Improving 1UPT

Hard rules are bad because they are frustrating. Take the "only melee units can take cities" which in the hand of the AI results in "I'm going to bombard this city state with my Longbowmen for another 25 turns even if it's defense is already at zero. Unfortunately, all my melee units that approach get killed, but I will bring them in one by one".

No thanks.

@Dunkah, So you can chose between a Division and "normal units". Why would anyone use 'normal units'? If you limit their movement to roads you're back with the clogged routes, if you have it as improved hexes, nothing much changes in the later game, but the early game becomes frustrating. So I need to build a costly road everytime I want to cross that plain. Costly and prevents rushing... Also, it's a hard rule. (and how does thatwork at sea?) I don't really see the difference honestly. I would go away from units totally and have agents (scouts, missionaries, trade units) and armies/garrisons (fighting units) where the first go more towards a roleplaying and the latter towards 'fast' tactical wars. But that's way off topic in regards with BNW.

mitsho said:

Hard rules are bad because they are frustrating. Take the "only melee units can take cities" which in the hand of the AI results in "I'm going to bombard this city state with my Longbowmen for another 25 turns even if it's defense is already at zero. Unfortunately, all my melee units that approach get killed, but I will bring them in one by one".

Click to expand...

Um....

I'm not quite sure what is meant by "hard rules".

Would removing a unit from play because its health reached zero be considered a "hard rule"?

What about a unit being unable to move once it expends all of its movement points? Would that also be considered a "hard rule"?

Dunkah said:

If you bring back the Army unit (I would call the unit a "division"), and allow three units per Division of any type then you eliminate the hard requirement of only 1 type per hex. Limit Divisions to roads or improved hexes (call it "being supplied"), and you create avenues of attack that the AI should be able to understand a little better.

The AI understood stacks of doom. If it knows to defend in force near roads it may do much better.

You would need to have a swap or pass-through movement action so two divisions couldn't occupy the same hex.

Click to expand...

The AI did not understand stacking at all. The reason that the AI fared relatively well with unlimited stacking is the simple fact that: adding a unit to a stack never made the stack worse. This meant that the AI did not need to understand stacking, it just lumped all available units together. Any bad decisions in stack composition were then easy to compensate by giving the AI more units.

Any form of limited stacking will add a new optimization problem to the game (i.e. creating the best possible stack given the available units). This is a type of problem that the AI is not particularly good at. (Because there is no simple formula defining what stack composition is "best"). The effective consequence of adding any form of limited stacking is that the AI has worse stacks that are in worse positions compared to a competent human player. And the bottom line is that the AI likely to do worse in with limited stacking than with 1UPT.*

Also note that abstract concepts such as "avenues of attack" are very hard to deal with for an AI in civ, which makes (most of) its decisions on a unit-by-unit basis. So that aspect of your suggestion is also likely to make things worse for the AI.


One modification that could help the AI (because it prevents log jams) is the following:
1) Allow an unlimited number of units on any tile.
2) Only one "active" unit any tile in one turn. I.e. Only 1 unit can attack from that tile or defend on that tile. In particular, two units that have acted in a turn cannot end their turn on the same tile.
3) If the active unit on a tile is destroyed all other units on the tile are destroyed as well. (Or taken as PoWs, which would be an interesting new mechanic by itself.)

The effect of this is that units are allowed to stack, but this something you would always want to avoid. This something the AI can deal with fairly well (simply by giving moves in which two units end up on the same tile a penalty). Any resulting bad decisions by the AI are fairly well compensated by having more units. The big advantage for the AI is that having more units becomes less of a liability because they less likely to be in each others way.

mintcandy said:

Um....

I'm not quite sure what is meant by "hard rules".

Would removing a unit from play because its health reached zero be considered a "hard rule"?

What about a unit being unable to move once it expends all of its movement points? Would that also be considered a "hard rule"?

Click to expand...

It makes sense that a unit dies when it reaches zero health, it doesn't make sense that a unit becomes impregnable to arrow fire when reaching 25 % of its health. I assume lots of people will complain about that.

You are right though that the movement points and all are hard rules, but a) they're a different sort and b) you need a minimum amount of those with the focus on 'as few as possible'. These are the basic rules: units move, units attack, units defend, units die. You are introducing a subrule into units attack, defend and die, an arbitrary limit, why 25 % and not 50 % f.e.?

Don't get me wrong, it's certainly a possibility, but I would rather go with rules that are to bring up a cliche "easy to learn" (and hard to master).

Trias said:

Also note that abstract concepts such as "avenues of attack" are very hard to deal with for an AI in civ, which makes (most of) its decisions on a unit-by-unit basis. So that aspect of your suggestion is also likely to make things worse for the AI.

Click to expand...

Agreed, that's why I think the way to go is to lower the number of 'units' on the map by replacing them by 'armies' and 'garrisons' or any other system and adding flexibility to how a army is made up so that the AI can react faster instead of the ~10 turns at minimum which are needed in civ 5 to produce or buy new different units. (i.e. wars should be fast in my mind in civ, not go on for an endless amount of turns).

A "soft" cap to ranged attacks might be better than a "hard" cap at 25%

If the damage done to a unit by a ranged attack were scaled by (the amount of defender health left +x)/(100+x) ,where x is a natural number selected as balance dictates,then your archers would be very effective at initially damaging a unit, and although they could kill it, they would be inefficient at killing it.

A ranged attack on an undamaged unit would still do the same damage it does now, but damaged units would take less damage from ranged attacks.

if x=20 this would result in requiring twice as many ranged attacks to kill a unit, but a set of attacks that would kill it now would still get it under 30 HP remaining; low enough for a single melee attack to finish it off.

Interestingly, this would naturally have no effect on ranged attacks strong enough to kill in a single shot, they would still be fatal.

This would make having melee units in your force a necessity. At the very least, my archer horde would bring a few mounted units to charge in, kill, and run away.

@Trias - Your idea would help with traffic jams greatly, but I can see 3) being highly exploitable against the AI (which I fear will inevitably leave large stacks in vulnerable locations). I would adjust this to "every unit in the stack takes damage" instead of "destroyed".

@AlazkanAssasin - Agree a soft-cap is ideal here. A change along these lines would make great strides toward balancing the power of ranged units!

mitsho said:

It makes sense that a unit dies when it reaches zero health, it doesn't make sense that a unit becomes impregnable to arrow fire when reaching 25 % of its health. I assume lots of people will complain about that.

You are right though that the movement points and all are hard rules, but a) they're a different sort and b) you need a minimum amount of those with the focus on 'as few as possible'. These are the basic rules: units move, units attack, units defend, units die. You are introducing a subrule into units attack, defend and die, an arbitrary limit, why 25 % and not 50 % f.e.?

Don't get me wrong, it's certainly a possibility, but I would rather go with rules that are to bring up a cliche "easy to learn" (and hard to master).

Click to expand...

I still haven't heard the distinction between a "hard rule" (which are apparently bad and undesirable) and a good one.

But if a good rule needs justification, how about this: In Civ IV, aerial attacks could only reduce a target's health to a certain threshold. It's been awhile since I've played Civ IV, but I think it was either 25 percent or 50 percent. Similarly, siege attacks (catapults, trebuchets, cannon, artillery) could only reduce a target's health to a certain threshold.

My suggestion isn't completely out of left field and it does have precedent within previous Civilization games.

@mintcandy - Hard rules are traditional Rules-with-a-capital-R while soft rules are mechanics in place to dissuade someone from performing an action.

In this example, your suggestion is "hard" because there's a rule preventing damage when the unit is below 25% health thus preventing you from attacking at all.

On the other hand, AlazkanAssasin's suggestion is "soft" because you'd be getting diminishing returns by attacking units with low health - the key is that you still can attack them if you want which is why it's "soft".

Hope that's clear.

@mintcandy:
Can only move on a raod is an example of a "Hard Rule". It limits what the AI can do and could perhaps make programing it to do the correct thing more difficult.

Can move over any tile is a "good non-rule" the AI can concentrate on the terrain and forming up a better line than adding in more rules.

Seek said:

@mintcandy - Hard rules are traditional Rules-with-a-capital-R while soft rules are mechanics in place to dissuade someone from performing an action.

In this example, your suggestion is "hard" because there's a rule preventing damage when the unit is below 25% health thus preventing you from attacking at all.

Click to expand...

:: shrugs ::

By that definition, my suggestion wouldn't be a hard rule, because I never said that a player wouldn't able to attack a unit at 25 percent health or below, merely that there would be no point, similar to having a soft damage threshold where further ranged attacks would do negligible damage.

:: shrugs ::

I guess the real motivation is to give incentive to making offensive melee attacks that don't necessarily capture a city. Whether that means a hard damage limit threshold for ranged attacks, or a soft damage limit threshold where further ranged attacks would do negligible damage, it doesn't really matter.

mintcandy said:

:: shrugs ::

By that definition, my suggestion wouldn't be a hard rule, because I never said that a player wouldn't able to attack a unit at 25 percent health or below, merely that there would be no point, similar to having a soft damage threshold where further ranged attacks would do negligible damage.

Click to expand...

Because I am confident the distinction is genuine I will attempt to lend a definition that will work for you to see what the other user means.

A hard rule is one in which the statement of it involves the word "cannot". A soft rule is one for which no statement of it can be contrived to include the word "cannot". Keep in mind that phrases such as "can only" are stylizations in natural language which are equivalent to some slightly different "cannot" statement.

A unit at <25% health cannot be harmed by archer fire. That is the statement of it. A mechanic that alters damage calculations fractionally cannot be contrived to include the word 'cannot' in the statement of it.

Diminishing returns for ranged attacks on land units below 25% would be a great solution... it's currently too easy to kill units with ranged fire, but making them impervious below a certain threshold was always silly.

The same rule should NOT apply to naval units. Diminishing returns on a land unit represents the unit's ability to disperse and take cover, which ships can't do.

I do see an issue with the diminishing returns idea though: the AI may not understand it and waste many ranged shots trying to get a kill rather than wounding healthy units (ie, using the mechanic as intended).

AriochIV said:

Diminishing returns for ranged attacks on land units below 25% would be a great solution... it's currently too easy to kill units with ranged fire, but making them impervious below a certain threshold was always silly.

The same rule should NOT apply to naval units. Diminishing returns on a land unit represents the unit's ability to disperse and take cover, which ships can't do.

Click to expand...

:: smacks my forehead ::

I hadn't even considered naval units being targeted by ranged attacks.

Yeah, I would agree that there should be no ranged damage threshold for targeting naval units, for the reason that Arioch stated.

Seek said:

I do see an issue with the diminishing returns idea though: the AI may not understand it and waste many ranged shots trying to get a kill rather than wounding healthy units (ie, using the mechanic as intended).

Click to expand...

That was part of the reason for my preference for a hard threshold: a hard threshold is binary....more discrete/less continuous than a soft threshold would be, so it may be easier to code the AI to recognize a hard threshold than a soft one.

The AI's mania for attacking damaged or vulnerable units, to the exclusion of healthy units which are actually a threat to them, should be addressed whether they implement a cap/diminishing returns or not. It's currently too easy to bait the AI guns with cavalry and get them to completely ignore the units that are actually killing them.

Another idea I was hoping to see with a 1upt paradigm: land combat frequently taking place away from cities, that combat could take place around advantageous geographic features, like chokepoints for defenders or open terrain for attackers.

Sadly, even with a 1upt paradigm, combat in Civilization V primarily takes place around cities. The argument for 1upt is that it is more chesslike, whatever that means.

I would love for Civilization's implementation of 1upt to eventually allow for wars of maneuver, but until combat stops being so centered around city tiles, I don't see how maneuver wars are possible.

So, basically, I've been trying to think of ways to get land combat away from cities. The conclusion I keep coming to is that while there is plenty of incentive for attackers to engage defenders a significant distance away from a city tile, there is virtually no incentive for the defenders to do so.

Sure, the defending city can't work tiles that are occupied by attackers, but the tile yield losses are usually insignificant compared to the considerable tactical advantages a defender has in fighting near one of their own city tiles.

Here's my suggestion: provide a happiness penalty for each hostile military unit from another civilization that occupies a tile claimed by one of your cities. Furthermore, the unhappiness penalty gets higher as the hostile military gets closer to a city tile. I don't want put discrete numbers on this suggestion, but I would expect the happiness penalty to get significantly worse as technology advances.

So, perhaps it would only be -2 happiness penalty for each hostile unit adjacent to one of your cities in the Ancient era, but -17 for each hostile unit adjacent to one of your cities in the Modern era...and perhaps -8 for each hostile unit two tiles away from one of your cities in the Modern era.

I'd also like for the penalties to be higher for developed cities than it is for nascent cities.

What do you guys think? Would something like this encourage defenders to engage away from a city?

I don't like to do it around the happiness mechanic, since that's one are where civ 5 is lacking and I'm not sure that tampering with the already unclear balance in peace time helps alot.

I'd rather see a option that tiles change ownership if occupierd for x+ turns (maybe 3 consecutive, or 5?). That way you could wage a war for tiles and not cities if you so want to... But I of course doubt the AI would understand it.

The easiest way would be to weaken cities again. But if you lower their range to 1, it becomes too easy to bombard them with the superior range archers which you can't lower to 1 range in the current system. The other way would be to dramatically increase the distance between cities (and make their population growth scale accordingly) so that the 2 range of the cities is worth less and you can avoid the dreaded focus fire (city + garrisoned archer + new archer + ship in city)... Still not possible for civ 5, so I'm out of ideas.

To get back to the original discussion, I see the problem in your proposition in one way:

You see that civ 5 has a problem with ranged combat in that it is the strongest one available. Your solution is to add a rule that its attacks are worth nothing if health < 25 points. I say that it's the wrong way to add more complexity, but instead revalue the ranged combat, make them deal out less damage in general or make them cost more etc. ...

The 'hard rule' bit doesn't really aim completey at what I am trying to say. It's more about a feeling of arbitrariness. Take Crusader Kings 2 f.e. where you need to fulfill condition 1-5 to be able to do action x, (i.e. have 4 counties as your own, be of age 30, 10 years ruling and of culture Norman, then you can found the Kingdom of x). Rules such as this become frustrating since in the game, the button "doesn't light up". It's puzzle solving. Your suggestion would make archers useless in that situation. If it's your only archer (because you are a newbie or it's a gift from a goody hut), you will lose it to the barbarian brute... To get back to the example, there's a reason civ 5 makes more money than Crusader Kings 2 (I assume, honestly i don't know). This adaptility is necessary for a good game, let's call it intuition. We assume that spears will be good against cavalry, but why would our arrows not hit our enemy? Why do I suddenly have a invincible enemy who can still attack me? That's unfair! And I feel that civ should avoid those feeling of unfairness. And also, less rules are better.

AriochIV said:

The AI's mania for attacking damaged or vulnerable units, to the exclusion of healthy units which are actually a threat to them, should be addressed whether they implement a cap/diminishing returns or not. It's currently too easy to bait the AI guns with cavalry and get them to completely ignore the units that are actually killing them.

Click to expand...

I think that's true. Although there's always the fear of insta-heal (which is the reason I usually try to kill weak units too).

Louis XXIV said:

I think that's true. Although there's always the fear of insta-heal (which is the reason I usually try to kill weak units too).

Click to expand...

Hmm, I never notice the insta-heal due to the fact that I only play with balance mods that remove these promotions precisely because of that. (which may also be why some of my analysis may feel skewed, I don't play vanilla, but GEM ).

mitsho said:

Agreed, that's why I think the way to go is to lower the number of 'units' on the map by replacing them by 'armies' and 'garrisons' or any other system and adding flexibility to how a army is made up so that the AI can react faster instead of the ~10 turns at minimum which are needed in civ 5 to produce or buy new different units. (i.e. wars should be fast in my mind in civ, not go on for an endless amount of turns).

Click to expand...

As I pointed out, "composite" units like armies just add another dimension for the AI to be bad at.

On top of that lowering the total units on the map also tends to hurt the AI, because the consequence of a bad decision is relatively greater.

Seek said:

@Trias - Your idea would help with traffic jams greatly, but I can see 3) being highly exploitable against the AI (which I fear will inevitably leave large stacks in vulnerable locations). I would adjust this to "every unit in the stack takes damage" instead of "destroyed".

Click to expand...

Why would the AI have large stacks in the first place. The whole point of the suggested set of mechanics is that it allows stacking, but having less units on a tile is always preferable. Since this preference is not situational it is very easy to communicate to the AI. Consequently, the AI should be very unlikely for the AI to cluster its units in a single tile. The AI possibly would have more double or triple occupancy tiles than a human would have, but that is in a large part due to the fact that the AI gets more units. (And any disadvantage it has from this is nicely covered by it having more units in the first place.)