Patch notes for 'Fall 2016' update have been released

[RefSteel]

The second is that I want to set up a Civ4 game that basically starts you in the Modern Era after a normally-paced landgrab happens

I would play the heck out this game. It sounds awesome.

Have you tried Adventure 60? Here's a link to the announcement thread! Closing day's obviously long past, but you can still play a shadow game; more reports are always welcome!

(I would happily play civ 4 SGs or a fast paced real time Civ 4 MP game (do they exist anymore?) if there's any appetite for that sort of thing.)

They do if you post a thread and find interested players - I'd say go for it! (Not a sign-up, sadly; I'd love to participate in a Civ 4 SG, but I don't foresee having time to play a turn set in a reasonable amount of time by the time we'd reach the mid-game.)

---

Back on topic: My computer is not Civ-6 compatible, but I'm following along anyway. How much patching and/or modding do you guys think is needed to make the AI interesting? It sounds like diplomacy needs to be more transparent (and, frankly, relevant e.g. to whether war gets declared) - but I've seen a host of other problems described with the AI. It seems from what I've read as if even with Deity-level economic boosts, the AI will lose every battle for most of the game just because it can't handle 1UPT and can't focus much more of its production advantage than the player onto any given tile. What if the corps/army mechanic were extended though: 2 units can be combined into a Phalanx in the classical era, 4 units or 2 Phalanxes can be combined into a Company in the medieval era, and so forth on up through the game? The techs to unlock each of these combinations might also add some interest to dry portions of the tech tree, and some of the later-game hammer costs could be reduced since a 32-unit-equivalent Brigade (or whatever) would function as quite an effective hammer sink....

[Gazglum]

Interesting discussion guys. And I agree with almost everything you said, Ruined Everything. You have not ruined the discussion!

No Civ game is perfect, and the problem is exactly what you say - that the single-player challenge abruptly ends once you pass the AIs, because they can't catch up. So I am very much a 'game abandoner', because the game ceases to exist after that tipping point. In Civ IV, it usually came around the Industrial era, but very occasionally a game could last right up until the finish line. That's one reason I like the Earth map, because it usually allowed for a serious rival on the other side of the world, and even the occasional actual space race.

In Civ VI, the tipping point comes in the Classical Era. The first 80 turns are probably the best any Civ game has been, the barbarian dance, then the tower defense against an AI invasion. I love it. But as soon as you beat back that first push, the game is over. Playing out a full game of Civ VI actually makes me disgusted at myself, because it seems like such a waste of time.

Right now, I don't know what kind of variant could make up for that. I think the truth is that, without serious modifications, Civ VI is not going to be a good game for any kind of single player challenges. I do think it could be a great multiplayer game, when that is finally implemented.

[Cheater Hater]

I feel like the problem can be separated into two main sections:
1. In games with a snowball, it's very easy to stay ahead, and Civ will always be like that since more land=more production (ironically Civ5 is the game where that isn't true, and everyone hates it--I wonder how the AI would be able to compete there if it wasn't crippled by 1UPT)
2. The more sub-systems in a game, the more complicated it is for the AI to understand it, and the number of subsystems increases as you go through the eras--imagine the situation where, say, an AI was extremely good at using air units or a navy; that could be how an AI "came back" against a player who didn't understand them as well

[fluffyflyingpig]

Yeah. It's just too bad the AI doesn't understand any of the subsystems except religion.

[Singaboy]

@Gazglum, I think you summarized my Civ6 experience pretty well. The first 50-80 turns are really nice and exciting and then it turns into a game you play entirely by yourself (as the AI is just a side note).

Over at civfanatics, a modder who came out with a quick AI+ Mod for Civ6 has explained the problems with the AI well (I can't of course verify whether he is right, as I am not a programmer)

This is what he wrote and it is not pretty:

"For those curious, a large part of the AI problems they have is because they decided to do a lot of decision making through what's called a 'behaviortree'. It's basically a collection of types of behaviors (move / attack / settle/ collect units/ upgrade / declare war/plan a units construction/ etc) that are done in some succession. Some of these are assigned to operations, which is basically a team of units doing some behavior together.
The idea may have originally been to make these types of behaviors easy to edit and even moddable. But I think it just ended up flopping horribly. The problem with it is mostly that these behaviortrees are almost completely blind to anything, there's very very few conditionals in there that would allow anything even resembling smart decision making. Units get constantly locked inside of behavior nodes, such as having them move to somewhere, without any way of getting them out of there again.

An example of a conditional that is missing is that there's nothing there that allows you to check whether your settler is on the same tile as a unit. You can tell the behaviortree to form formations, but it only seems to do that if the units are already on the same tile, otherwise itll just pass the 'formation making node' and continue on as if nothing went wrong.

Another example is that when units are locked into an operation, you cant have them switch over to another task. Capturing civilians is done through a different operation/behaviortree than say attacking cities is. So if a unit locked to 'attacking a city' sees an unescorted civilian settler, its gonna run right past it.

And when attacking a city, there is nothing there that allows you to check the health of the enemy city. If humans see that a city is at 5 health, they'll attack it quickly before its too late. The AI can't see it, and will be distracted by anything it would ordinarily be distracted by. Move a unit out of the city a turn before its captured, and its rather likely that all of the enemy units will decide to chase after that unit. You buy a turn, get some city health back, find some reinforcements, and the war is won."

[Gaspar]

I think at this current point in its development, Civ6 should be looked at as two games. The first 70-80t, where the ratcheted barbarian pressure and potential AI invasions make for a somewhat interesting competitive experience, is the first game. One that generally is solvable for the human, but certainly one where mistakes can be costly or even fatal. The second game is the remainder, which should be looked at as some composite of puzzle and strategy game, where humans try and take the various inputs, make smart decisions about city planning and trade route optimization, and the AI provide some very light contextual background. Both games are fun, if you have the right mindset. And I think it is pretty fair to say there's a lot of scope for both parts of the game to improve to such a situation where they will continue to be fun in the future, since there is a lot of room for things like optimization and variant play.

[Gaspar]

Here's a little something I saw on reddit which made me nod along. Not sure if we discussed it in the other thread so thought I'd port it over here...

In fact, the whole district limit system cripples the AI. In Civ IV and Civ V, the AI could basically do OK by constructing every possible building in every city. Obviously, human players who prioritized the important buildings were much more efficient. But with a 100% or so production modifier, the AI could do fine. Sooner or later, it would get around to the important stuff, like libraries.
In Civ VI, the AI cannot implement this strategy because of the district limit. If it builds a holy site and an entertainment complex as a city's first two districts, that city is going to be useless for a long time. In that case the AI literally cannot build the important stuff like campuses.
The district limit makes prioritizing the right districts really important. For a human player, that's fun and interesting. For AI performance, that's disastrous.

Hadn't really thought about it before but this makes a lot of sense for why the AI seems so comically inept at development. There's obviously more to it than that, the above comment about decision trees certainly explains some of the frankly insane behavior I see with AIs and settling - if you read my Adv1 report I saw both Egypt and Japan wander settlers through the wilderness for ridiculous numbers of turns. I'd guess in the current setup AI are basically prioritizing most districts similarly. They plunk down that holy site almost immediately and then the next that comes available, usually a campus or an encampment. Then they just start spamming out the buildings for those districts and so on. I wonder if simply relaxing/removing the district cap from AIs would make a meaningful improvement in how they play.

[Singaboy]

I'd guess in the current setup AI are basically prioritizing most districts similarly. They plunk down that holy site almost immediately and then the next that comes available, usually a campus or an encampment. Then they just start spamming out the buildings for those districts and so on. I wonder if simply relaxing/removing the district cap from AIs would make a meaningful improvement in how they play.

Of course there would be a potential human exploit when taking over those AI cities as they would have more districts than a human would be allowed to build.
Maybe the AI needs to be more scripted. 1 Commercial 2 Industrial 3 Campus

That way they would at least ensure that the AI is more competitive.

[Ruined Everything]

You could get around this by giving the AI massive growth/amenity bonii, so that they could rock size 12-15 cities very, very quickly and effectively have no district constraints.

Even with the AI we have now, there's probably substantial room for quick fixes by throwing bonii at them.

[Singaboy]

If the AI needs such obscene help, why not let them start with pop 10 cities right away with all infrastructure. Not sure, whether I want to play such a game, but hey, at least all the failings of designing a game that has the AI in mind can be forgiven.