Adventure Fourteen Discussion

[mbuna120]

I didn't report on this game, simply because I accidentally erased my savegames and photos. I start many games on my own free time, trying out different stuff and bettering my early turns, so I always delete my files afterwards. But I lost as well. I don't remember what year I lost in, but it was early. I had just founded a city near the crabs? and the elephants (my 3rd city) and had a total of about 6 defenders overall, when persia came at me with a stack of 4 immortals to trounce my latter two cities, and then a smaller stack of two archers and an immortal towards my capital. And as he so graciously smacked my cities down, caesar was ready with settlers to found TWO new cities in between them, effectively locking me to my capital, even if I had somehow managed to weather the attack on my final city... which I doubt was probable. It was actually funny to me to be trounced so early. I had even decided to try this one without slavery. But when I bent research to bronze working, only to have no copper... Oh well. It was a fun 20 turns or so

[T-hawk]

The randomness of early war declarations was always a problem for the balance of Civ 3 events, but I agree that it's a much bigger problem in Civ 4. Here's some reasons that come to mind:

The AIs in Civ 4 are opportunistic about weak targets. Also the AIs are just better, in that they can deliver dangerous and more coherent military stacks sooner and on lower difficulty. Civ 3 rarely saw a serious early wipeout after 2500 BC or so except on Deity. This is the root problem on which everything else rests, but we aren't going to weaken the AI in order to solve it.

Later and less tech trading in Civ 4 means that it's harder for a player to instantly pick up an appropriate military tech in case of emergency. Also, in Civ 3, a player that sniffed out a sneak attack coming could bilk free techs and cash out of the offending AI by buying on credit before the war declaration. That's gone in Civ 4 and isn't coming back.

Civ 3 made it far easier to buy alliance help in the event of an early war declaration. Civ 4 not infrequently sees help entirely impossible to buy, and that's a core element of the smarter AI behavior.

Civ 3 Conquests also had the one-stop military solution of the Statue of Zeus wonder. No such silver bullet (thankfully) exists in Civ 4.

In Civ 3, the player could get a far higher kill ratio by exploiting offense/defense strength discrepancies between units via tactical maneuvering, and using the free bombardment damage capability of catapults. The mechanics of unit strength and siege weapons in Civ 4 don't permit that. You can get some advantage with specialized promotions, but the AI can use that almost as capably as the human. This is also just part of Civ 4's core game and won't be changed.

The width of the Civ 4 tech tree means that players can get a serious economic advantage with depth-first economic research while bypassing military techs. In Civ 3, there was no such thing as building universities while bypassing spearmen and archers. This is a key problem for competitive events: success will go to those who gambled the thinnest on military and got away with it. It can be mitigated by variant rule (along the lines of the Walls rule in Adv 13), or by making the scoring conditions not dependent on the early economic growth curve, but both of those are tricky to implement well.

Without resources in Civ 4, you're just screwed for capable military units in the early going, all the way up to longbowmen. In Civ 3, you could usually handle incoming swordsmen with spearmen, archers, and catapults; but in Civ 4, nothing stops axemen and swordsmen except for other axemen, and you can't even counter horse archers without spearmen or elephants. Conversely, an AI without metal poses a minimal threat, and one also without horses poses none. This of course is entirely solvable by scenario and map design, but it would definitely take away some of the excitement if we got used to Sirian always providing us with copper. And there's also the risk of overcorrecting, allowing a player with metal to easily conquer a neighbor without, making that required for competitive success.

So overall, I'm afraid a large part of the problem rests in the core game mechanics, with only the resources and depth-first economic gambling being solvable with scenario and map design.

In Warlords, the Protective civ trait makes the archers (your early resourceless unit) significantly stronger; using Protective civs on high difficulty may be a good idea. The trait kicks in if you need it, and doesn't give an advantage to anybody who doesn't get attacked.

[uberfish]

The width of the Civ 4 tech tree means that players can get a serious economic advantage with depth-first economic research while bypassing military techs. In Civ 3, there was no such thing as building universities while bypassing spearmen and archers. This is a key problem for competitive events: success will go to those who gambled the thinnest on military and got away with it.

There is no real solution to this in any economic based game. Diverting resources away from defence and reinvesting them into the economy always results in a stronger economy. The only time you actively want to build military from an economic point of view, as opposed to being forced to build it, is if you plan to use it to attack someone. (And the attacking players in civ4 do pretty well for themselves.) Part of the skill in the game, especially at the highest difficulty levels, is deciding how little defence you can get away with in builder mode without completely folding if you get attacked.

The early wipeouts could be mitigated though just by not starting the player next to Rome too often. CG1 Archers with cultural defence or Walls do defend all right against swords even if you can't stop them pillaging. Praetorians are another matter.

One thing I don't like about the hidden military resource system is that it is a true gamble whether you guess the right resource or not, and the timing of your opening is screwed up if you guess wrong. I'd prefer it if there was some way to start tournament games with metals and horses revealed.

[Blake]

Sirian's issue here will go unchanged by Blake's AI changes, the level will just be lower, instead of the dice rolls of who lines up againist whom being able end anyone's game on Emperor, maybe the change will make it Monarch- but the issue will still be out there that until about 500BC an Emperor AI war declaration wil probably result in a loss- there is no fixing it outside of some patch that disallows early war declaration. I think that those playing the Extreme games just have to shrug it off, it is no big deal.

Wrong. Everyone who has said this is wrong.

My AI primarly enhances mid-late game performance, in many ways the AI are even slightly weaker in the start due to more focus on economy - not much weaker mind you, but a little.
But the main thing is the AI keeps re-pulling ahead, you can catch up, and watch in wonderment as the AI again leaps forward, with 1.61 AI it's very rare to catch up and then NOT be in front place permamently.

Unfortunately we wont see REAL competitive from the AI's until some can actually conquer their weaker neighbours, because ultimately a player with twice as much land is going to just have a much stronger economy, so warfare is always a crutch towards victory. The AI still suffers from the key flaw of throwing the game away the moment it declares war, regardless of when that declaration is. AI's never profit from war unless it wipes the human out in the early game. AI's are always too tough for other AI's to break open and in the worst case they pretty much stop space-racing just because someone declares war on them.

[Blake]

I posted my "report" primarly to show HOW a defeat happens to a good player. It usually involves multiple strokes of bad luck, for example in my case the worst luck was having my allies fight each other instead of Rome. Next bad luck was the combat results in an attacking archer beating my defending archer (3 vs 4.8 ) and an attacking axeman beating an Elephant (5.5 vs 8.8 ), both of these being battles where my unit had 60% more strength (and I consider anything above 50% to be a shoe-in).
Finally was that Rome also had his praetorians, so not only did he have luck on his side, he didn't even need it for many of his battles.

I'm pretty sure the whole fight would've been a cake-walk if I could've bribed Cyrus onto Caesar, I would've lost Bombay still but then the Roman attack would've been halved in strength and broken against Madras, after that I could've pushed back and taken most of Rome with Elepult.

I don't think the Praets were the biggest factor... the combat rolls are ALWAYS a big factor at emp+, Praets just make it more likely to get bad rolls.... and clearly in my game Caesar didn't need Praets to curse me with bad rolls.

Generally speaking, at Emperor+ level, being metalless AND being up against certain AI's is going to present a problem...

AI's which are a genuine danger by themselves:
Caesar, Napoleon.
Caesar has the strongest UU in the hands of AI and Nappy builds the most units of any AI and he seems to just love stacks of axes and swords.

Then there are AI's which are possibly not dangerous by themselves, but having them as neighbours can be a severe complicating factor.
Monty, Alex, Mongols.

These guys aren't so hard to befriend and don't tend to produce wonderful unit mixes but you can kinda rely on them attacking if not befriended, on the plus side if befriended they make extremely useful war-allies - in fact in many ways it makes it too easy for the warmonger because it ensures 2 on 1 wars.


I believe that all the other AI's can be dealt with in the vast majority of games. Exactly how to deal with them varies, like Tokugawa and Mansa Musa require quite different approaches. But it's not essential to have metals or war allies to deal with them.

[ThERat]

Without resources in Civ 4, you're just screwed for capable military units in the early going, all the way up to longbowmen. In Civ 3, you could usually handle incoming swordsmen with spearmen, archers, and catapults; but in Civ 4, nothing stops axemen and swordsmen except for other axemen, and you can't even counter horse archers without spearmen or elephants. Conversely, an AI without metal poses a minimal threat, and one also without horses poses none. This of course is entirely solvable by scenario and map design, but it would definitely take away some of the excitement if we got used to Sirian always providing us with copper. And there's also the risk of overcorrecting, allowing a player with metal to easily conquer a neighbor without, making that required for competitive success.

This is precisely one of the key weaknesses of CIV. In C3C it was possible to overcome a game without access to iron/salt/etc by a good combination of archers/horses/cats etc. This way you could fight your way back into a game. Not in CIV

[Kylearan]

Hi,

The randomness of early war declarations was always a problem for the balance of Civ 3 events, but I agree that it's a much bigger problem in Civ 4. Here's some reasons that come to mind: [...]

While I agree with all your observations, they only list reasons for why it's more problematic for the human player to fend off early (random) declarations of war, but not why they are happening in the first place - which is the main problem for the balance of Civ events.

The main problem is the AI's lack of long-term goals and mid-term objectives. If declarations of war from an AI could be traced back to its needs and long-term strategy, there would be less randomness, and if some players get attacked and others do not, at least there would be a more sensible explanation which would make it easier for the attacked player to accept the declaration of war.

I've thought a lot about a more sensible, goal-based AI which is capable of going for specific victory types (and even switching from one type to another, if needed) during the past weeks, and I think it is quite doable. The CIV designers put too much emphasis on AI personality and effects of diplomacy. For example, most of the time it would be more sensible for an AI to declare war on a friendly neighbour, who perhaps owns a critical resource, than to declare war on an AI with a different state religion who is far away. Depending on a chosen long-term goal, it should be possible to formulate and decide between mid-term objectives, like capturing some cities from a neighbour as soon as the economy supports this in case of an aggressive victory type, or turtling up and emphasizing economic techs and buildings in case of a run for a spaceship or cultural victory (if the current ampire is deemed sufficiently large to reach that goal). This would reduce randomness a lot, although you have to make sure it doesn't become too predictable.


Digressing a bit, I think it would also be better if the AI were more flexible with its units and their assigned roles. Using some kind of "threat" map [1], it would be possible to identify which cities can be defended only by a weak garrison, and which cities are in danger and should be fortified better. This would also prevent the AI from putting a unit on a resource deep in its core. This in turn would free up more units for attack stacks, which should help to quickly capture two or three cities from an enemy before he will be able to muster a defense, making war more profitable for the AI.

But this becomes quite off-topic, and I'm still working on these concepts. Maybe I will be able to implement some of these ideas, although my envisioned AI architecture would most certainly require a complete rewrite from scratch...

-Kylearan


[1] You could start with all tiles on the border, and depending on relationships, power and state of war regarding the neighbouring civ, assign a threat value to that tile. The same could be done to tiles containing known enemy stacks, the more dangerous the stack, the higher the value. Then, using a breadth-first graph traversal of the map and decreasing that threat value for each adjacent tiles, the AI could identify which areas of its empire are (relatively) safe, and which cities are in danger of getting attacked. This threat map should be generated each turn, then defenders assigned according to this map. This would be more like a human would defend his cities, and would free up more units for other purposes.

[uberfish]

This is precisely one of the key weaknesses of CIV. In C3C it was possible to overcome a game without access to iron/salt/etc by a good combination of archers/horses/cats etc. This way you could fight your way back into a game. Not in CIV

The map scripts do a pretty good job and it's rare to be missing all of iron, copper and horses. I would say 5% or less and even that estimate is based on OCC standard pangaea which is all I have time for in private games lately. Trouble is if you don't have them, you don't know you have to stretch to grab them until it is too late.

[Atlas]

Wrong. Everyone who has said this is wrong.

My AI primarly enhances mid-late game performance, in many ways the AI are even slightly weaker in the start due to more focus on economy - not much weaker mind you, but a little.

How so? What ever tweaks you make to the AI- it does not change the discounts that it gets- that are in the beginning overwhelming. If the patch changes affect the mid-late game- then they don't really address the issue that Sirian wrote about- that early war declarations will end your game on high levels and are rather random. The AI whipping or prioritizing Pottery/Alphabet don't address the root of the issue, which is that the early discounts are insurmountable in the begining. It takes hundreds of turns to overcome the discounts through smarter play.

[T-hawk]

While I agree with all your observations, they only list reasons for why it's more problematic for the human player to fend off early (random) declarations of war, but not why they are happening in the first place - which is the main problem for the balance of Civ events.

I disagree with that last bit. The balance problem for tournament events is not that early war happens to some players. It's that early war crushes some players. If the attack can be fended off, early war becomes just a temporary obstacle rather than a game-ending brick wall. In Civ 3, good players could fend off pretty much anything below (and even on) Deity difficulty. In Civ 4, even Emperor difficulty can see a crushing defeat. Even I barely held on in this game, thanks to luck from the combat RNG and in the Rome AI's unit movements.

I don't see any way in the current or future game mechanics to remove the threat of early war. That AI behavior is not subject to control by the scenario designer, short of the Always Peace option. There are two ways to avoid war: get to Friendly diplomatic status, or build your military sufficiently larger than the would-be attacker's. Both are impossible in the early game on high difficulty. So no matter what the scenario designer does, there's a substantial threat of the dice rolls coming up craps for the player. So I propose that the solution for tournament play is to make sure the player's civilization always has the resources (literal and figurative) to hold off such attacks.

You propose solving the problem by implementing a strategically-oriented AI, such that the player could be assured of having a development plan that would avoid war. There could be value to that approach (if it turns out to be feasible to use, either in the core game or via modification). Or it could just shift the problem. Economic success for the player would then depend on figuring out exactly how weak you can be while still avoiding the AIs declaring war, and then walking that tightrope.