Do you want an AI that can make reasoned choices?

[TheMeInTeam]

The upshot is yes, indeed, the computer players (a.k.a. the AI) are trying to achieve one of the victory conditions.

Side note: in rare instances, a sufficiently large Civ 4 AI will switch to "domination mode", there IS logic for it there. You can occasionally see this on high difficulties if you go to negotiate peace between two AIs and see the hover message "we'd rather win the game, thank you very much" .

To what extent this is true varies by Civ game, and even patches/expansions within that title.

For example, the AI didn't have a functional culture victory script in Civ 4 until BTS. After that, it was one of the fastest ways a high-difficulty AI would threaten to win as well. But these sorts of AIs only sort-of "try". Aka in past civ games the AI will pick a victory condition at some point and do things that progress towards it (usually poorly, but at least broadly aligned with victory). However, I have yet to see a Civ AI operate specifically to deny victory to either player or other AI civs...and in fact will sometimes actively throw by intercepting attempts to stop another AI/player winning inadvertently. Its handle on victory vs players' is really glaring in those instances.

Most players would probably not enjoy well-timed AI coalitions designed to stop anybody else from winning, I suspect. Even less so if the AI were truly good at evaluating the game state and executing those coalition attempts. I think Solver is absolutely right about that, and also suspect that a ML AI would probably do it (though likely not against players, who would be too far behind lol).

I have not yet seen evidence that the Civ AI logic effectively pursues victory conditions that require sustained, successful military action.

I think there are a few reasons for this. Top of the list is that it requires the AI to defeat itself militarily, at literally equal skill. Devs haven't really pushed AI tactical capabilities much either, again Solver's almost certainly right that if the AI were really good at it, most of the community would be frustrated/not play.

The other reason is a design inconsistency that has always existed in Civ. Civ offers you many different ways to win...but in reality, none of these victory conditions will work for you if someone rolls up your empire like a blanket and you have no cities. If you can do full conquest of the other Civs, it implies you could win any of the other VCs easily too (though "diplo victory" is more of an inside joke at that point, reminds me of nuking down pop to win "UN victory" in Civ 4). A consistently military dominant nation can not only win reliably, it can deny any other VC to other nations on the way. The mechanics of the game imply that this is a 4x conquest game first and foremost, with empire management/economy secondary.

But that's not the vision the devs had for the game, nor is it the expectation from (most of) the players. Thus I would not expect to see this emphasized in the AI. When you watch how PvP matches usually go, however, they do wind up prioritizing military capability first and foremost, because you can't win if you're conquered, and other humans will do that if they manage it w/o getting knifed themselves.

 

[rgp151]

I think people would play against harder AIs that actually pursue military victory. My favorite 4X games are Master of Magic and Master of Orion (1 & 2). In all of those games the AIs are very aggressive military and will also take out each other. It seems to work just fine. I think those games are definitely more challenging that Civ 6 and also a lot of fun to play. Civ 6 has other things going for it of course, but the AI is certainly a weak point, because in terms of military, its really only a matter of being capable of taking down cities. There is essentially no military strategy required to the game, and very little in terms of tactics.

In both MoO and MoM the military threat from the AI is very real and very persistent. In addition, while you can certainly master the tactics and out maneuver the AI, it does require some experience and level of expertise.

The lack of military threat from the AI influences the whole rest of the game. Indeed it ends up making domination by the player much more advantageous, because the main threat the AI poses is winning via Science and the best counter to that is simply to destroy their cities. If any AI starts to win, al you need to do is attack them basically. Likewise, you don't really need to worry about your own defense, so its easy to amass an army and go offensive with it.

In games like MoM, you absolutely need to keep units, often most of your units, at home on defense. There are definitely cases where you can't pull your troops to attack a city because you have to keep them on defense due to the treat of attack.

Now, in MoM and MoO, conquest is the main mode of victory. You can essentially win diplomatic victory in MoO and you can win something like a Science victory in MoM, but mostly you win by conquest, especially in MoM. But very often you will see on those games that one civ has eliminated another AI civ, and this is often a big threat because it means that civ is getting much stronger. Of course, in those games there is zero penalty for going wide, you simply go as wide as you possibly can. The biggest issue is just the number of troops required to secure it all.

But yeah, I do wish Civ 6 had AI more like MoM.... a game about 20 years older than it....

I think that is the AI were more capable in Civ 6 it could enhance other victory conditions, by favoring defensive play and securing Science or Cultural victory. It could also make diplomacy much more interesting. I could see sort of building a fortress and launching the rockets to hold off an aggressive civ that is difficult to defeat militarily.

But really, in Civ 6 all victory conditions other than Domination are just ways to win without the tedium of needing to run the military. I think in every case, except maybe on very large maps, if you win by anything other than Domination, you could have won by Domination faster, you just don't because you don't feel like putting in the effort. If the AI were harder to defeat military, there would be times when you opted for other victory conditions because Domination was too difficult.

 

[TheMeInTeam]

I think people would play against harder AIs that actually pursue military victory.

No doubt some people would, but this isn't really the selling point for Civ 6, and would be a deviation from what majority of its audience expects.

Regarding tactics: I think people underestimate their importance. They mattered a lot as early as Civ 3, and have ever since. To the point that player vs player wars in Civ 4 looked completely different from SP wars. I've seen SP-minded people in 4 attempt PvP, go in with the old "stack of doom", get their army wiped to collateral initiative, and die in impressively short order. I've also seen people rage quit because groups of guerilla 2 crossbows blitzed their cap before they could whip enough units to react to the unexpected move speed in their territory (and difficulty of actually attacking guerilla 2 xbow on hills), or similarly because it's actually pretty hard to fully garrison all coastal cities against large groups of units on transports (with any captured city getting razed).

Similarly in 6, vision advantage is huge because you can see where units are, know where they can/can't move, and better predict opponent choices. There's also at least enough space for tactical skill difference to allow players to outright kill AI an era ahead in tech, without losing units. AI is bad, yes, but if there were really no significant impact/limited impact from tactics, we wouldn't expect that to be possible.

 

[Archon_Wing]

n my experience with CivIV (on a lower difficulty than you play), the AI pursue the victory conditions, including Culture Victories and Space Victories. I've not seen them push hard to complete a Domination victory, though they do build large armies and vassalize smaller factions.

A domination victory by the AI should not be possible unless the player is severely misplaying because it naturally involves bowling over said player in most maps. So a player never actually sees the AI making said victory, because they probably would have been eliminated or they resign.

However, the AI would achieve other victories like science or culture if you didn't do anything about it even on Prince. In Civ 6, I just don't even see the AI ever having a chance of winning anything besides on Deity, and even there it's just kinda there.

It seems like the AI does not comprehend the new science victory featured in GS and above as they just stop at a certain point.

Part of this is due to feature bloat resulting in much faster win times even if you don't try to win, but even considering that the AI has effectively no plan. I suppose they do cover it up by trying to make you lose even if it means they can't win with early rushes and such, which has long been a thing in the series.

Now there was a point during Gathering Storm where the AI did try to spam a bunch of rock bands and actually looked like a threat, but as of the newest NFP patch, the AI is simply crippled and can't even keep up an army or anything. And sadly this looks like the final state of the game so this discussion is kinda moot. I'm not even looking for an AI that can make reasoned choices when the it just made it past the front gate.

 

[rgp151]

A domination victory by the AI should not be possible unless the player is severely misplaying because it naturally involves bowling over said player in most maps. So a player never actually sees the AI making said victory, because they probably would have been eliminated or they resign.

Yeah, but that's fine. In those cases, you wouldn't see the victory, but you would lose. In Master of Orion it definitely lights a fire under you when you see that some AI civ has eliminated another civilization from the game. The problem with Civ 6 is that the map gets to a pretty static position fairly quickly and it has a lot to do with the starting positions of the leaders. Everyone expands out until they can't peacefully anymore and then things just lock in place. In games like MoM or MoO the map is constantly changing, with some civ/race or another expanding and others contracting.

It would likely be the case that when playing continents, militaristic civs would end up taking over entire continents, that's true. But that's fine really.

Generally, the overall "best way" to win the game IMO, is to conquer early, eliminating one or two entire civs, and then go for Scientific or Cultural victory. This is especially true on continents, where you can just get an entire landmass to yourself. But, the AI never even tries that. In games like MoM and MoO you have to worry about one AI steamrolling the other AIs and getting really powerful, but in Civ 6 its not really even a consideration. Having AIs that will try to conquer other civs would add a lot to diplomacy adn make alliances and such more meaningful. There would be motivation to come to the aid of weaker civs to help them survive against an aggressor, etc.

The irony is that it makes militaristic civ AI opponents a joke, while in the hands of a player the militaristic civs are the most dominant. In games like MoO, the strong civs are strong, whether in the hands of a player or the AI.

 

[Archon_Wing]

It would likely be the case that when playing continents, militaristic civs would end up taking over entire continents, that's true. But that's fine really.

Hah, that's why I never like continents in any of these games. It's decided by just taking over your continent. And sadly AI will never have a chance there because they can't handle water. (Now, that I at least understand).

 

[rgp151]

Hah, that's why I never like continents in any of these games. It's decided by just taking over your continent. And sadly AI will never have a chance there because they can't handle water. (Now, that I at least understand).

Yeah, that's also a big difference between Civ and Master of Magic. In MoM the AI dominates the water so hard its rarely worth contesting it. And also, you can't just send units across the water, you have to load them into boats, and if your boat carrying units get sunk, then oh well, big loss for you. But this also makes guarding the water very important because you can intercept enemy boats, filled with units and sink them before they land with a doom stack. I've definitely played games where it was critical to kill the transport boats because it would not have been possible to kill the units once they unloaded. Too bad you don't see stuff like that in Civ 6. It would definitely force more use of boats. As it is now, you may as well just sent your Musketeers swimming...

 

[Leyrann]

Side note: in rare instances, a sufficiently large Civ 4 AI will switch to "domination mode", there IS logic for it there. You can occasionally see this on high difficulties if you go to negotiate peace between two AIs and see the hover message "we'd rather win the game, thank you very much" .

To what extent this is true varies by Civ game, and even patches/expansions within that title.

Anecdotal evidence, but I remember once playing a game of Civ V (this was probably King or Emperor difficulty) where the AI Napoleon became absolutely huge. Both me (I was also going domination) and him had conquered almost half the map, with some other AIs still surviving in the corners, and this was by the Industrial Era or something. I ended up never playing that game out (not sure why to be honest), but I was expecting an insanely big war for world domination. I think if I'd been playing peacefully (and wouldn't have reacted to his conquering in the way a player would), he'd have won a domination victory before anyone else would've achieved a victory.

The other reason is a design inconsistency that has always existed in Civ. Civ offers you many different ways to win...but in reality, none of these victory conditions will work for you if someone rolls up your empire like a blanket and you have no cities. If you can do full conquest of the other Civs, it implies you could win any of the other VCs easily too (though "diplo victory" is more of an inside joke at that point, reminds me of nuking down pop to win "UN victory" in Civ 4). A consistently military dominant nation can not only win reliably, it can deny any other VC to other nations on the way. The mechanics of the game imply that this is a 4x conquest game first and foremost, with empire management/economy secondary.

But that's not the vision the devs had for the game, nor is it the expectation from (most of) the players. Thus I would not expect to see this emphasized in the AI. When you watch how PvP matches usually go, however, they do wind up prioritizing military capability first and foremost, because you can't win if you're conquered, and other humans will do that if they manage it w/o getting knifed themselves.

That's a really interesting observation, and it makes a lot of sense.

 

[civac]

The other reason is a design inconsistency that has always existed in Civ. Civ offers you many different ways to win...but in reality, none of these victory conditions will work for you if someone rolls up your empire like a blanket and you have no cities. If you can do full conquest of the other Civs, it implies you could win any of the other VCs easily too (though "diplo victory" is more of an inside joke at that point, reminds me of nuking down pop to win "UN victory" in Civ 4). A consistently military dominant nation can not only win reliably, it can deny any other VC to other nations on the way. The mechanics of the game imply that this is a 4x conquest game first and foremost, with empire management/economy secondary.

But that's not the vision the devs had for the game, nor is it the expectation from (most of) the players. Thus I would not expect to see this emphasized in the AI. When you watch how PvP matches usually go, however, they do wind up prioritizing military capability first and foremost, because you can't win if you're conquered, and other humans will do that if they manage it w/o getting knifed themselves.

I disagree with this. At least for slow (Pitboss, Pbem) Civ4 FFA games economy and empire management are more important than tactics. While winning an FFA requires successful conquest it is superior economic management which enables said conquest via a tech and production lead. Moreover, despite what you say, on land maps the stack of doom is the primary tool of conquest. There simply is no other. Raids with guerilla/woodsman units or more commonly cavalry are of course employed. However those are merely tactics. They don't win wars by themselves. It may be different in fast (online) MP games. There, often very small maps combined with a lower standard of play (blazing timer) and rules like city elimination could elevate 2-mover tactics to a higher status.

 

[TheMeInTeam]

I disagree with this. At least for slow (Pitboss, Pbem) Civ4 FFA games economy and empire management are more important than tactics.

Sure, nobody is beating someone of similar or even somewhat inferior tactical ability while an era behind, in any civ game, unless the empire size differential is unrealistically large.

There simply is no other. Raids with guerilla/woodsman units or more commonly cavalry are of course employed. However those are merely tactics. They don't win wars by themselves.

They can, though as you say it's more likely the lower the skill level and faster the lobby/turn timer, and significantly more likely with city elimination.

Main problem for stack of doom (generally) is collateral initiative. To make an offensive work, you need not just a larger stack, but a *much* larger stack, otherwise you get catapults or cannons or w/e slammed in and cleaned up by units that can move faster due to friendly roads. Defensive terrain will somewhat help, if available, but you're a lot better off with era lead where number of siege sacrificed to get high odds of victory becomes much higher. Or if target simply doesn't have the siege count and you know it, of course. Can't do collateral initiative without unit count at least in ballpark of target. And players do have some incentive to cheat on unit production in favor of economy to fish for that era lead in tech, so it's not like this can't happen.

 

[civac]

Yeah you need an enormous advantage in Civ4 to attack someone skilled. There are certain breaking points which make it viable to conquer if you have a decent advantage. Knights are one (if war elephants are banned or absent), rifles another (drafted rifles are a fair bit broken). Cuirassiers or cannons can also enable sucessful attacks in the right circumstances.

I'm curious how the situation is in Civ6. It seem the defender also has a huge advantage with overlapping city fire and loyalty pressure.

 

[Leyrann]

Yeah you need an enormous advantage in Civ4 to attack someone skilled. There are certain breaking points which make it viable to conquer if you have a decent advantage. Knights are one (if war elephants are banned or absent), rifles another (drafted rifles are a fair bit broken). Cuirassiers or cannons can also enable sucessful attacks in the right circumstances.

I'm curious how the situation is in Civ6. It seem the defender also has a huge advantage with overlapping city fire and loyalty pressure.

I can't speak about this in detail, as I don't play multiplayer myself, but I have watched a number of tournament streams. An early game war almost always leads to both players becoming irrelevant compared to the rest in terms of development, and is therefore rarely declared unless the game is a team game or someone is really desperate (or in an amazing position comparatively). Some civs have unique units that are useful to do a push with; I remember seeing a strong Conquistador push in one game. However, most wars are fought in the later stages in the game, usually with a tank push (very occasionally cuirassiers are used). This basically means you want to get armies unlocked and pre-build cuirassiers or even knights, you want to get a great general, the Fascism government (which gives +5 combat strength to all units) and, if possible, a golden age with the right dedication. With the Better Balanced Game mod, Military Academies also become very important because they generate oil (to make the game less coinflippy based on whether or not you happened to have oil in your empire).

Then, when you unlock the technology for tanks, you immediately upgrade all your knights/cuirassiers, and then push forward with your tank armies, supported by usually a trio of artillery armies (artillery, similarly upgraded from earlier siege units, is unlocked one tech earlier). The best counter to this is... tank armies. Yup. It's tank armies or lose, pretty much. Usually this is paired with beelining the bottom tech tree, meaning planes don't come in until a while later, but tank (+artillery) pushes are absolutely devastating. We're talking on the order of conquering a city every 1-2 turns here. Even with your own tank armies it can be tough to defend, depending on other bonuses that may be present. Oh, yeah, and some military engineers can be useful to build railroads, if unlocked.

The only thing that trumps tanks is nukes. If someone gets nukes and has uranium, they're pretty much the winner of the game by default, because a single nuke can destroy most of an army (as in, the attacking force, not a single combined unit) and two can pretty much destroy all of one. Regarding this, also keep in mind that players will usually want their units to be as close together as possible, as that gives them flanking and support combat bonuses, which can be very strong (to the point where, for naval combat, players will often build scouts when defending - these can be placed 'beneath' ships and give an extra +2 combat strength to adjacent units when those units are defending), which of course makes them more vulnerable to the area of effect annihilation that nukes bring.

I'd say city fire doesn't do enough damage to be relevant, while loyalty pressure doesn't work fast enough to avoid the attacker receiving a loyalty foothold.

Also, if you're wondering about Modern Armor (tank upgrade), the extra combat strength quite simply isn't worth the increased cost and research time. As I understand it, players sometimes even avoid researching them because they'd rather use the somewhat weaker, but significantly cheaper tanks. Also, if you can get to Modern Armor, you can probably get to nukes as well.

 

[Solver]

However, I have yet to see a Civ AI operate specifically to deny victory to either player or other AI civs...and in fact will sometimes actively throw by intercepting attempts to stop another AI/player winning inadvertently. Its handle on victory vs players' is really glaring in those instances.

Most players would probably not enjoy well-timed AI coalitions designed to stop anybody else from winning, I suspect.

So this is a very interesting feature because it's obviously polarizing. We have exactly this as an optional toggle in Old World - the AI will aim to deny you victory. It keeps track of your victory progress and gets more aggressive as you get closer, until eventually the "prevent victory" goal overrides several aspects of normal decision making. Most prominently, the AIs become willing to declare war even with an army they'd normally deem too weak, and they will refuse any peace proposals.

As it should be expected, the option is not very popular, but the people who use it seem to really love it.

 

[Leyrann]

So this is a very interesting feature because it's obviously polarizing. We have exactly this as an optional toggle in Old World - the AI will aim to deny you victory. It keeps track of your victory progress and gets more aggressive as you get closer, until eventually the "prevent victory" goal overrides several aspects of normal decision making. Most prominently, the AIs become willing to declare war even with an army they'd normally deem too weak, and they will refuse any peace proposals.

As it should be expected, the option is not very popular, but the people who use it seem to really love it.

I really like the idea to make it a toggle. I'm hoping Firaxis will do something similar for Civ VII. Whether or not I'd actually use it would probably depend on the quality of the AI in general and my current mood when starting the game, but sometimes, I'd really love it.

Random thought: What about being able to turn it on or off during the game, depending on whether you want to have the AI roleplay or give you the best challenge it possibly can? Or is that too likely to mess things up?

 

[rgp151]

So this is a very interesting feature because it's obviously polarizing. We have exactly this as an optional toggle in Old World - the AI will aim to deny you victory. It keeps track of your victory progress and gets more aggressive as you get closer, until eventually the "prevent victory" goal overrides several aspects of normal decision making. Most prominently, the AIs become willing to declare war even with an army they'd normally deem too weak, and they will refuse any peace proposals.

As it should be expected, the option is not very popular, but the people who use it seem to really love it.

This is exactly how Master of Orion and Master of Magic work. The AIs will ally against you and certainly work to deny you victory. I really don't see how it could be any other way. It always seemed odd to me that civs would ally with you when you are about to win the game. Doesn't really make sense.

 

[TheMeInTeam]

Yeah you need an enormous advantage in Civ4 to attack someone skilled. There are certain breaking points which make it viable to conquer if you have a decent advantage. Knights are one (if war elephants are banned or absent), rifles another (drafted rifles are a fair bit broken). Cuirassiers or cannons can also enable sucessful attacks in the right circumstances.

I'm curious how the situation is in Civ6. It seem the defender also has a huge advantage with overlapping city fire and loyalty pressure.

I haven't done any competitive PvP in Civ 6. I did it so long ago in Civ 4 that the community has long moved past me in skill too, though it seems the units most commonly used to push an advantage haven't changed much in 4 Other than knights those are the common SP options too.

I would have to imagine Civ 6 to favor defender a lot in MP. You can make completely ridiculous plays in terms of troop count vs AI, to the point of being functionally immune to attack at tech parity with not too many defenders. Cities function as a high strength unit with enormous health, and until late game they match or outrange opposing units/do decent damage. Don't forget that the encampment can also fire, not just the city. Some spots can take 3 hits from nothing but cities, and all of those can house ranged units to pile on even more (while being immune to damage in return until the cities/encampments are captured).

Another underrated advantage to Civ 6 defender is vision; if attacker is pushing into borders defender can see a larger % of tiles that are relevant than attacker, who might miss that fast movers can swap in/out and concentrate damage on a few of their units/kill them. For much of the game (including most that matters) it isn't always trivial to scout interior of others' territory...nothing that can move around closed borders to see like a civ 4 spy for example.

 

[aieeegrunt]

I haven't done any competitive PvP in Civ 6. I did it so long ago in Civ 4 that the community has long moved past me in skill too, though it seems the units most commonly used to push an advantage haven't changed much in 4 Other than knights those are the common SP options too.

I would have to imagine Civ 6 to favor defender a lot in MP. You can make completely ridiculous plays in terms of troop count vs AI, to the point of being functionally immune to attack at tech parity with not too many defenders. Cities function as a high strength unit with enormous health, and until late game they match or outrange opposing units/do decent damage. Don't forget that the encampment can also fire, not just the city. Some spots can take 3 hits from nothing but cities, and all of those can house ranged units to pile on even more (while being immune to damage in return until the cities/encampments are captured).

Another underrated advantage to Civ 6 defender is vision; if attacker is pushing into borders defender can see a larger % of tiles that are relevant than attacker, who might miss that fast movers can swap in/out and concentrate damage on a few of their units/kill them. For much of the game (including most that matters) it isn't always trivial to scout interior of others' territory...nothing that can move around closed borders to see like a civ 4 spy for example.

Play Gaul, have walls, oppidum and encampments in your border cities, have one good melee and one good ranged unit and you basically don’t need an army

 

[TheMeInTeam]

Really as any nation, if you have a great general + fortified units, the AI basically can't kill them if it can't concentrate damage. Same deal if these units are adjacent to cities; contemporary units basically can't capture a city if attacking from only 1-2 spots, even if they have a means to deal with the walls...something AI does not consistently manage anyway.

 

[Leyrann]

I would have to imagine Civ 6 to favor defender a lot in MP. You can make completely ridiculous plays in terms of troop count vs AI, to the point of being functionally immune to attack at tech parity with not too many defenders. Cities function as a high strength unit with enormous health, and until late game they match or outrange opposing units/do decent damage. Don't forget that the encampment can also fire, not just the city. Some spots can take 3 hits from nothing but cities, and all of those can house ranged units to pile on even more (while being immune to damage in return until the cities/encampments are captured).

The issue here is that, in the late game, it is trivial to reduce a city to zero health in one or two turns using three artillery armies, and then a tank army to conquer it. As tanks don't suffer zone of control, encampments aren't actually of that much use, as those ranged attacks won't do more than dent a few units. On top of that I'm not sure if city/encampment combat strength is actually boosted by corps/army, but it might be. If it isn't, it practically makes their attacks irrelevant no matter what.

 

[rgp151]

Cities just seem defensively powerful because the AI is so inept at attacking them. Against a real opponent they aren't that strong, which is how it should be, its just that the AI should be better at attacking.