Will they ever fix broken AI for 1UPT, Civ VI ?

[jokii]

When next Civ 6 comes close I will have only one question for devs... AI...? 1UPT....?

Or maybe few more questions like... are you sure AI ? ....1UPT ?....are you relly, relly sure.... AI ? ....1UPT ?..... AI?

I don't care about the damn graphic and leaders, and other crap...
Is it to much to ask for some ( god damn ) competition in a strategy game????

Moderator Action: Please do NOT over use bold text. This is considered shouting.
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889

 

[Gamewizard]

You have to start somewhere, Rome wasn't built in a day, or 4 years for that matter. Firaxis is still new to the 1UPT game. Things are certainly better than they were day 1 of Civ V. They are well aware that the AI is a weak-point in their otherwise stellar games. I certainly expect it to improve going forward, just gotta be a little patient...or buy lots of Firaxis stock so you can make the business decisions. Would your AI heavy budget approach make more or less money than they make now?

 

[Drowsy Emperor]

Its not fixable. The reduced scale of the maps combined with the increased unit coordination needed to make warfare work in 1UPT ensures that the AI will never be even remotely competent.

 

[nokmirt]

It needs to become fixable...otherwise perhaps it's time to stop making civ games altogether.

 

[Kozmos]

Its not fixable. The reduced scale of the maps combined with the increased unit coordination needed to make warfare work in 1UPT ensures that the AI will never be even remotely competent.

Well it's not impossible, but its an effort that will never be expended by devs to make it work. They could at least fix the tendency to move every single unit every turn contributing to those annoying turn times, but I wouldn't even bet on that either.

 

[cazaderonus]

We can only hope they focus on AI for CIV VI....

For BE, we're screwed

 

[AriochIV]

I'm certain it's possible, but I'm not sure the strategy game industry has the right people to make it work. It seems like the guys writing the AI for Civilization are game designers first and programmers second. Perhaps Firaxis could stand to staff up with some genuine computer scientists.

 

[Drowsy Emperor]

No we could just dump 1UPT and go back to what works. Or inflate map size to make it work, but then we'd have to deal with civ maps three times the size of the typical Civ IV map.

And that would just give the units room to maneuver, there's no guarantee you could program them to actually do it.

Endless Legend tried to get around this by separating war on a different hex screen but it didn't work there either. Too few options to make for involving combat. The AI's entire strategy is targeting the units that his units are strong against. Failing that, its just a straight up brawl.

 

[AriochIV]

I know not everyone shares this opinion, but I find 1UPT with questionable AI is still more fun than smashing stacks of doom together.

 

[ThERat]

As most design decision were a consequence of 1upt and how to avoid too many units, I'd say it's a complete and utter failure.

No need to talk about the AI's inability to use 1upt. Civ4 was so much better in this regard. Civ6 should take the best features of all the previous civs and improve on them. 1upt is the worst feature of them all.

 

[Drowsy Emperor]

I'm certain it's possible, but I'm not sure the strategy game industry has the right people to make it work. It seems like the guys writing the AI for Civilization are game designers first and programmers second. Perhaps Firaxis could stand to staff up with some genuine computer scientists.

Which strategy game had dynamic combat of this type that depended on precise unit placement and worked?

I say dynamic because they took the whole concept from Panzer General II where the maps are large and the AI is essentially passive.

You'd have to program the AI to make a proper army, appropriate to the task at hand and then use it on maps that bottleneck most of the time and that can be hard for a human player to navigate.

You could easily program it to be competent for a particular scenario with fixed troops and unit placement but getting it to come up with strategy on the fly... good luck with that.

 

[cazaderonus]

I know not everyone shares this opinion, but I find 1UPT with questionable AI is still more fun than smashing stacks of doom together.

This. Massive stack of dooms in CIV 4 i found so stupid that i stopped playing.

Real issue with 1UPT is the city\unit\map scale. If you had maps 2 times bigger, and cities with bigger border limits and reachable tiles, things would be more interesting.

 

[AriochIV]

Which strategy game had dynamic combat of this type that depended on precise unit placement and worked?

None that I'm aware of, ever. So?

It's never been done, so it can never be done? That doesn't sound like a proper credo for anything computer related. We'd still be playing Pong.

We have algorithms that can recognize voice and parse natural language, that can run on a smartphone. There's no way that making a decent game AI is any harder than that. I'm just saying that game AI's don't have as many PhD's and guys with a dozen patents working on them. And maybe they should.

 

[ThERat]

Civ is an incredibly tough task for any programmer to deal with when it comes to decent AI behaviour. The task was pretty well solved in Civ4, especially with the help of the modding community here, Dale's Better AI comes to mind.

1upt is an insurmountable task to do with the countless variations in maps, situations etc. I doubt it will be done for a very long time to come.

 

[isau]

We have algorithms that can recognize voice and parse natural language, that can run on a smartphone. There's no way that making a decent game AI is any harder than that.

Sorry, but no. AI in a complex game is MUCH harder than parsing language, in part because language has thousands of people plugging at it, and in part because complex AI is far, far more difficult than most people appreciate.

 

[Wizard-Bob]

What of a hybrid approach between 1UPT and stacks? Maybe allow 1 Melee + 1 Ranged + 1 Siege + 1 Air per Hex? It would streamline movement a bit, but would that allow for a stronger or weaker AI overall?

 

[Drowsy Emperor]

I'm all for it, but in the span of typical game development I doubt anyone will make an AI that works for that particular need.

The trick is to have game framework that is conductive to making a seemingly smart AI. I say seemingly because all AI is eventually revealed to be limited and with holes that can be exploited. But a stack composition is easier to program than 1UPT units maneuvering on the world map.

I would rather see Civ VI embrace a new framework entirely that may not be either of these two things, because I feel that the Civ V framework has skewered the game. Even if you fixed combat AI, you'd still have to deal with the reduced scale, the tight maps, the rates of production slowed to a crawl to keep the map from filling up etc. etc. Everything about the game was changed to accommodate a different combat model, for the sake of 1UPT. Was it worth it?
Is it worth fixing at any cost?

I'm not biased either way, but the stacks worked in general, for 4 games, many players and the AI and 1UPT doesn't seem to be working as it should 2 games and 2 expansions down the road.

 

[Drowsy Emperor]

@Wizard: limited stacks, "armies" in other words occurred to me as well. But you'd still have to balance production so that the player is unable to make a train of these armies and just throw them against the enemy.

Then there's the whole somewhat boring aspect of it. If you force me to compose my armies in a particular manner then everyone is going to have the exact same armies especially since Civ doesn't have that many different units per era.

 

[Greizer85]

What of a hybrid approach between 1UPT and stacks? Maybe allow 1 Melee + 1 Ranged + 1 Siege + 1 Air per Hex? It would streamline movement a bit, but would that allow for a stronger or weaker AI overall?

With a very good UI maybe (ha!)... I'd have to be able to see at a glance which units are already present in a hex, so that there'd be no logjams with my own troop movements. It sort of breaks immersion, though, since what would be the in-game justification for such an arrangement? 'I'm sorry sire, but we already have one Tank here. The, err, parking rules forbid any further steel beasts from entering!' (splat goes the advisor as the King takes his own 80-ton advice instead of his )
EDIT: Plus what Drowsy said above. It's not a hopeless idea though, but it has more moving parts than stacks, and the less parts there are for the AI to handle, the better.

Imo we should return to stacks in Civ VI if 1upt cannot be sorted. When we have self-aware holographic quantum computers (in two years ), we can return to it. I suspect though that the unwashed masses that play on Settler like blasting away clueless AI hordes so much that a return to stacks is not economically feasible. It would also mean admitting to one's mistakes, which is something that companies do no better than people (although they have more reason to).

 

[Ryika]

What of a hybrid approach between 1UPT and stacks? Maybe allow 1 Melee + 1 Ranged + 1 Siege + 1 Air per Hex? It would streamline movement a bit, but would that allow for a stronger or weaker AI overall?

Being able to put melee + ranged on the same tile pretty much destroys the purpose of 1upt, doesn't it? ^^ If anything, then allow units of the same type to go onto the same tile.

Although I have to say I like how pandora handles that stuff: ALL ranged units do dmg to the whole stack, that sort of kills stacks of doom right away. And I think that a bit of stacking management would be a lot easier for the AI than 1 upt. But still... I actually prefer 1upt. Maybe they should just remove many of the restrictions that the terrain has, give units indirect fire by default etc. - would also make conquering in really tight spots less painful.