Is Cavalry Overpowered?

[pcasey]

The Franco/Prussian War, Prussian/Austrian war, American Civil War, and Crimean war were all dominated by massed infantry attacks. All these wars were fought between combatants with rifled muskets.

Cavalry was never a war winning battle arm after the late middle ages.

[Bad Ax]

pcasey:

Just because the unit looks like a Sherman (on which point I'm not particularly convinced) doesn't mean that it will have the unusually high performance characteristics of a Sherman. The average WWI-WWII tank broke down all the time. Even if I buy your point, the logistics easily explain the reduced speed.

You're somewhat right on the subject of the wars, except for the fact that the Crimean War rather famously involved the use of Hussar troops in battles. However, in those wars many of the tactical decisions were dominated by the ability of cavalry to make quick, highly damaging hit-and-run attacks. The success of the Confederacy in the Civil War's early years was due in no small part to their superior cavalry. Cavalry did not make or break a war, to be sure, but the relative ineffectiveness of infantry charges in the face of massed fire (Longstreet famously estimated a need for 3:1 forces, with an expectation of 66% casualties, for victory in such a charge) meant that cavalry played a critical role for detecting foiling infantry advances until their army's own infantry could entrench. That cavalry battles don't stick out in your mind when you think of these conflicts (though they should) is more because the opposing armies neutralized each other's cavalry with cavalry of their own than because the cavalry troops were just ineffective. Additionally, these conflicts involved infantry in a primary role because they were cheaper and easier to equip.

Sidebar: a great weakness of the resource engine is that a single resource is always enough to power your entire empire. If I have only one iron deposit, I can build 15 swordsmen units as easily as 1 and a single oil deposit can fuel 30 cities worth of cars and tanks. I think the game should limit this such that a resource can produce only a limited number of associated units and buildings at a time. So, 1 horses resource means you can build up to 5 cav simultaneously, and if you get 2 horses then you can build ten. This would also make resource trading more meaningful.

Back to the point. Cavalry should be just fine if you knock down the defense (relative to rifleman offense) and increase the price. Or you could exercise a little bit of self-control and use realistic tactics without modding them.

[ranald]

Originally posted by Comrade Tribune
I agree with Vel. The whole Horseman/Knight/Cavalry line is overpowered, because they are the Undead. Mass them, and you can overrun an entire civ, nearly without loss. The retreat ability is so powerful it defeats the concept of combined arms.

Actually it's just the opposite.

The retreat ability is so powerful that it *requires* the other
side to counter it with combined arms. Your damaged, retreated
cavalry would be dead meat if the AI could properly use mobile
units for local counterattacks. Or if the AI were smart enough to
keep a fast unit in each city... well.

[pcasey]

I disagree, even a smart AI isn't going to fare well against a mobile cavalry strike force. Counterattacking is all well and good if I don't take the city and I leave my cavalry sitting out in grassland for him to attack.

I won't though. I'll bring enough troops to take the city in one turn, then I'll move in infantry/riflemen/musketmen to defend the city, and move the cavalry inside the city.

A well managed cavalry rush never, ever leaves cavalry exposed to be attacked.

Alternately, if you want to play a pure cavalry rush and not bother with infantry, you can just take the city, pillage all the approach routes, and park your cavalry inside the city. The counterstrike force will have to slog across unimproved terrain to get to your city, giving you a chance to chew them up as they approach.

If you're careful, you'll never have to defend at all.

As for the historical usefulness of cavalry, I'm actually going to resist the temptation to get involved in a pedagogic debate. Against my ordinary nature, I'll just say this.

It doesn't matter if it was or was not the war winning arm historically. All that matters is how it plays out in the game. From a game balance standpoint, I think its overpowered.

If we were going to insist on strict historical accuracy, there'd be no way a swordsman would have a 50/50 chance of defeating a cavalry unit, or for that matter, no way an entrenched bronze age spearman could have at least a gambling chance to do some damage to a tank.

Ultimately, the only thing that matters is how a unit plays out in the game. Its a game about history, not a simulation.

[Velociryx]

One final point before I head home....all of the above assumes a "standard" war with "standard" game objectives, but it should be pointed out that an all cav strike force makes the ultimate "slash and burn" army. Swoop in faster than your enemy can react, raze his border towns, and leave (since a wounded cav is just as fast as a healthy one). With road pillaging, he'll never catch you before you can get back to your borders (where you've got troops just waiting to pounce if he follows).

Rinse and repeat as needed...OUCH!

-=Vel=-

[Bad Ax]

pcasey:

I, on the contrary, think what this thread is showing is that it's not the unit that is overpowered, but rather the AI that is underpowered. I agree that tweaks to cav are needed, but I think that the effect of the retreat is only important in light of an enemy civ that seems incapable of combining fast-attack and slow-attack units. Granted, balancing attack styles in the cities won't defend you against an overwhelming rush of cavalry units, but there's no guarantee that removing retreat, or anything short of giving cav a ridiculously underpowered attack, would get rid of speed-unit rush tactics. Retreat or no retreat, if I bring 20 cavalry up against an enemy city with 2-3 entrenched defensive units, I'm going to win. It won't matter whether the defense rating of the appropriate countering defensive unit is comparable or not.

After all, the AI has a simple expedient to prevent the cavalry rush, which is to build cavalry. It could do what the real life armies did and counter the deep threat of cavalry with cavalry of its own. That the AI fails to do this does not demonstrate that the unit has problems so much as that the AI has them. After all, if you thought that the AI could counterattack with an equal cavalry force, you probably wouldn't be so keen on the force concentration you describe. Of course, if you've managed to get infantry while your enemy still defends his cities with musketmen, then you likely won't be as worried about the counterattack. In that case, however, I think the gameplay is working out fine. You've gained a technological advantage, which yields a strategic advantage. What's to complain about?

[ranald]

Originally posted by pcasey
I disagree, even a smart AI isn't going to fare well against a mobile cavalry strike force. Counterattacking is all well and good if I don't take the city and I leave my cavalry sitting out in grassland for him to attack.

I won't though. I'll bring enough troops to take the city in one turn, then I'll move in infantry/riflemen/musketmen to defend the city, and move the cavalry inside the city.

A well managed cavalry rush never, ever leaves cavalry exposed to be attacked.

In that case, you must have such numerical superiority to the AI
that it doesn't matter anyway. Because if the numbers were
anything like equal, and the AI knew how to fight, this would be
happening *to* you just as often.

As someone else said, what it really demonstrates is that the
AI doesn't fight very well.

Alternately, if you want to play a pure cavalry rush and not bother with infantry, you can just take the city, pillage all the approach routes, and park your cavalry inside the city. The counterstrike force will have to slog across unimproved terrain to get to your city, giving you a chance to chew them up as they approach.

Interesting, but with your now culture 0 city, you have no
territorial radius to prevent the computer from building roads
right up to you, and using them to counterattack in force. That's
what I'd do.

Again, you have found a case where the AI doesn't know how to
react to a certain situation, but that doesn't mean a human
couldn't. It's not the unit, it's the AI.

If you're careful, you'll never have to defend at all.

True, but that has nothing to do with the merits of any individual
unit type.

As for the historical usefulness of cavalry, I'm actually going to resist the temptation to get involved in a pedagogic debate. Against my ordinary nature, I'll just say this.

It doesn't matter if it was or was not the war winning arm historically. All that matters is how it plays out in the game. From a game balance standpoint, I think its overpowered.

If we were going to insist on strict historical accuracy, there'd be no way a swordsman would have a 50/50 chance of defeating a cavalry unit, or for that matter, no way an entrenched bronze age spearman could have at least a gambling chance to do some damage to a tank.

Ultimately, the only thing that matters is how a unit plays out in the game. Its a game about history, not a simulation.

I agree with almost all of this, and think it bears repeating. The
one thing I disagree with is that I think cavalry isn't too powerful,
it just looks that way because it lets you do things the AI can't
seem to either imitate or counter.

[David Weldon]

Cavalry are not the equivalent of Riflemen, because it's quite possible to rush for cavalry and have them at least 5-6 techs before an equal comptetitor could have Riflemen. If the competitor spends time on the optional techs, it could be 10 or more techs before they get Riflemen. They must be viewed as compared with Musketmen.

The only real issue is that there _seems_ to be a general status-quo throughout the game where the best defensive unit defends better than the best offensive unit attacks. Cavalry are pretty much the only glaring exception to this, and the question is: does that mean they're broken?

Since cavalry are clearly intended to be used as an offensive force to capture resources that you learn about in the Industrial age (rubber, coal, oil), I think the fact that they're available _long_ before then is probably the problem, not their particular stats.

I think that retreating is very powerful, but don't forget that bombardment will allow your slow units to survive forever as well. They can even make up some speed because an artillery/musketman army won't have to spend _any_ time healing.

All the quick-strike arguments are definitely arguments against the AI, not arguments against Cavalry.

Increasing the cost of Cavalry won't really change much because people that abuse them pop-rush them anyway, and therefore won't be worried about spending a few extra captured workers to make them. Since they never die (almost), their actual cost is almost irrelevant. I would suggest that they are delayed until they can, in fact, be the contemporary of Riflemen and Infantry.

[Sevorak]

I just want to put in a word about how fascinating this thread has been. It's already inspired me to alter my (as-yet-unreleased) mod, especially in terms of the cavalry/rifleman issue. Thanks all.

The substance of the change, if anyone's interested, is the massive inflation of Military Tradition's tech cost, such that the tradeoff of going for it is delaying entry into the Industrial Age by a few techs.

-Sev

[Velociryx]

Query: If we're all pretty much in agreement re: Riflemen/Cav being the appropriate matchup, how 'bout moving Cav to Nationalism??

-=Vel=-

[zeace]

I agree that a well timed cavalry rush can easily destroy any AI civ, but these cavalry become relatively useless in the modern era. The comparable defensive units (Musketeers/Riflemen) both upgrade to Infantry against which Cavalry is much less useful. So while your Cavalry horde will be great at knocking out a few civs, your army will become next to useless if one of your oppenents learns how to upgrade all his defenders...

Ze Ace

[Sevorak]

Vel,

Move Cav to Nationalism and Scientific civs light everybody up at the beginning of the Industrial age. Correct me if I'm wrong, but I've NEVER gotten any tech other than Nationalism when advancing from the Middle Ages to the Industrial age. Letting Scientific civs pick up a good defensive unit right away is all right. Letting them pick up both Rifles and Cav, for free, as soon as they snap up the last middle age tech they need might just be a bit too much.

-Sev

[halley]

Shouldn't Pikeman get a bonus against fast units? Does Horseman/Knight/Cavalry count as fast units? Maybe the way to fix it is to give entrenched defenders in a city bonuses against Cavalry etc. I mean how silly it is, to attack cities with horses.

[Venger]

Cavalry - too powerful here, too weak there. A lack of HP differential means you can make the argument that they are too strong against tanks (almost a 50% chance to defeat a tank on attack - that's INSANE) and too weak against swordsmen (50% chance of losing). It's nice to compare them to riflemen, but keeping blinders on to how they relate to everything else in the game isn't helpful.

Cavalry movement should also be reduced to 2 (actually I think 2.1 would be good), or maybe 2 with treat all squares as road.

Keep in mind that I thought cavalry was too strong in Civ2 - I moved them from 8/3 to 6/3. They had the same attack as tanks (though only 2 HP instead of 3), and that led to cavalry rushes against tanks - gain, total drek.

Venger

[Nic]

Originally posted by Velociryx
Vs. Human opponents:
A little more uncertain, but for drawing first blood, it's no contest. If there ever IS MP, I promise you that you'll see "Mongol Hordes" when humans attack each other....mounted troops pouring en mass over your borders. Even if you have mounted troops in numbers yourself, the fact that they'll dominate MP games implies that they're superior in some way, and I think that retreat, and it's implications where capturing cities is concerned, is a major part of that.

Combined arms should dominate the game and probably will in MP unless something more like a killer-tactic can be found than pure mongol horde. Yes cavalry can over-run cities, especially if they significantly outnumber the defenders. On the other hand those cavalry are dead meat if they lose the battle against a human opponent because the counterattack will devastate 1hp units and the defender can use roads to rush all mounted troops within 5 tiles (8 if they have cavalry) in for the purpose.

Even where the cavalry have sufficient numbers to win against comparable defenders they are instant soft targets for counter-attack, a counter that will only face their rather weak defence of 3 and many units down to their last hp. A victorious combined force can use the newly captured roads to rush a few riflemen in, a pure cavalry horde is vulnerable even in victory.

Against the AI the cavalry horde is effective because the AI does not counterattack well and has clearly never heard of the exploitation phase of battle. Against a human opponent it will work gloriously sometimes and fail badly other times, it is a bold but desperate gamble.

One final point, the lifespan of cavalry swarms as a vaguely viable tactic lasts as long as your opponent does not develop railways. The counter-attacking power of railways is overwhelming against units so weak defensively.

--
Nic

[Thunderfire]

Originally posted by Bad Ax

After all, the AI has a simple expedient to prevent the cavalry rush, which is to build cavalry. It could do what the real life armies did and counter the deep threat of cavalry with cavalry of its own. That the AI fails to do this does not demonstrate that the unit has problems so much as that the AI has them. After all, if you thought that the AI could counterattack with an equal cavalry force, you probably wouldn't be so keen on the force concentration you describe.

Speed 3 is the problem of cavalry. Their defense raiting doesn't matter if
I am the one who is attacking. Speed 3 + railroads = overpowerfull first
strike. The AI will loose 2-3 cities when I attack him and I can fill these
instantly with my own rifleman. The AI has to deal with a dozends defenders
istead of the 2-5. The AI always counter attacks in my games with his own
units. Most of the AIs units will die against a well designed defense.

[Allemand]

It appears that the real problem is railroads, which are grossly overpowered.

I like cavalry the way it is, except the speed is just a little too much. If someone wants to change something, use the editor.

[Nic]

Originally posted by Thunderfire


Speed 3 is the problem of cavalry. Their defense raiting doesn't matter if
I am the one who is attacking. Speed 3 + railroads = overpowerfull first
strike. The AI will loose 2-3 cities when I attack him and I can fill these
instantly with my own rifleman. The AI has to deal with a dozends defenders
istead of the 2-5. The AI always counter attacks in my games with his own
units. Most of the AIs units will die against a well designed defense.

Then the AI is simply being poor at counter-attack. With a city radius reduced to 1 for being captured and no chance to pop-rush until the resistance is quelled you should be outnumbered and outgunned. An equal number of cavalry counter-attacking should massacre the exhausted troops outside the city (before you have a chance to move them again) and cause pretty severe carnage to the ones within the city. With working railroads to within one tile of the city your opponent should be able to do more than just manage equal numbers unless you have such an immense superiority that their defeat is inevitable anyway.

If you have dozens of attackers and dozens of defenders waiting in the wings then either your opponent should have had dozens of defenders or it was a mis-match from the beginning and the precise stats of the units concerned is only a detail. Against a smart player this should only happen against an opponent who is on the verge of defeat or using a builder strategy, in the latter case you should be worried about losing the city and army to culture (if you raze it you probably do not gain the road control to bring your infantry in so you can't afford to raze it unless you want to see your exhausted cavalry massacred).

To be honest I don't think we will see how this pans out in MP until we have MP. Even so I think there are plenty of balancing factors against aggressive attack in the game and so the attacking units have to be pretty decent to avoid a permanent (and very dull) state of trench warfare. I'd still like to see pikemen have a decent bonus against mounted troops so we can get classical musket and pike formations against those pesky cavalry attacks but even without this I think the rules on culture, resistance and control of roads limit the effectiveness of mongol hoard strategies. It is still a worthwhile strategy - as it should be - but I really don't think its a game-killer.


--
Nic

[Pyrodrew]

After reading the posts here & considering some recent games... I strongly agree Cavalry is too fast, 2 would be better for game balance. Specifically, because you can attack cities deep in an enemy's culture border. Don't even need railroads! They can travel on 2 roads (yours or neutral) then 2 enemy tiles then attack. Railroads are gravy.

Deadly Combo = Cavalry + Democracy.

Cavalry is so fast you don't need to have any units sitting in enemy territory (war weariness). 10 Cavalry have no problem taking over 2 Infantry... if you send tanks (1 turn longer while sitting in enemy territory) the AI often drafts 4 Infantry units (in Communism) making it harder to take that city. Once you take the city you can send defenders for your wounded valiant cavalry since you now own that territory/roads & end your turn with your cavalry on your new territory... or round up some more cavalry to take the next city.

[David Weldon]

Regarding moving cavalry to Nationalism:

I don't mind increasing Scientific in this way, for two reasons. First, it's a poor attribute to begin with, not the worst, but definitely not in the top 3 either. Second, the advantage that they gain will only be for 1 tech (normally 4 turns by then). That's not a huge time advantage. The way it is now, the uncontested rule of Cavalry can last for 10 or so techs.

The problem I see is that if Nationalism gives Cavalry, and everyone generally agrees that armies are extremely inefficient in a closely contested game, then Mil. Trad. becomes a non-tech, as there's no reason to research it just for the Pentagon.

Then again, maybe that would free up another tech space for something more useful anyway...

Other Thoughts:

All in all I think we should consider nerfing the unit that is currently called Cavalry, and then adding a unit very similar to it that becomes available later (around nationalism or steam power), to be used in the accquisition of strat. resources in the Industrial Age.

I have to admit that I like the idea of making Mil. Trad. way more expensive to encourage players that missing out on all the wonders of that era isn't worth it. Go back and research it _after_ you have steam power and have found that you need to go get some coal...

[Comrade Tribune]

I still insist that the real problem is a combination of three things that, by their cumulative effect, unbalance the game:


1- Retreat ability makes fast units immortal

2- Forced Labour is too powerful

3- Walls/City defense are too weak

[ranald]

Originally posted by Thunderfire


Speed 3 is the problem of cavalry. Their defense raiting doesn't matter if
I am the one who is attacking. Speed 3 + railroads = overpowerfull first
strike. The AI will loose 2-3 cities when I attack him and I can fill these
instantly with my own rifleman. The AI has to deal with a dozends defenders
istead of the 2-5. The AI always counter attacks in my games with his own
units. Most of the AIs units will die against a well designed defense.

If you can successfully defend against cavalry (when the AI
counterattacks), why couldn't the AI defend against your
cavalry in the first place?

The difference is not in the cavalry- both you and the AI are
using cavalry to attack, right, but only one of you is getting
anwhere, why is that?

The basic problem is that the AI doesn't fight well. It's better
than the Civ2 AI, but it still has a way to go.

[ranald]

Originally posted by Comrade Tribune
I still insist that the real problem is a combination of three things that, by their cumulative effect, unbalance the game:


1- Retreat ability makes fast units immortal

2- Forced Labour is too powerful

3- Walls/City defense are too weak

I think you're on to something here, though I think the real
problem is 2. Forced Labor is too powerful.

They made it easier to grow your population. As a counter to
that and to the ICS it could easily lead to, they introduced
corruption.

However, there's a backdoor. You can now convert population
into production. And to top it off, that population is immune to
corruption. So now you have a situation where it's easy
(too easy?) to increase your population, and you have an easy
way to bypass the one restriction on "excess" population.

The output from Forced Labor should be subject to corruption.

[jbrians]

Originally posted by Comrade Tribune
I still insist that the real problem is a combination of three things that, by their cumulative effect, unbalance the game:


1- Retreat ability makes fast units immortal

2- Forced Labour is too powerful

3- Walls/City defense are too weak

I think it would be a good idea to fix 1 OR 3, not both. Expansionist war is already fairly difficult in this game. If both 1 and 3 were changed it would be impossible.

-Brian

[pcasey]

I have to disagree strongly with the assertion that expansionist war is hard in Civ III. On the contrary, I find it way too easy to rush an AI civ with a swarm of fast units.

Once you discover the awesome and abusable power of pop rushing high speed attackers, defense is impossible for the AI.

If you've turtled and its not the late modern age and you're surrounded by high culture size 25 cities, a pop rush conquest stategy will be tedious because of the propensity of cities to revert. There's an easy solution to this. Scorched earth. Raze everything, conquer nothing, plant one of your settlers where the city used to be next turn after the entire AI civ has been obliterated by your horde of modern armor.

I at least would strongly support massive as in doubling, the defensive impact of walls and size 12+ cities. I'd also support walls being addative with city size. Plenty of modern cities had massive ring of fortifications making them harder to take in practise than there mere size would have offered e.g moscow/leningrad in WW II.

As it is defense is basically impossible in this game. The reason most players don't tend to run into this issue is that we play against the AI which I happily agree is horrible on the offense.

The problem that the AI can't run an offensive merely serves to mask the core gameplay flaw that defense is impossible against a fast unit rush.

Since the AI can't attack, we as players don't find ourselves having to deal with the impossibility of facing a cavalry rush. I guarantee you, not even a human player sitting across from me can defend 10 cities with 50 riflemen if I have 30 cavalry. He'll lose 3 or 4 cities on my first attack round, basically guaranteed.

Of course he'll counterattack and I'll lose 3 or 4 cities, but that's the problem in a nutshell. Neither of us can defend ourselves.

Defense is impossible.

[rid102]

I'm quite surprised no-one has mentioned Dragoons here.

I think Cavalry are actually underpowered because of some short-sightedness in their introduction. Cavalry are available pre-Industrialisation, post-Musketmen when really this kind of role is much better suited to Dragoons (a la Civ2). Hence, Cavalry in Civ3 are expected to last until around Infantry/Tanks (WW1) yet are given to the player post-Industrialisation. So, you have a unit that is relatively powerful when it's introduced but hopelessly outpaced later one (when historically it would be coming into its own).

Cavalry should be slightly more powerful than it is now but importantly shouldn't come until later on (around post-Riflemen). Although I do agree that it should only have 2M, so would be more like 7/4/2 or 8/4/2.

Dragoons should then fill the gap between Knights and Cavalry at something like 5/3/2 and also gives better unit matching through the ages:

i.e.

Spearmen->Pikemen->Musketmen->Riflemen->Infantry

Mounted Warrior->Knight->Dragoons->Cavalry->Tank

So each mounted unit type also has an infantry unit type counterpart.

[Nic]

Originally posted by pcasey
I have to disagree strongly with the assertion that expansionist war is hard in Civ III. On the contrary, I find it way too easy to rush an AI civ with a swarm of fast units.

In my (still rather limited) experience the AI is over-reliant on static defence and a bludgeoning counter-attack. In neither regard is it particularly smart. Swarming a dumb and predictable opponent is more to do with the limitations of the AI than the power of any particular unit in the game.

The most obvious sign of its stupidity is the habit of building roads and even railways right up to the border with another civ, roads that the human player is going to use to get all those supporting infantry into its captured cities. A smart player would build a minimum of border roads/railways and in defensively weak areas would place forts on those that are needed (for trade or resources). A smart player also places cities as much for defence as for growth, a city on a hill or the other side of a river is significantly harder for a cavalry rush to over-run and as we have discussed a failure to win the battle will often result in the loss of the cavalry army. Even if the cavalry win, their footsoldiers will be lagging behind without roads and the counter-attack should hammer the defensively weak cavalry unless they outnumber the defending civ by a wide margin.

If there is a problem here it is in the ability to pop rush large armies in short times. There has been plenty of discussion on that subject and I agree that it does make attack/pillage strategies very powerful. I still think that the defensive advantages of road usage, culture (and forts) should allow less despotic play-styles to hold out against the onslaught but we will only know for sure when we have MP.


--
Nic

[Thunderfire]

Originally posted by ranald


If you can successfully defend against cavalry (when the AI
counterattacks), why couldn't the AI defend against your
cavalry in the first place?

The difference is not in the cavalry- both you and the AI are
using cavalry to attack, right, but only one of you is getting
anwhere, why is that?

I can sucessfully defend because I am prepared for war.
Even the human player will be overwhelmed when the AI
starts a war and attacks with alot of speed 3 units. The
Chinese got 5 of my towns because they attacked me with
several dozend riders. They lost 4 units I lost 30...

[rid102]

Even the human player will be overwhelmed when the AI

Speed is a much bigger deal than before because of restricted movement in enemy territory and only being allowed one attack per turn (most of the time). This makes any M3 unit very dangerous.

I may try reducing Cavalry to M2 and seeing what effect that has.

[Comrade Tribune]

I think there is also a cost/effectiveness mismatch between Cavalry and Infantry. I am just now creating a Mod where Muskets and Rifles are reduced in price, while Cavalry is up to 120 shields!

Supply and Demand should regulate unit prices.