification to develop more coherent long-term strategies for the AI.

This can be fixed fairly easily once the .dll is out. Yes, I expect someone with real experience to do it, but then again, I didn't expect much from Firaxis when Sid made comments about how stupid the 'AI' should be.

This is why I mentioned working [...] The devs, obviously, agreed since they implemented those XML fixes into a subsequent patch without any apparent change to the underlying code.

Amusingly, ICS for civ 5 came out of watching the AI settler spam all over the map. Including locations that the player couldn't possibly settle. The AI does it so much better than anyone else. Even with the changes, the AI is still better than the player at ICS, and yes, it still does it.

No one is taking anything away from stuff Thal and some others have done. Then again, some of it was obvious and from Civ 4 in the first place. But that's a totally different discussion.

Given that, no one can seriously claim that balance can't be improved without a DLL change, since this has already happened. So why is it so hard to believe that there might be more changes like this that can improve the balance even further?

Minor changes are minor. That's obvious. Real changes require more.

To make the existing AI really function better, yes. However, to "add" AI (for example, forcing the AI to do a particular beeline that Deity players have identified) is just as easy to code with Lua as C++. It's difficult programming, no doubt. But if no one is interested in doing this programming in Lua (where it is possible) then I have little confidence that someone is going to do it in C++ when the dll is out.

I'm less concerned with doing what Firaxis did and more likely to implement something better.

As per Lua vs. C++ -- I know more C++ Computational Intelligence programmers than Lua programmers. (and since the former group is small, the latter must be worse) C++ is more robust than Lua, and realistically, a real programming language. (hence the core is still C++, not Lua) But then again, I'm biased.

It would be nice if some of the Deity players (MadDjinn and others that post a lot in the strategy forums) tried out some of the more developed "balance" mods. These are the folks that would be the very best testers -- zeroing in on "metagaming" tricks (or "exploits" or whatever you want to call them) where the human can do it but the AI doesn't compete. There are many examples where the AI really just doesn't even play. In Civ4, you could be hugely behind and catch up by tech brokering.

Check my blog... There's stuff there that I absolutely consider issues. But here's some issues that you can't fix with Lua:

removing the minimal one damage from ranged or other attacks closes loopholes and balances combat a bit more. This is something that human players use a lot, but the AI doesn't quite consider.

That's something that needs balance, and I support it. Can you fix that with just Lua/XML? I don't think so. (not without a small change to the AI to stop trying to attack units/cities where it can't actually do damage).

By the way, that's something else the AI sucks at: adding 'extra' attacks on cities/etc when they do zero damage.

Here's one I discovered recently: the AI won't hook up his water-based Oil deposits. Ever. [...]
It's obviously a bug. [...]

I make fun of my CS ally (and therefore the AI) in my recent England LP on Youtube for exactly this issue. It's a bug, not a mod requirement.

In the same vein, the AI will not use land ranged units (or air units) against embarked units. You cannot possibly consider forcing the player to do the same. Those are bugs, yes, but it's just something the AI doesn't even try to do. That needs fixing, and loopholes will be closed.
 
The AI does attack emarked units with ranged units. I've seen that happen.

I've got hours of video footage showing the opposite. This is in the latest patch.

It, like other issues, seems to have come with the latest patches. (such as AIs almost entirely refusing to upgrade units and the map scripts not aligning with the types)
 
Well, maybe it changed with the last patch, but they at least used to.

it's likely.

Back in the early days of the game, the AI used to mass upgrade every unit as soon as they could. Now they seem to just want to build units from scratch and might upgrade one if they can't do anything else.

the latest patch really messed up the AI and some parts of the game. I'm still hoping that the silence from 2K/Firaxis means that they're trying to solve all of those issues, but are just worried that telling us will set up too much expectations.
 
Back
Top Bottom