Improving 1UPT

Trias said:

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

Click to expand...

Agreed, but I don't want "composite" units, I want to do away with unit types like spearmens, archers and so on. It always felt strange to have entire armies just made up of one 'weapon' and to have units based on 'weapons'. The "build unit" -> "move unit and use it for the rest of the game or until death" mechanism of civ seems like a simplification originating in ~1992 when the first civ came out...

I'm not sure that a system of mobilization is the way to go, but I do feel that the "unit" system fails at illustrating the various different combat styles found in history (from the ancient era pitched battles to hit-and-run horse archers, trench warfare and surgical air bombardments). Again, that's nothing for BNW though

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.

Click to expand...

But that's mainly because the developers chose the "freebies" route of giving the AI more stuff to make it a competitor instead of giving it "scripts" or real intelligence. I'm confident if you have a system that lets the AI react quickly instead of making it have to plan ahead a long time, the game will become a better competition (though not necessarily more fun ).

mitsho said:

But that's mainly because the developers chose the "freebies" route of giving the AI more stuff to make it a competitor instead of giving it "scripts" or real intelligence.

Click to expand...

No it is not. Overall scaling down of the number of units means that the relative importance of each move becomes bigger, and that AI mistakes (which are inevitable) hurt the AI more.

(Also, "scripted" AIs are useless for games like civ, because they cannot cover all possible situations, and are therefore easily exploited by presenting it with situations not accounted for by the script. You seem to be making easy accusations without having given the problem of programming an AI for civ (or other 4X games) as serious thought.)

I'm confident if you have a system that lets the AI react quickly instead of making it have to plan ahead a long time, the game will become a better competition (though not necessarily more fun ).

Click to expand...

The AI's problem in civ5 is not getting units on the map, but what it does with units once they are there. So, no improving reaction time is not going to do much for the AI.

Trias said:

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.)

Click to expand...

I think you answered your own question: It would

just add another dimension for the AI to be bad at.

Click to expand...

....

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.

Click to expand...

In Comunitas/GEM the AI gets free XP and is programmed to produce fewer units (ie, no carpet of doom) and it performs better than vanilla/G&K. When the AI dows you and you see a force of only level 4 or 5 units approaching (much like you would playing a human) it can be a frightening prospect. Also this system alleviates the traffic jams the AI usually ends up in, which is a serious problem for it. Not a perfect system, granted, but better.

Trias said:

The AI's problem in civ5 is not getting units on the map, but what it does with units once they are there. So, no improving reaction time is not going to do much for the AI.

Click to expand...

I think you misunderstand mitsho here. He's referring to the AI's boneheaded stubbornness in completing certain objectives in spite of changing situations. For example, an English AI has determined to attack one of your border cities, you bribe the German AI on the other side of England to help you in your war but England keeps pushing the original city despite the fact that the better play would have been to stay and defend it's flank.

Seek said:

I think you answered your own question: It would

Click to expand...

Maybe read it again? I quite clearly explained why there would be no large stacks to exploit.

In Comunitas/GEM the AI gets free XP and is programmed to produce fewer units (ie, no carpet of doom) and it performs better than vanilla/G&K. When the AI dows you and you see a force of only level 4 or 5 units approaching (much like you would playing a human) it can be a frightening prospect. Also this system alleviates the traffic jams the AI usually ends up in, which is a serious problem for it. Not a perfect system, granted, but better.

Click to expand...

You are misunderstanding whats been said. The suggestion was to change the game mechanics to decrease the number of units across the board. This in general works as to strengthen the advantage of human player. What you are talking about is replacing add bonuses from having more units to having stronger units, which is something I have advocated since the release of civ5.

I think you misunderstand mitsho here. He's referring to the AI's boneheaded stubbornness in completing certain objectives in spite of changing situations. For example, an English AI has determined to attack one of your border cities, you bribe the German AI on the other side of England to help you in your war but England keeps pushing the original city despite the fact that the better play would have been to stay and defend it's flank.

Click to expand...

No he clearly isn't.

No he clearly isn't.

Click to expand...

Well, I wasn't. Doesn't change that it's a good point seek brings up.

Trias said:

No it is not. Overall scaling down of the number of units means that the relative importance of each move becomes bigger, and that AI mistakes (which are inevitable) hurt the AI more.

(Also, "scripted" AIs are useless for games like civ, because they cannot cover all possible situations, and are therefore easily exploited by presenting it with situations not accounted for by the script. You seem to be making easy accusations without having given the problem of programming an AI for civ (or other 4X games) as serious thought.)

Click to expand...

First, it's clear that I'm not a professional programmer. But there's no need to get personal in here. It's obvious that on a site called civFANATICS you will not be able to have discussions on a academic level. And while I don't have experience in programming, I would like to share the obversations I have made from playing these games. I don't feel like I made accussations, I try to look for solutions.

So you say that with fewer units, mistakes matter more. That for me doesn't tell me we should strive to have as many units on the map as possible. It rather tells me that we can keep the number of units only so low as 'the mistakes of the AI' allow it. I would nevertheless strive to have fewer units on the map, see congestion issues and all that.

The AI's problem in civ5 is not getting units on the map, but what it does with units once they are there. So, no improving reaction time is not going to do much for the AI.

Click to expand...

I often see that I can stay defensively while destroying the AI's army. Afterwards it's easy to destroy the single units the AI sends or produces. First, because it likes to send them one-by-one, and secondly because there's no place on the map for those units to hide. If you are under siege, you can't produce a new unit in the city because there's no place for it to go. This effect is apparent with naval units especially.

While 'getting units on the map' isn't the problem, it's forming a 'battle group' that can withstand a few turns of combat.

As for scripts, I am confident that certain 'battle group compositions' may help the AI more than the 'flavour' system. Obviously, there needs to be a bit of randomness added.

Seek said:

I think you misunderstand mitsho here. He's referring to the AI's boneheaded stubbornness in completing certain objectives in spite of changing situations. For example, an English AI has determined to attack one of your border cities, you bribe the German AI on the other side of England to help you in your war but England keeps pushing the original city despite the fact that the better play would have been to stay and defend it's flank.

Click to expand...

While I didn't mean exactly this, it does play into it. Another example, conquest AI's often try to attack the weakest civ. Which often means that they will attack across unfavourable terrain and over a middle-to-large-distance. That's why when Montezuma is your neighbour you build up your military as a deterrent. Most often however, the best way is to attack your nearest neighbour, it doesn't matter if his military is 10 % stronger than the Indias half a continent away. The strength difference is lost on the way and with you not being able to bring up reinforcements swiftly.

Regarding combat units, that means that the production of your army takes up a number of turns ranging from 10 - 90 (so, many turns), and that includes research, upgrade, production and moving them to where they matter. This long-term approach and need to build the units is a big challenge to the AI, don't you agree? I would prefer to eliminate that 'production' cycle. It is after all 'unrealistic' that it takes me 400 years in the beginning to give a few men some clubs who then continue to stay on like that for the next 3600 years... Wars (or unit movement) in civ are scaled wrongly to the rest of the game like researching technologies, building wonders and so on. In my ideal game, those would be decided in a few short turns.

EDIT: And again, I want to stress that I don't want 'units' on the map, but 'armies' that work more like the air units and their recon and bombard missions, so with range and stations.

Trias said:

Maybe read it again? I quite clearly explained why there would be no large stacks to exploit.

Click to expand...

Yeah sorry, my response was pretty glib in hindsight. I'll explain in detail what I meant.

You clearly explained that the programmer could tell the AI to try to avoid stacking it's units, but that would essentially defeat the purpose of the model, which is a simple change that would eliminate congestion issues and therefore get around the bad pathing rules or pathing AI in Civ 5. So let me qualify my earlier statement by saying "I can see step 3) in your model being highly exploitable against the AI without a major rewrite of the AI's current approach to moving units, ie pathing." If a major pathing improvement rewrite is a simple matter I think we would have seen more advancement in this area in the numerous patches released in the last few years, so my suspicion is this is a difficult prospect.

Let me explain the problems that would need to be fixed:

(I have no idea how well you understand the current AI and it's pathing so I'll attempt to explain myself as clearly as possible so we're on the same page.) When sending a unit across the map it currently selects the target tile and the "pathing AI" takes over (I'm not a programmer so I hope I'm getting my terms correct here). As anyone who's played the game for any length of time knows, the "pathing AI" is quite horrible in that it chooses one available path when the original order was given and doesn't deviate unless another unit enters said path (which is often), and then it completely interrupts the order and requests new orders. When the AI sends a force of multiple units across the map the problem is amplified, because each unit's pathing is completely ignorant of the other units' pathing and they frequently get in each others' way - thus when the force encounters a chokepoint we get the congestion problems. Not only this, but when the force is intended attack a city we get not only the congestion problems but problems where the AI just shuffles it's units around without attacking because it's still trying to place it's units correctly in the face of a changed landscape.

In your model with unlimited stacking and the "top" unit's death would mean death to the entire stack, the pathing would indeed be improved when it comes to transferring a force across the map and it would help the AI get into attack positions more easily, but could fail horribly when approaching a city or if ambushed by an unexpected force because the pathing of an attacking force is actually just the pathing of individual units attempting to land on a specific tile. (Your model could potentially harm the AI in this instance.) Because of this the AI won't use the unit effectively until it reaches it's target destination so the AI will - without significant changes to it's pathing protocols - inevitably end up with large stacks that are extremely vulnerable to enemy fire along the way. Thus my suggestion that "If the active unit on a tile is destroyed all other units on the tile take some damage as well" rather than "If the active unit on a tile is destroyed all other units on the tile are destroyed as well". I truly hope I've made myself clear.

You are misunderstanding whats been said. The suggestion was to change the game mechanics to decrease the number of units across the board. This in general works as to strengthen the advantage of human player. What you are talking about is replacing add bonuses from having more units to having stronger units, which is something I have advocated since the release of civ5.

Click to expand...

You're right, I took the comment out of context, but I'm happy to see we're in agreement regarding better units over more units.

mitsho said:

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.

Click to expand...

The rule addition is a means to an end: to provide incentive/motivation to encourage something I would like to see more of or to discourage something I would like to see less of.

In this case, I'd like to discourage ranged units being the foundation of ancient-medieval Civilization V warfare. That's not to say that I want to eliminate ranged units completely in that era, just that I want to give skilled players some benefit to diversifying their forces. If doing so means that "new hard rules" need to be implemented, then so be it.

Seek said:

(I have no idea how well you understand the current AI and it's pathing so I'll attempt to explain myself as clearly as possible so we're on the same page.) When sending a unit across the map it currently selects the target tile and the "pathing AI" takes over (I'm not a programmer so I hope I'm getting my terms correct here). As anyone who's played the game for any length of time knows, the "pathing AI" is quite horrible in that it chooses one available path when the original order was given and doesn't deviate unless another unit enters said path (which is often), and then it completely interrupts the order and requests new orders.

Click to expand...

I think you are confusing what the pathing AI does for human players, with how the AI itself works. The civ AI players loops through all units every turn reconsidering their orders. (At least it always did in previous interations of the game.) This probably with a weight for sticking to its "previous orders". (Note that even if it didn't, as you said the current pathing already interrupts pathing when the tile where a unit would end his turn is occupied.)

This makes it rather easy to add an negative weight to moves in which the unit ends on the same tile as other units. It is just as easy to make this negative weight increase exponentially with the number of units already on that tile. (i.e. moves where 2 units end on the same tile are not penalised as much as moves where 3 units end on a tile, while moves where 10 units end on the same tile are penalised by such a big amount as to be practically forbidden.

This allows the AI just that bit more "air" when it needs it to resolve congestion. Yet the situation where the AI clusters all units on the same tile is effectively avoided. (This is an example of the earlier discussion here about the AI being better at handeling "soft rules" - i.e. avoid stacking units - than at handeling hard rules - i.e. you cannot stack units.)

Thus my suggestion that "If the active unit on a tile is destroyed all other units on the tile take some damage as well" rather than "If the active unit on a tile is destroyed all other units on the tile are destroyed as well". I truly hope I've made myself clear.

Click to expand...

I'll try to explain why this suggestion would be counter productive. The "if the top/active unit dies, the whole stack dies" rule was included to ensure that stacking units is always a bad idea. However, if you change this to "if the top unit dies the others are damaged", then this automatically means that the tile is still defended after the top unit dies, and the opponent is prevented from entering the tile. This provides an incentive to stack units. (In fact, this could be abused to create invulnerable tiles by stacking more military units on a tile than can be destroyed in a single turn.)

The consequence is that the AI would have to be programmed to recognize when to stack and when not to stack. (Instead of simply trying to avoid stacking as much as possible). This likely not to bulletproof leaving much more openings for exploiting the AI, achieving the exact opposite of what you want to achieve with that rule.

I love where this conversation is going. Will it ever make it into Civ V? Probably not but I like it...

Seems to me that rather than an army system where you are capable of putting 3 units into an army and you would have several units kicking around separately or armies without the full compliment of units in it that there should be something similiar like a division type of unit where there are no other "miliatary units" just the divisions. Flavors could dictate what a leader put in the division but ultimately you would build the division unit and equip it with the types of units you wanted. (or you could have differnt types of divisions that are hard wired with certain types of units).

You wouldn't build separate units, you would build a division then everything associated with it would appear at the same time.

I am now thinking there should be like 5 parts to it. or perhaps that gets larger or smaller depending on what type of division you choose to build. An infantry division would have say 5 parts consisting of at least 2 melee units and then a mix of melee, scout, ranged or siege.

This would still allow all of the unit types currently found in civ, and the grapchics could still represent them but they would be put into fighting groups. You could upgrade your spreamen to pikes but the archers or catapults wouldn't get upgraded etc.

Two different types of movement. Strategic movement would allow you to move via roads rail or air greater distances to get to the battlefield. Tactical movement would be once there is an enemy within 3 or 4 hexes and would require units to slow down to one or 2 hexes per turn.

This would allow larger expensive more powerfull units than we have now, to move quickly to an area and then line up like we see currently and fight it out.

I wish some modder would just make a mod, which would put my head to rest.. I love civ5 - HAD IT NOT been for the 1upt-rule. I think it has to many downsides, and seem to chess-like. (Just my oppinion). However, I used to enjoy Kael's Legions Mod - unlimited stacking. I would stack units for easier moving, and would spread to receive my bonuses. (Not sure that everybody would, but personally I don't really care). It's very unfortunate that players are not atleast given the choice which the personally would like to play. Kael haven't updated his mod, and it seems it doesn't Work proberly under the new expansion-packs.. Would someone know if there's another mod which does stacking available though??

I believe some subtle changes like allowing great people to stack with workers and also allowing you to purchase a military unit first and then moving the older unit to some other tile(assuming there is room for it). Or in other words, the purchase option for military should only be disabled if there is absolutely no room for another unit. Also I believe there should be an option to automatically stack a escort unit to a civilian unit , then both units should move together in the pace of the slowest one.

Bottomline, I think 1UPT is awesome but there should be some game play improvement to avoid boring tasks.

rferrao said:

Bottomline, I think 1UPT is awesome but there should be some game play improvement to avoid boring tasks.

Click to expand...

I think this is a pretty good summary.

I think I still prefer to use a mod that allows for a small sized stacks in every tile... three seems to be the number i like.
Though, I'd be also be interested in that army system where you throw in different unit types, too. Could be used to encourage melee/range diversity...

The problem with stacks of three is a tile that fails to have three is inferior. This means three is the new one, you still could have crowding problems, and you lose any tactical feel because any unit weak defensively can just be protected by units strong defensively. To me, small stacks is the worst of both worlds, not the best of both worlds.