1. We have added a Gift Upgrades feature that allows you to gift an account upgrade to another member, just in time for the holiday season. You can see the gift option when going to the Account Upgrades screen, or on any user profile screen.
    Dismiss Notice

The 1UPT system (as we know it) has to be expunged first, if we want a better A.I

Discussion in 'Civ - Ideas & Suggestions' started by Abade69, Nov 2, 2016.

  1. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    I think you mistake me. A Civ that's capable of domination will necessarily be coded to be militarily stronger than any other Civ in the game. That's how it's done. so you can't make them all tactically brilliant. If they're all equally good, then you end up with... Civs that are all incapable of domination! The point isn't that they're bad. They point is that they're EQUALLY bad. If you make them EQUALLY good, then they're no better off.

    When you're asking for a Civ that's capable of domination, what you're asking for is coding it so that a Deity bonus level AI lives on the same continent as a Prince level one. That's capable of domination.

    You're not getting the point. The point is that having to dig through all that in order to uncover someone taking your capital on the verge of Science Victory is seriously unfun. This is not me or other players. This is basic game design. You don't do that. It's bad design. The AI CAN declare Surprise War without denouncing you. In fact, it can (and has) done that to me while we were Declared Friends. It's part of what it's designed to do - fool you into complacence. But it wouldn't be fun if it did that and then you lost the game without being able to do anything about it, because you considered Intel a secondary concern.

    But seriously, don't take my word for it. You don't even have to necessarily lose. Just losing that much stuff randomly would suck for many players. Ask around.
     
  2. Abade69

    Abade69 Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    40
    Gender:
    Male
    Not necessarily. That's why we got diferent districts and different great persons for different victory types, the flexibility to attain any victory with any civilization (in most cases) applies to A.I's too. It has to do with agendas. If the A.I are coded to be equal, the difference would be on the playstyle and victory condition they are after (which dictate what they build every turn and how they behave to others), it doesn't make sense to make a code for each agenda as you suggest: "coding it to be militarily stronger".

    The current stalemates you see on the A.I Wars are not just because they are equally good or bad as you imply. The A.I cannot handle the carpet to a city and seige it as a human can because it IS USING LAGGING LINEAR ALGORITMS of the 1UPT, defensive buildings are to strong, they are incapable of upgrading their units to match the walls they are sieging, it is as simple as that. I have seen very pathetic attempts to capture cities with modern units (ancient ones defending it).

    To prove my point i will use the AI on Civ IV as an example: On a map with the same difficulty level A.I's you can still see domination all over, why is that? Is one better coded than the other? No. Two civilizations can have the same "domination" mindset and will not necessarily netralize each other. The imbalance comes when one of them sacrifices science, culture instead of military, resources, terrain, luck, etc. A real neutralizing effect would be both having same army size, technology and resources. On Civ VI, well you now know why they cannot even touch each other.

    Simply put, if an A.I decides it wants to conquer the world it will focus on military and sacrifice science, religion, culture or expansion, no need to put a Deity level AI with a Prince to make a rollover. The game is unbalanced and the 1UPT you are defending only makes it a bloody stalemate which would not happen in a stack system. You got what i am saying? One thing is what the A.I is TRYING to do, and the other is what it GETS TO DO. In the current state it almost never gets what it is designed for.

    You are right when you say i am not getting YOUR point. Someone taking your capital on the verge of a science victory is something you are supposed to prevent IN game, it is part of the risk, and it's part of the fun because this is an strategy game, things are supposed to go badly if your strategy is bad, got what i am saying? Or I should call Firaxis to put that "win anyways" button. :lol: It would make life a lot easier and predictable.
     
  3. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    Abade69:

    I will counter your Civ IV example with the same. It's not a matter of being "better" coded. You need to get that out of our conversation because it has no place in it. A Civ that's coded to dominate will dominate. It's not a matter of being better or not. It just rolls other civs in the military game. The Civ IV AIs don't necessarily prioritize one or the other. It's not quite that specific. It's more than a certain script will prioritize war a little more AND the Civ abilities match well to conquest, giving it a dominating edge. So we see Shaka dominating over and over. Because it is a better AI at domination. If you code things like that, then you will get a runaway nearly every single game. In fact, one of the most overriding concerns in high level play in Civ IV is WHICH Civ (not whether anyone was) was dominating if a dominating Civ was present. Every game. I know. I played that game. It was fun the first few years. It gets old.

    And it's not necessarily that Civ IV AI plays the stack game better. I've stalemated a massive AI stack between two cities just by manipulating the AI. Can't do a thing. Just moves from tile to tile. Other player manipulate the AI to take each other out, eliminating any domination threat to them whatsoever. Bleh.

    Stacks are dead. Get over it already.

    DON'T TAKE MY WORD FOR IT. Ask around. See if random city losses are good game design.
     
  4. CaiusDrewart

    CaiusDrewart King

    Joined:
    Mar 18, 2015
    Messages:
    834
    Location:
    Los Angeles
    I do play a lot of Civ IV. It's great :)

    You claim to be content with an AI that dutifully marches its units up to yours to be slaughtered, and that's great if that's your thing, but for me, that's boring and unrewarding. I want a sense of danger, excitement, and challenge. 1UPT can't deliver that. Stacks can.

    Of course you're right that the AI needs economic bonuses to compete on high levels. Figuring out ways to win anyway, sometimes by conquest, sometimes not, is really fun. I think Civ IV is the best strategy game anyone has ever made. That's why I'm a civfanatic.
     
    Thorik and Abade69 like this.
  5. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    We're both Civfanatics. I think Civ IV was okay for its day, but I don't play it now. Too many games of it, I'm afraid. Tends to run the same after a while. The Civ IV AI ALSO marches its units up to yours to be slaughtered at your leisure. It's exactly the same. If you're fighting a war against an uncontrollable stack with too few units, you've already lost. Likewise, if you're in Civ 6 and the Barbs are rushing you with 6 Horsemen and you have 1 warrior, you've lost as well. Different situation, same thing, broadly. Hundred unit stacks are tedious. I'd rather manage 16.
     
  6. Beorhtric

    Beorhtric Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    17
    Gender:
    Male
    To be honest I'm fed up of seeing all of these "AIs can't cope with 1UPT"/"Allowing stacks will fix the AI" threads... It wont. I've seen plenty of games where the AI Is more than capable of dealing with 1UPT and behaving sensibly with it. Heck some of those games were on a computer that only had 32KB of memory in which to run the game (it was a 48K machine, but 16K was used for various OS related things and video) - that's 0.000002 of the memory modern PCs have,

    Stacking certainly wouldn't have made a difference to the AIs attempt to take one of my isolated cities last night.

    The AI bought 6 Knights, 2 Catapults and a Battering Ram, the city was defended only by a Crossbowman and Ancient Walls. But it failed.. why? Firstly because it only attacked twice the whole 8 turns it had units in striking range of the city, secondly because whilst the Knights arrived quickly the siege units arrived one at a time two turns apart.

    Now in those 8 turns it could easily have attacked an average of 4-5 times each turn and that would of allowed to take the city (the two attacks it did do nearly wiped out the city defenses).

    If the AI's problem at this point had been "not enough firepower to take the city" then suggesting stacks to help the AI may have some weight to it... but that again could be fixed by adjusting unit balance (and altering the formula the AI uses to determin the number of siege/ranged to bring) rather than stacking.

    Bring back stacks may feel like it's fixing the AI, but in the same way that papering over that crack in the wall feels like you've stopped the house from falling down.
     
  7. Deggit

    Deggit Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    14
    Gender:
    Male
    In one word: YES.

    Think about it like this. Suppose you, the player, have just lost a war to an AI. What are some acceptable reasons for a player to lose a war?

    "I lost the war because..."
    • I didn't build enough units, so I was outnumbered.
    • I was behind in military technology, so my units were easily beaten.
    • I built the wrong unit mix, so my armies were easily defeated by combined arms.
    • I didn't pay attention to the map, so I was caught with my pants down when all the armies massing on my border invaded.
    • I managed my diplomatic, cultural and religious relations poorly, resulting in my neighbors teaming up against me.
    • I didn't perform enough espionage/diplomatic visibility, so I was taken by surprise by the enemy's strength.
    • I didn't evaluate the strategic geography so I attacked the wrong city while ignoring the more valuable one; or I attacked over mountains when there was an easier route; or I failed to defend the city that the invading force was obviously heading for, etc.
    These are all STRATEGIC reasons for losing a war. The important point here is that you gain advantages and disadvantages in war based on how well you manage the game's other subsystems: production, technology, exploration, and diplomacy.

    If you the player can do all of the above right and then lose a war anyway because you stupidly threw archer after archer into melee combat with a tank, then WTF was the point of all the other game subsystems?

    The same goes for the AI. the AI should win or lose wars for strategic reasons and this is hard enough to program already.

    The AI must already...
    • judge the right amount of military units to produce for the situation
    • build the right mix of infantry/cavalry/artillery/naval
    • judge how much to prioritize military technology
    • watch the map and guess when neighbors have aggressive intentions
    • manage diplomacy with its neighbors to try to avoid dogpile (often instigated by the human player)
    • when at war, figure out what cities & targets are important and the most favorable invasion routes to attack/defend.
    Far from being Tic Tac Toe, huh? ;)

    These are already big challenges for an AI. Yet without meeting these challenges the AI cannot even fight a war much less win one.


    Actual combat should be easy for the AI to manage. That means minimize opportunities for the AI to make stupid mistakes. Every time you throw another granular little tidbit into the combat engine (e.g. flanking bonus is doubled when you are flanking from a hill) you're giving the player one more way to defeat the AI that can already barely handle combat.

    What it comes down to is this: every added detail in the combat engine is effectively another opportunity for the human to nullify the strategic layer of the game when warring with the computer. That's bad.
     
    Last edited: Nov 3, 2016
    n0thing likes this.
  8. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    Eh. Bad is relative. Civ isn't a war game at its core. It really isn't. 90% of your time is spent building things. There are better wargames out there if you are looking for that. Civ 6's 1UPT is a toy soldier system, not a real war game. Because if it were, it chose Panzer General's systems and that's actually incredibly punishing, EVEN FOR PLAYERS. The reason the AI is bad is because it's a very difficult system to get a handle on, even for people who spend a lot of time on it.

    So we go to tic tac toe? I dunno.

    War has ALWAYS been a way for the human to nullify the AI's strategic bonuses. This was always true. The singlemost consistent advice you can give to a Civ 4 player who was losing strategically is to wage war. You have to, in some games, just to make the difference. Let them build the wonders and then you take them over. Always has been this way. If this is bad, then Civ 4 is catastrophically bad and the stacks didn't save it. The AI is no more competent at stack combat than it is at hex combat. NO MORE COMPETENT. It's equally bad. That's why it needs a stack twice as large just to have some semblance of a chance.

    So the solution isn't stack combat. It's to make Civ combat exactly tic tac toe. Great. Good luck with making your new game. It'll be amazing, I think.
     
  9. Deggit

    Deggit Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    14
    Gender:
    Male
    I'm not saying that the AI is good at stack combat, I'm saying something a little different.

    I agree with you that Civ4 and Civ3 AI was bad, the truth is that the AI has always been bad at the strategic layer.... it only becomes remotely competitive when you give it free units (either directly or through production bonuses). The AI has also always been terrible at the tactical 1UPT layer added in V. Even after six years of development.

    The significance of the difference between 1UPT and stacks is that stacks are less granular. The 1UPT carpet of units gives the player ample opportunities to defeat the AI piecemeal. This nullifies any strategic advantage the AI has managed to attain over the player. Stack combat means fewer playing pieces on the board. That lets the AI prioritize thinking about how big the stack should be, what unit mix it should have, and what invasion route it should follow. These are already ways the AI can lose, but if it manages these tasks correctly then the AI can sometimes successfully leverage a strategic advantage, when it has one, against the human. Doomstacks are scary precisely because they give the player fewer opportunities to exploit his/her way to victory against an AI which has massively outproduced the player.
     
    Last edited: Nov 3, 2016
    n0thing likes this.
  10. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    You're implying that the AI is marginally better at stack combat. It's not. If it were, equal stacks would have equal chances. They don't. Civ IV stacks are rock-paper-scissors. Assuming you can tell what the AI stack is made of (and you can), you can always make a counter-stack in the time it takes to get to you and massacre it hardcore. So they don't play the stack combat any better. They're just as bad at it. EQUALLY BAD.

    The core frustration fueling all this fury is the idea that the AI is some magical genie in the box that thinks. It's not and never will be. It's a script. It can always be exploited, unless your game is literally tic tac toe. At which point it's pointless to have a system in the first place.
     
  11. Deggit

    Deggit Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    14
    Gender:
    Male
    Exactly like you said, Civ is NOT a war game, that's why it has always been wrong to take the 1UPT system from wargames like Shafer did.

    It is a game about building things and in general the player who is more successful at building up his productive forces, should win the damn game either by building more soldiers than everyone else, more culture than everyone else, more science than everyone else etc. Adding a strategic war layer (where you have to at least prioritize military research & production enough to defend yourself even if you're going for a culture victory) doesn't nullify the strategic game because it's still a contest of productive forces. That is, do you have enough production capacity to build up your culture while also prioritizing military a little too?

    But adding a tactical combat layer does nullify the game because it is no longer about who has the most productive forces. An AI with a far larger and higher tech army gets thumped by a player sitting with one archer on a hill chokepoint. See what I mean about making combat more complicated creates more opportunities to screw the strategic layer? Whether it's player exploits or AI stupidity doesn't matter - the player usually doesn't feel rewarded for her cunning by pulling off the archer trick. It's an obvious, stupid exploit that just highlights the AI's inadequacy. That's why people are complaining and want a harder AI.

    The archer on a hill literally nullifies all the effort the player put into building up her cities and districts, claiming and settling new land, researching technology and correctly allocating production capacity to different tasks, because in the end all she needed to defeat the AI was one archer. Meanwhile the AI on the other side of the hill was either correctly scripted to settle/develop in a highly efficient way, or was given cheats to catch up, yet all of its productive forces cannot defeat the player because its combat script cannot figure out that it should go around one puny hill.
     
    n0thing and Abade69 like this.
  12. criZp

    criZp Emperor

    Joined:
    Jul 19, 2013
    Messages:
    1,882
    Location:
    Nidaros, Norway
    What bothers me most is the time spend on moving around the individual units during wartime, it's just not fun.

    Production times are off the charts too, my 5 population capital can pop out a warrior every single turn, and they have 0 upkeep, just ridiculous. The city had like 15 production and warriors cost 40, but still, 1 turn per warrior...

    I would like to see how things turned out if warriors had maintaince and the military policy slots were disabled, how much the unit spam would be reduced.
     
    Abade69 and TehJumpingJawa like this.
  13. Azem.Ocram

    Azem.Ocram King

    Joined:
    Sep 15, 2015
    Messages:
    637
    Gender:
    Male
    Location:
    WA
    Might I suggest a limit of 3 per type and 2 types per tile? 3 identical units would have the strength an Army currently does while an actual Army should be stronger. Protecting a corps of crossbowmen with a corps of swordsmen would be manageable with the AI. Friendly nations with open borders would allow units to pass through each other if it doesn't break the rules of 3 units per type and 2 types per tile.

    City-States should also decrease unit spawning or allow the Suzerain to command them to decrease unit spawning.
     
    Abade69 likes this.
  14. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    One archer on a hill defeats a whole army of mamluks? That's weird. Because the last time I had one archer on a hill, the Barbarians attacked it down and went on to raze everything. So it's not an AI problem in general. It's probably a specific script and well within something that's fixable. Find another example. Or just make your tic tac toe game already.

    Even if you successfully assert that Civ combat should be tic tac toe, I have poor predictions for your efforts in making that stick.
     
  15. Namaspamus

    Namaspamus Warlord

    Joined:
    Aug 5, 2010
    Messages:
    206
    Am I the only one to find that Civ V BNW's AI was slightly less dumb? It sure did a ton of stupid moves, but focused fire better I find, and tried to take each occasion to kill a unit of ours. Sometimes it made good coordinated attacks on your units (often just after you refused a peace treaty, remember?) I still haven't played the VI enough to be sure, but so far it seemed worse to me.

    In fact it (Civ V AI) seemed even nerfed by devs to me, I noticed its ranged units like archers had, like catapults, the need to "set up" prior to firing. If it could move and fire in same turn, it probably resulted in inavoidable frustrating kills so devs introduced that (I think). Not saying it was smart either.

    And I strongly disagree with the idea of bringing back stacks of doom. Maybe, as it's been suggested, AI's silliness should be compensated with more free promotions rather than ability to crank huge carpets. Stronger units would make up for the mistakes better than larger armies I feel.
     
    Abade69 likes this.
  16. Abade69

    Abade69 Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    40
    Gender:
    Male
    Civ V A.I was improved over the years and even used it on Beyond Earth (the combat bits). But for some reason they prefered to make a new one for Civ VI. Some things are better (units are not that suicidal now) and some things are worse (settlers not settling). No matter how good the A.I becomes, they will not be able to make tactical decisions on a map flooded with units, so the tactical mechanics become more or less pointless.

    The system i was thinking about has to do with limiting the production of units by linking it to infrastructure (which links to economics indirectly), policies (which links to culture and faith), and of course hammers (pretty unbalanced right now). Both stacks and carpets of doom (as seen in past civ games) have to go.

    But the A.I IS IN FACT better at stack combat since it just has to move around its units to the next target with NO OBSTACLES (those being other friendly units), it is not on the script (magical or not), nor the rock, paper, scissors; it is on the board, in which it may flow better (to lose or win, it doesn't matter). I don't get how you still don't understand the point made on 1UPT. If we reduce the number of obstacles (granulation) the units flow freely and can use those numbers in its favour and still use tactics (on a flooded map you CANNOT, Which also defeats the purpose of having tactics).

    1UPT:
    - Own units = obstacles.
    - Other units = obstacles.
    Stack system:
    - Own units = NON-Obstacles.
    - Other units = Obstacles.

    If you are still defensive on the idea, i am not talking about having the Civ IV, III, UNLIMITED stacks. I am talking about LIMITED stacks and carpets, keeping the tactics alive (you could still play as a turtle).

    Instead of this headache, CPU power consuming, carpet of lag:



    You can have a system that lets units of the SAME type to fuse power on a single tile, but should never rise to infinity as you have a supply based on infrastructure, policies and techology (not just gold, and strategic in nature). Both carpets and stacks would be predictable on size and would not flood the map in any way, AND you could still use turtle tactics.



    To clear my point a little futher, you would not be seeing this on your map. You would be seeing a single calvary unit with a defined "tall power" (promotions, numbers, or whatever fits in your mind), another single infantry unit with a defined "tall power" to the side, and a cannon open to the other side. It's not the stack system i am promoting as you can see. It does not consider other units to fuse or split, as a matter of fact, it would be impossible to "join" units of the same type. It's more like a hybrid that allows tactics and is linked strategically to other aspects of your civilization, not just tactics. The "tall" would come from diferent policies, infrastructure, technologies and other perks, always limited to a set number ("wide" too).

    The point is to get rid of the stacks and carpets of doom as both systems FAIL HARD. I agree on your views of the classical stack system as a failure and hindrance.
     
    Last edited: Nov 5, 2016
    n0thing and Azem.Ocram like this.
  17. Deggit

    Deggit Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    14
    Gender:
    Male
    The core of the problem is that the game engine produces too many units.

    If they are spread out in a carpet they are a nightmare to manage and the AI can't handle it; if they are concentrated in a stack it feels pointless. And either way movement and battles take way too many clicks and too much time.

    The reason the game engine produces so many units is because each unit has to cost a relatively low amount. That way you feel like you're accomplishing something. If it took 80 turns to build an army of chariots, that would take away the "one more turn" feeling. So instead the game has you take 8 turns to build a chariot, and you can do it 10 times to build a stack of chariots.

    It also increases the flexibility offered to the player - they can take a QUICK break from infrastructure or wonders to pump out one or two military units when needed. You can also build a lot of units quickly by switching ALL your cities to chariot production at the same time.

    This is all good. It adds strategy. The problem is the end result: way too many units to keep track of and manage, for both the player and AI.

    The answer is to abstract away units. Just have armies instead. You get one army for every city you have, and you can unlock a few more with technologies or civic policies. A city can only switch to military production if an army is located inside one of its districts. Whenever a city builds a military "unit," that unit automatically gets added to the army. The army moves as one piece. When an army encounters another army, it's a one-click battle - the outcome depends on strength, tech, unit mix and terrain - and armies are usually not destroyed, they just lose some strength and are forced to retreat.

    With this approach you combine the "one small piece at a time" of the production side of the game with a simple "6-12 playing pieces to keep track of on the map" approach that the AI can handle no sweat.
     
  18. Roxlimn

    Roxlimn Deity

    Joined:
    Feb 11, 2005
    Messages:
    3,526
    The current stacking system of Civ VI allows you to stack up to 4 units on a tile - an Army and one support unit. the problem with carpets of doom is the AI, not the system. Fix the AI.
     
  19. Beorhtric

    Beorhtric Chieftain

    Joined:
    Nov 2, 2016
    Messages:
    17
    Gender:
    Male
    Pretty much this.

    It's a case of tackling the root cause of the issue and not the symptoms. Which means adjusting the AI so it doesnt need to produce so many units (and even in it's current state it doesnt need to).

    I suspect it is also because the AI is able to cheat at production and therefore doesn't have to make quite so many "unit or improvement" decisions the player makes.. leaving it more time with nothing to do but churn out units.

    Not entirely sure I agree with the "just armies instead" as that has the potential to feel a little odd to the player - an explorer as an army makes no sense really nor does an archaeologist army - make combat/domination require no thought - you're pretty much doing the equivalent of just shuffling one warrior per city around - there's no tactical decisions to be made when it comes to it.

    I do support the concept of armies as a unit - not the civ 3 style or 6's non-armies - and have indeed been working on my own idea for armies and how such a system might work to keep combat interesting, require the player to make meaningful choices and keep a sense of immersion. I'll post it when I've got a few issues nailed down (modern era's proving a challenge since modern armies tend to be wider a looser than the strict formations of prior armies)[/QUOTE]
     
    Abade69 likes this.
  20. TehJumpingJawa

    TehJumpingJawa Warlord

    Joined:
    Aug 15, 2008
    Messages:
    204
    Abstracting away units into armies is where I'd like to see civ go too. (You'd also get rid of all the other unnecessary civilian units; builders, traders, religious units, etc, a huge declutter of the map)

    Not sure precisely what'd determine army count; pop? Tech? Great people? Government Civics? civilisation traits? Strategic resources?
    You'd want keep it low, and the tactical benefit of more distinct armies would necessitate a fairly uniform cap. Cities would need to be able to defend themselves (Total War garrison style)

    The strength of your armies(and city garrisons) would be under your control; directly linked to how much industrial capacity/population you dedicated to it.(build up time and maintenance would be considerations, as would something like eu4 army tradition)

    It'd radically shake up the entire game, in particular the start (exploration would be much slower).

    We'd be left with something closer to Rome2 TW / EU4 army model.

    Done right, I think it could propel the civ franchise in a positive direction.
     
    SCBrain likes this.

Share This Page