[R&F] When will we get a AI overhaul mod?

[ashendashin]

DLL changes are also nothing but band-aids, just stronger ones. It's not like you're suddenly magically free from all resource limitations; you're making an artificial distinction where there is no actual distinction.

Really now? Rewriting the source code itself is a band-aid? Adding functionality in areas that never had any and even entirely new mechanics for the AI to work with is somehow the same as being restricted to the single way the AI can function?
There is a very real distinction. The source code is barebones and still manages to fail or not even exist in many areas(I can say that with certainty for Civ5, at least. Doesn't seem like a stretch here). Adjusting what currently exists is a band-aid over gaps that won't close.

 

[Ryika]

Really now? Rewriting the source code itself is a band-aid? Adding functionality in areas that never had any and even entirely new mechanics for the AI to work with is somehow the same as being restricted to the single way the AI can function?

Yeah, obviously it's still a band-aid. Even in Vox Populi, the AI still acts silly when you look at the overall picture more often than not. Even with the tons of changes they made, it's still an AI that's limited by what you can do within the framework of the game, without killing performance. Those are just the limitations of a script-based AI in a heavily randomized environment.

There is a very real distinction. The source code is barebones and still manages to fail or not even exist in many areas(I can say that with certainty for Civ5, at least. Doesn't seem like a stretch here). Adjusting what currently exists is a band-aid over gaps that won't close.

There's simply no way to fully fix the AI, period. Obviously there is a huge gap between the performance changes of a basic, flavor-based AI mod, and something like Vox Populi, but both types of mods do not "fix" the AI. You can't on the one hand claim that flavors are a band-aid and on the other hand claim that a DLL-based mod "fixes" the AI when the end-result is still an AI that barely functions on a grand-scale tactical level.

Either both "fix" the AI on different scopes, or one put band-aids on the Ai in different scopes. Your claim that one is a band-aid and the other one is a "fix" is still an artificial distinction.

 

[ashendashin]

Yeah, obviously it's still a band-aid. Even in Vox Populi, the AI still acts silly when you look at the overall picture more often than not. Even with the tons of changes they made, it's still an AI that's limited by what you can do within the framework of the game, without killing performance. Those are just the limitations of a script-based AI in a heavily randomized environment.


There's simply no way to fully fix the AI, period. Obviously there is a huge gap between the performance changes of a basic, flavor-based AI mod, and something like Vox Populi, but both types of mods do not "fix" the AI. You can't on the one hand claim that flavors are a band-aid and on the other hand claim that a DLL-based mod "fixes" the AI when the end-result is still an AI that barely functions on a grand-scale tactical level.

Either both "fix" the AI on different scopes, or one put band-aids on the Ai in different scopes. Your claim that one is a band-aid and the other one is a "fix" is still an artificial distinction.

Oh boy, the "but it's not real AI" argument on a more intricate post.
Look, when it comes down to it the AI is just a band-aid on the fact that we want a single-player game with opponents. Designing a game entirely to play on the computer's strength wouldn't be fun. That the AI can't learn or plan is separate from actually using game mechanics properly. For a video game, AI is part of the design (or at least it should be). Within the design of Civ, adjusting flavors is a band-aid compared to actually making the AI capable of playing the game with minimal boosts. I don't see why I have to clarify that this is all relative to the game. Making the AI play it exactly as a human player would be way beyond the scope of fixing.

 

[Trav'ling Canuck]

Making the AI play it exactly as a human player would be way beyond the scope of fixing.

Personally, what I want from an AI is three-fold:

1. Simulated behaviour that makes it seem like I'm taking on leaders with distinctive personalities. This primarily shows up in their diplomacy interaction. How I maintain good relations with Teddy should feel different from how I maintain good relations with Pericles. Ideally, how Teddy responds to the world around him should differ from how Pericles does. While I like the leader agendas overall, on the latter point I feel like Civ 6 has moved backwards from earlier iterations of Civ. Civ 1's Gandhi was more likely to nuke you, Civ 1's Genghis was more likely to attack you, etc. Now it feels like Teddy's as likely to cut down trees as anyone else, even if he doesn't like you doing so, Pericles is as likely to attack a City State, everyone's as likely to attack you as everyone else. Perhaps that isn't true, but that's the sense I have as a player.

2. A "pace car" that sets a realistic challenge at the higher difficulty levels for when I need to win or else I will lose. If playing Prince, that should be a slow pace car with plenty of time for playing around with the game. If playing Deity, I should need to be very efficient/be forced to actively interfere with the AI's plans or else the AI will achieve it's victory before I can achieve mine. Civ 6's AI right now meanders to victory quite slowly. Whether that's because of the production priorities it has or an inability to use key systems (like Builder charges to speed up space projects, theming to boost Tourism numbers, etc.) I'm not sure.

3. A "speed bump" that interferes in my plans if I don't allocate resources to maintaining good relations/defending myself. That requires a certain level tactical ability that I actually think Civ 6's AI is pretty good at for military units, except for air force. Religiously it's hampered by coding that sends waves of Missionaries that eliminate more of its Religion than it spreads. The overall impact of the AI on the player in Civ 6, however, is uneven and often non existent for long periods of time, even if you aren't paying any attention to their agendas, etc.

 

[Ryika]

Oh boy, the "but it's not real AI" argument on a more intricate post.

Not at all.

Look, when it comes down to it the AI is just a band-aid on the fact that we want a single-player game with opponents. Designing a game entirely to play on the computer's strength wouldn't be fun.

No, but there's always a way to skew things into one direction or the other. To stick with Civ 5, that game is VERY much geared to be enjoyable by a human player, with nearly no regards for the AI. One of the reasons for that is that the strong strategies mostly rely on putting together a set of puzzle pieces that form a strong strategy. The AI simply can't do that, it's designed to "explode onto the map". A part of the problem here is not with the AI itself, but rather with game balance. Synergy and minimalism are way too strong when compared to strategies that dominate the map. (<- This is again for Civ V, I know that has changed quite a bit in Civ VI)

That the AI can't learn or plan is separate from actually using game mechanics properly. For a video game, AI is part of the design (or at least it should be). Within the design of Civ, adjusting flavors is a band-aid compared to actually making the AI capable of playing the game with minimal boosts. I don't see why I have to clarify that this is all relative to the game. Making the AI play it exactly as a human player would be way beyond the scope of fixing.

I don't think making an AI play like a human player is even a goal that should be aimed for. People who want that are better off playing against actual humans.

But just getting an AI to a level where it can play without large modifiers is not "fixing" an AI. What the AI is supposed to do in my opinion, is to provide a challenge, to try to mess with the strategies of other players for their own benefit, to force you to pay attention to what is going on around you, and to play in a way that seems immersive without doing thing that make me go "Did it just really do this?" - or at least doing these things extremely rarely. If an AI is actually competent at doing all of these, that's what I would be willing to consider "fixed".

Flavor-based mods obviously don't achieve that status, and neither does DLL modding.

 

[Gedemon]

DLL changes are also nothing but band-aids, just stronger ones. It's not like you're suddenly magically free from all resource limitations; you're making an artificial distinction where there is no actual distinction.

Well, you've just taken a thermonuclear weapon and said that it's nothing but an explosive, like a firecracker, just a stronger one.

Because, yes, that's the scale of the difference between trying to improve the AI now and doing so with access to the source code.

 

[Ryika]

Well, you've just taken a thermonuclear weapon and said that it's nothing but an explosive, like a firecracker, just a stronger one.

Yeah, I think you're overplaying DLL modding a bit there. Saying that it's like comparing a firecracker to TNT or something like that might come close though.

But saying that both are just explosives is entirely reasonable as a response to a person who claims that TNT is basically a laser rifle.

 

[ashendashin]

But just getting an AI to a level where it can play without large modifiers is not "fixing" an AI. What the AI is supposed to do in my opinion, is to provide a challenge, to try to mess with the strategies of other players for their own benefit, to force you to pay attention to what is going on around you, and to play in a way that seems immersive without doing thing that make me go "Did it just really do this?" - or at least doing these things extremely rarely. If an AI is actually competent at doing all of these, that's what I would be willing to consider "fixed".

Flavor-based mods obviously don't achieve that status, and neither does DLL modding.

...Have you played VP at all? It does all of that. Maybe not fully "fixed" as you define it as it is of course not perfect, but it keeps you on your toes if you're playing the right difficulty.
Keep in mind that that the full VP package is closer to a new game than just an AI overhaul. I've been focusing purely on the AI aspect because honestly the vanilla AI is so bad that I don't understand why anyone is fine with it.

1. Simulated behaviour that makes it seem like I'm taking on leaders with distinctive personalities. This primarily shows up in their diplomacy interaction. How I maintain good relations with Teddy should feel different from how I maintain good relations with Pericles. Ideally, how Teddy responds to the world around him should differ from how Pericles does. While I like the leader agendas overall, on the latter point I feel like Civ 6 has moved backwards from earlier iterations of Civ. Civ 1's Gandhi was more likely to nuke you, Civ 1's Genghis was more likely to attack you, etc. Now it feels like Teddy's as likely to cut down trees as anyone else, even if he doesn't like you doing so, Pericles is as likely to attack a City State, everyone's as likely to attack you as everyone else. Perhaps that isn't true, but that's the sense I have as a player.

That'd be a different game design, I think. Since it's both leader and civ going through the eras there's gotta be malleable abilities to fit the theme. I think that'd be more interesting with a dynamic leadership Civ design where either the actual civ leader changes or new intricate court systems were put in place so it's more about dealing with the entire nation's leadership than one person that has the same agenda from clubs to nukes. Could be cool to go through different channels for trade deals and such.

 

[Ryika]

...Have you played VP at all? It does all of that. Maybe not fully "fixed" as you define it as it is of course not perfect, but it keeps you on your toes if you're playing the right difficulty.
Keep in mind that that the full VP package is closer to a new game than just an AI overhaul. I've been focusing purely on the AI aspect because honestly the vanilla AI is so bad that I don't understand why anyone is fine with it.

Yeah, I've played it quite a bit. It's certainly a staggering difference, but does it make the AI feel like a natural part of the world? No, not quite.

But yeah, I very much agree with this...

Maybe not fully "fixed" as you define it as it is of course not perfect, but it keeps you on your toes if you're playing the right difficulty.

...but in saying that you very much agree with what I said in post #20, it's just a stronger band-aid- Much stronger of course, and much closer to "fixing" the AI than you could get with flavors alone, but there is no DLL mod that has "fixed" Civ 5 AI.

I feel like we're mostly talking about semantics though. It seems that we both agree that DLL modding can do much more to improve the AI than anything that is possible without it, and afaict the rest is just us discussing about categorizations and the meaning of words.

 

[ashendashin]

I feel like we're mostly talking about semantics though. It seems that we both agree that DLL modding can do much more to improve the AI than anything that is possible without it, and afaict the rest is just us discussing about categorizations and the meaning of words.

That's...the thing. DLL modding compared to adjusting flavors is like stitching, cleaning, and bandaging a wound compared to slapping on a band-aid (I mean I guess that can be stretched to a strong band-aid depending on how you look at it). Imperfections are inevitable but Civ6 clearly needs a lot more than a band-aid. It's not like I've been calling anything perfect. The point is to make it work, and pure difficulty isn't what I'm looking for.

 

[Leyrann]

Not about to get into that again with you. Strictly speaking this thread is about getting the AI fixed. Adjusting flavors is akin to putting a band-aid on it.
There's a long partial list Siesta Guru made about the AIs faults if you wanna see if any bits were crossed off yet. I don't give enough attention to Civ6 to go beyond my observations in the community.

It's the best we have so stop complaining about it. Take it or leave it.

And, no matter what way you put it, RF Flavor Deity DOES improve the AI significantly. Band-aid, cheating or whatever, it improves AI performance and I don't feel cheated in the process.

 

[tzu]

Thank you both, Ryika & ashendashin, for working out, that we need the DLL source code.

a thermonuclear weapon and [...] a firecracker, just a stronger one.
Because, yes, that's the scale of the difference between trying to improve the AI now and doing so with access to the source code.

The performance difference between the execution of compiled code and that running in an interpreter is one thing.

The other is, that if you have no control over the DLL, there is interference between whatever you code fresh via LUA and the functionality which is already in the DLL. Example:

[...]
Unfortunately, the way operations are executed leaves some room for the inbuild AI to make decisions. These operations could be seen as 'guidelines' that I can now give the AI. But since it doesn't listen perfectly, it causes some issues. For example, if I tell the AI to move a unit one tile using these operations, the AI can decide after that to move its unit back to the tile it came from. Similarly, I can give an 'attack' guideline, but it might still decide to attack another unit, or might be too scared to attack.
It's also impossible (not sure yet?) for me to order moves in any way, which makes micro unpredictable. A common ingame scenario might be to have these units: -RM-E where R is a ranged unit of mine, M is a melee unit of mine, E is an enemy, and '-' are empty spots. It tends to be wise to first move the melee unit to the empty tile, then move the ranged unit and shoot, and then attack with the melee unit.
But unfortunately, since I can't order things as I want them to, it might up trying to walk with the ranged unit first, and since the intended tile is blocked, the inbuild AI may then decide to actually move it somewhere else.
So instead of the intended result with a unit kill: --RM-
It may end up resulting in this instead: R--ME in which case I not only missed out on a kill, but leave my melee unit in a place it might die and my ranged unit useless.
[...]

 

[Aristos]

Yeah, I think you're overplaying DLL modding a bit there.

That right there seems to expose your lack of understanding. Are you being serious? The entire engine runs in the dll; the code for what to do with all those flavor values is right there. The old IT adage is valid as ever: GIGO. Only in this case, the garbage may well be inside the dll, and not the input per se.

Have you ever at least skimmed over the dll source code for vanilla civ 5? Have you ever compared it to some of the dll modders work, or try to do some of it yourself? "Little" things like the dreaded "archers cannot shoot and move" from vanilla civ 5 (which in reality was "any ranged unit cannot move and shoot")? In fact, that horrible bug was in the end a complete absence of code in the dll for such a vital, basic ranged functionality... try to solve that with flavors. You'll earn a Nobel Prize.

 

[ashendashin]

It's the best we have so stop complaining about it. Take it or leave it.

Or, you know, I could keep complaining anyways. I like Civ's concepts and I don't want to be limited to Civ5's version of VP forever as one definite upgrade Civ6 got was being made 64 bit. I see no reason whatsoever to be happy with being far more limited compared to previous games. They even claimed that this would be the most moddable Civ ever. Might as well tell all negative reviewers to shut up and go away. That doesn't make sense.

 

[Ryika]

That right there seems to expose your lack of understanding. Are you being serious? The entire engine runs in the dll; the code for what to do with all those flavor values is right there. The old IT adage is valid as ever: GIGO. Only in this case, the garbage may well be inside the dll, and not the input per se.

Have you ever at least skimmed over the dll source code for vanilla civ 5? Have you ever compared it to some of the dll modders work, or try to do some of it yourself? "Little" things like the dreaded "archers cannot shoot and move" from vanilla civ 5 (which in reality was "any ranged unit cannot move and shoot")? In fact, that horrible bug was in the end a complete absence of code in the dll for such a vital, basic ranged functionality... try to solve that with flavors. You'll earn a Nobel Prize.

Can you please stop this stupid crap? I already said around 5 times now that I'm well aware that there's a HUGE difference between a flavor-based mod, and a dll-mod, stop pretending like I didn't. My point is that even with dll modding the AI still does plenty of stupid crap, and that I do not consider it "fixed" even in Vox Populi, and that saying that it is "fixed" in Vox Populi but not "fixed" with a flavor mod, is an artificial distinction.

If you can say it's "fixed" because it does fewer stupid things with a dll mod, even though it still does plenty of stupid things that even a bad player would not do, then a person who has a lower standard with what they're looking for can also say it's "fixed" with a flavor mod. Both do not actually "fix" the AI.

If you are willing to consider Vox Populi's AI "fixed" even though it still willingly sends its armies into very obvious dead-zone bottlenecks to be shot down one-by-one, then a guy who's just looking for an AI that does better economically and gets exactly that, can also call that AI "fixed". Those are personal declarations that do not actually translate into a truth claim, because in both cases, the AI still isn't "fixed".

 

[Gedemon]

Can you please stop this stupid crap? I already said around 5 times now that I'm well aware that there's a HUGE difference between a flavor-based mod, and a dll-mod, stop pretending like I didn't. My point is that even with dll modding the AI still does plenty of stupid crap, and that I do not consider it "fixed" even in Vox Populi, and that saying that it is "fixed" in Vox Populi but not "fixed" with a flavor mod, is an artificial distinction.

If you can say it's "fixed" because it does fewer stupid things with a dll mod, even though it still does plenty of stupid things that even a bad player would not do, then a person who has a lower standard with what they're looking for can also say it's "fixed" with a flavor mod. Both do not actually "fix" the AI.

If you are willing to consider Vox Populi's AI "fixed" even though it still willingly sends its armies into very obvious dead-zone bottlenecks to be shot down one-by-one, then a guy who's just looking for an AI that does better economically and gets exactly that, can also call that AI "fixed". Those are personal declarations that do not actually translate into a truth claim, because in both cases, the AI still isn't "fixed".

It's not because one DLL mod doesn't "fix" it in your eyes that it can't be "fixed" by a DLL mod.

I know you like to play with word, and of course we can keep talking of the definition of "fixed" until the release of Civilization XX in 2074, but "fixed" in any conversation about mods means "doing the best we can do using the current technology and hardware".

A mod is not perfect ? well, try to do it yourself, the tools are here, because you know what, you (or anyone) may do better with enough (free) work.

But anyway civ5 and civ6 are very different beasts when it comes to modding, I'm not going to list the pro and con here, it's not the subject, but in relation to the AI, the DLL would make a bigger difference with civ6. My previous comparaison with a thermonuclear device is not exaggerated, especially if we start mentioning the tactical AI ("obvious dead-zone bottlenecks to be shot down one-by-one")

Because with civ6 we can't do anything with flavors and Lua for the tactical AI while we will be able to do everything allowed (using the current technology and hardware for a turn-based game) with the DLL.

In civ5 we could at least control AI units with Lua, allowing "real" band-aids without the DLL. My WWII mod used Lua only to "fix" the air AI long before Firaxis started to improve it in a patch.

In civ6, it's impossible. Period. If it starts bleeding somewhere (like when it was dancing with units around cities it could take in one turn), the only thing you can do is to watch it die slowly or wait for a Firaxis patch, no modders band-aid allowed there.

 

[Ryika]

I know you like to play with word, and of course we can keep talking of the definition of "fixed" until the release of Civilization XX in 2074, but "fixed" in any conversation about mods means "doing the best we can do using the current technology and hardware".

That's again a very specific definition that seems designed only to make an artificial distinction. Why not "doing the best we can do with the tools available to us right now"? Because then a well-rounded flavor mod would also be considered "fixed"?

A mod is not perfect ? well, try to do it yourself, the tools are here, because you know what, you (or anyone) may do better with enough (free) work.

Nah, I've enough other things to do, and little to no experience with actual programming. I'm sure the guys who work on VP have done a much better job than I ever could, and the overall difference between the normal game AI and VP is staggering; but that does not change that the AI still shows some pretty clear flaws.

This passive-aggressive "If you don't like it, then do it better!"-tone is also entirely uncalled for. I've never claimed that I could do it better, I've not claimed that VP does a bad job, and I've said since the beginning of the thread that I think the AI-work in VP especially is amazing, and that I enjoyed my time playing it. All I'm arguing against is this nonsensical idea that the AI in VP can be considered "fixed".

That's me disagreeing with your definition, not me bashing VP.

 

[Gedemon]

That's again a very specific definition that seems designed only to make an artificial distinction. Why not "doing the best we can do with the tools available to us right now"? Because then a well-rounded flavor mod would also be considered "fixed"?

I said technology and hardware, I didn't say software or "tools" .

The software is the limitation ATM, give us the source and the limit becomes the hardware.

Nah, I've enough other things to do, and little to no experience with actual programming. I'm sure the guys who work on VP have done a much better job than I ever could, and the overall difference between the normal game AI and VP is staggering; but that does not change that the AI still shows some pretty clear flaws.

This passive-aggressive "If you don't like it, then do it better!"-tone is also entirely uncalled for. I've never claimed that I could do it better, I've not claimed that VP does a bad job, and I've said since the beginning of the thread that I think the AI-work in VP especially is amazing, and that I enjoyed my time playing it. All I'm arguing against is this nonsensical idea that the AI in VP can be considered "fixed".

That's me disagreeing with your definition, not me bashing VP.

It's not about liking it or not, it's about doing better.

You're taking a mod as an example, dont try to run away from that example when I say that it could be made even better.

 

[Ryika]

I said technology and hardware, I didn't say software or "tools" .

The software is the limitation ATM, give us the source and the limit becomes the hardware.

Yeah, and my question is, why that definition? There's no reason for it other than that you want it to be the definition.

I mean, where else would that definition work? "Well, your TV still has some real problems with displaying the signal correctly, but at our current technological level, this image that looks alright 70% of the time is the best we can do, so as far as I'm concerned, your TV is 'fixed' now."

That's not at all a sensible definition of something being "fixed" in my book. The TV is clearly not "fixed", it has been band-aided as much as possible.

But again, you're free to disagree with my definition.

It's not about liking it or not, it's about doing better.

You're taking a mod as an example, dont try to run away from that example when I say that it could be made even better.

I'm taking the, to my knowledge, most advanced AI as an example.

Surely it only strengthens my argument that a DLL mod does not actually "fix" the AI (as requested in the OP), if you say that even after... how many years now? ...of having access to the DLL, the best mod out there still doesn't provide an AI that is free from showing major flaws in some situations and you're arguing that it could be done even better.

 

[tzu]

1. Fundamental AI issues won't be 'fixed' during the civ6 lifecycle. Fortunately we just need an AI, which plays "pretty darn good".
2. Without access to the DLL source code several types of modifications are impossible. Many/most of the possible modifications are comparably inefficient relative to the invested time of those, who volunteer to develop the improvements.
3. With the DLL source code (in principle) every modification is possible. Not just workarounds & band-aids.