[R&F] The State of Combat Units (Post Spring 2018 Patch)

[Sostratus]

Now that the spring patch has boosted the 3 medieval UUs of Norway, Georgia, and Japan, I thought it might be time to finally make a post I've been thinking about. A lot of discussion on the same units occurred before the full patch notes dropped; I'm feeling lucky! I can't lose!

For starters, it's good to be quantitative in a topic full of numbers. Let me make some remarks on unit lines in general. It appears to me that the game was originally designed on the following principles:

• Unit strength, within a class, increases by 10 per era
• Unit classes have semi-persistent characteristics to shape a contemporary battlefield

The first is not necessarily intuitive unless you actually start charting out unit strengths. The second becomes apparent once you've charted. I'm primarily going to focus on things relating to unit strength and not movement/ZoC, an entirely separate topic.

Down the rabbit hole we go!

First, let us look at the unit classes themselves and see if we can ferret out that +10 strength per era.

And the raw data for the graphs:

Now, before you kids get uppity about me tossing out colorful shapes all willy-nilly, let me explain:
The Y axis in every graph is combat strength. The blue shape in the backgrounds is the trend line for the unit class. It's a line sloped at 10 strength per era. The axis is spaced by era, the units themselves are named. I consider the atomic and information era to be the same because no unit has an atomic and information era representation (other than maybe air units?)
I included a few unique units because people often suggest filling in gaps in the unit tree in certain spots. Unique units all have the same characteristic: they have the template of a base unit, plus special attributes. So from this, "Khevsur" is really a "Longswordsman" (to borrow from Civ5) with a hills bonus. Redcoats and Gardes Imperiale are really "Rifleman" units with continent based bonuses. Domrey=Trebuchet, Rough Rider= 'Cuirasser', Malon="Lancer."

As can be seen, the 10 strength per era rule is almost perfectly followed. The two types of deviations:
Ancient era units and the "post industrial shift."

Ancient Era units: there's 3 units that seem to buck the trend without bucking the trend. To me, that's the slinger & warrior, as well as the archer. The slinger and warrior are available to everyone, turn 1. They are almost "proto-ancient" units, and they seem to share the attribute of -5 combat strength from a 'regular' ancient era unit. Archers, to prevent from being either totally useless or completely OP, sort of straddle the ancient and classical eras. For this reason, they sit right between that 20 and 30 ranged strength that the trend line suggests.

The post industrial shift refers to, essentially, artillery becoming dual field/siege units, and mounted forces becoming armored (tanks.) As I'll touch on later, Heavy Cav goes from being essentially +3 strength over contemporary footmen to becoming armored beasts with a +10 strength advantage, although mech inf. close this gap a little. Melee is pretty consistent through the ages, though.

Now that I think I've established the +10/era relationship, let's look at all these units in aggregate and see if we can't divine some class-specific characteristics.

What a mess. Do you like tables? I like tables- let's look at a table of the raw strengths and a table of deviations from a trend line!

Now that's a little more enlightening.
What conclusions can we draw? Well, the average deviation is -0.5. That tells me this trend line- the "contemporary unit" - is a pretty snug fit.

• Melee: these boys seem to fit almost perfectly. They are the standard unit of civ6, so this isn't surprising. My note about warriors/slingers is why i think they are at -5. Swords were 35 strength on release (they never changed the legion or Ngao with that patch - hence Ngao is stuck at 35.) In the late game, I think they were intended to slip a bit so tanks and other units could be viable against all infantry armies.
• Anticav: These guys really get the short end of the stick. Not only are pikes terrible, but AT units stink as well- their bonus only matches them to armored units, it doesn't give them a strength advantage!
• HeavyCav: early on they have +3 strength as a sort of "top dog" perk. Which is fine- historically these units did trample everyone underfoot. Later they get bumped up when other units - melee, anticav, and ranged- fall behind, making them incredibly dominant for their production cost.
• LightCav: I'm not sure why this class exists at all, but they go from having one unit in the first part of the game (Horseman) that is just incredibly strong, to being about -3. I guess as "light" cav intended for raiding, they can withdraw from harm, so that's not so bad.
• Ranged: if we recall my point earlier about archers, it looks like ranged units all have -5 strength vs their contemporaries. Except machine guns, which are the red headed step child no one wants to fix, it seems. They are mostly obsoleted by air units and artillery late game, though. Ranged units all universally feature a melee strength 10 points below their ranged strength. This means they take 50% more damage from melee attacks.
• Bombard: Bombard units fit well early game, then become doom engines at the end. I think this is intended to give them a use case over bombers, and keep late game warfare form being stagnant. Note that bombard units feature -12 melee strength vs their bombard strength, with artillery and rockets having -20. This corresponds to taking 62% and 125% more damage from melee attacks, FYI.

Okay, so, that's how unit strengths stack up. But these units also cost production, which certainly comes into play. Let's dig into that now.

I'm repeating what I did in this post . Read it to understand methodology. The gist is that given strength & cost, two units can be compared to determine how efficient an upgraded unit is over its precursor. The important point about strength is that in combat, +X strength means both +Y damage dealt and 1/Y less damage received. So strength confers advantages on two sides of the coin- hence a unit that had +17 strength does double damage and takes half, resulting in 4x the combat effectiveness. "+17 strength, you say? That's what an Army formation gets, but it only costs 3 units for the power of 4?!" Exactly- that's why Zulu is OP and military academies are extremely powerful production structures.

Anyways, let's go back into the land unit lines:

All unique units are compared to their predecessor, but in all cases except the khevsur, the next base unit is compared to the earlier base unit. E.g, tanks are compared to knights and not rough riders. I compare muskets to khevsurs because the recent patch.
A few other points:

• Before the patch, the khevsur was at 0.69 efficiency. It was literally worse than building swordsmen. This was the only case in the game of an efficiency of less than one. They are now worth building, although they aren't brutally better than swords are- perhaps fair for most civs not having medieval melee.
• Samurai now stand at 1.47 efficiency compared to swords. They are basically less mobile knights that upgrade on the melee line.
• If pikeman were buffed to 45 strength and 180 cost, getting a similar treatment to what the other Military Tactics units got, then they would have a 1.79 efficiency score. IE, they wouldn't totally suck.
• Crossbows might also need a cost reduction, although the power of ranged suggests they are okay.

For those who are wondering, if buffed pikes would make knights to vulnerable, consider this: Pikes would be as effective against knights as muskets are. Given that you need to grab a leaf tech, I think that's fair. Knights still would hit melee, ranged, and bombard units incredibly hard. Knights need a counter that actually counters them properly- instead of forcing those without iron to wall up and watch the crossbow bolts ping off their chivalrous armor.

~~~~~~
Wow, what a long winded post. Please, share what insights you glean, things that just seem off playing through the game that may have a basis in the data presented here, or anything else that comes to mind!

 

[UWHabs]

You can definitely see the pike is in a screwy place here. I'd guess if you changed the pike and crossbow to both be cost 160, that would probably balance things out a lot more, right? Then as expected, the Knight becomes the most expensive medieval weapon, however it is the strongest of them.

 

[Trav'ling Canuck]

Excellent analysis.

Now, who wants to build on it by bringing in the relative benefits of:

• Differing promotion trees (Commando, anyone?); and,
• Differing production cost reduction cards.

EDIT: Forgot: Also differing government combat bonuses by unit type

 

[Eagle Pursuit]

@Sostratus Did you address the change to efficiency in the Berserkers? They had a cost reduction in addition to buffs to the attack strength and defense penalty.

 

[Flaxton]

Maintenance costs should be looked at. Why cavalry is the same cost as melee or anti-cav to maintain? To nerf cav I would a) remove tile defensive bonuses b) hamper them against walls (and don't allow rams or siege towers for them.) and increase the maintenance. They are a hard hitting unit that should be used sparingly and in open terrain.

 

[Sostratus]

Did you address the change to efficiency in the Berserkers? They had a cost reduction in addition to buffs to the attack strength and defense penalty.

Berserkers are a funny unit- nominally they are identical to Khevsur in that both cost 160, and have base 45 strength; but in practice a berserker is either dealing +10 or -5; when they were at +/-7, it would even out over many combats, but they've definitely gotten a little boost from that as well. I'll have to do some figuring to work out just how to account for that odd bonus.

Maintenance costs should be looked at. Why cavalry is the same cost as melee or anti-cav to maintain?

Initially with this post, I wanted to address the "top line" differences between unit classes- their strength. It's the starting point for balance.
To really flesh out that second leg of balance, though, and especially account for movement, maintenance, resource needs, etc, there definitely needs to be a hard long look at class characteristics in the vein of what you are referring to. Ideally, one needs to start with a design concept for the unit classes- what each of them are intended to do and why- otherwise we end up with a numerical soup that prioritizes everything but fun. Which is perhaps the most important part!

 

[pgm123]

You can definitely see the pike is in a screwy place here. I'd guess if you changed the pike and crossbow to both be cost 160, that would probably balance things out a lot more, right? Then as expected, the Knight becomes the most expensive medieval weapon, however it is the strongest of them.

I was about to say the Pike and Shot wrecked the curve.

 

[Boris Gudenuf]

Excellent Analysis!
From a military historian's standpoint, if I may, when it comes to 'tweaking' the units and their relationships, there are several places where the Civ VI factors/relationships are simply historically inaccurate, and possibly correcting those inaccuracies would correct a lot of the perceived problems with the units.
1. Heavy Chariots are historically overpowered. While there are not a lot of reliable 'battle narratives' from the period of chariot warfare, it is interesting that there is no account anywhere, as far as I know, of melee chariots beating disciplined close-ranked spearmen. They could literally run over lighter troops, like archers, slingers, scout-equivalents, and even unarmored (bronze) swordsmen, but against spearmen, not so much.
2. Pikemen, at least the late Medieval/early Renaissance sort, were relatively lightly armored and Cheap. In the game, their initial cost is much higher than it should be compared to Knights, or even compared to Swordsmen - mail or plate armored like a late-Roman swordsman represented far more expense than the pikeman's simple helmet, and the training time required to keep swordsmen effective is Massive compared to pike drill.
3. Just for a moment, let's look at the 'timeline' for some units:
*late 11th century: the couched-lance armored horseman, the Knight, appears in Europe. We know the Normans were not couching their spears at Hastings in 1066, but by the end of the century Normans in Sicily are and Byzantines are commenting about the 'ferocity' of their charge.
*late 12th century: the steel-spring crossbow, technically the 'Arbalest' appears in European illustrations - less than a century after the knight, and preceded by a 'heavy prod' wooden crossbow that was armor-piercing, so the Knight and the Crossbow are essentially simultaneous-appearing units.
1302 to 1315 CE (beginning of the 14th Century). In three battles, Pikemen demonstrate dramatic superiorities over knights:
*1302: Battle of the Spurs - Flemish militia pikemen hang up 700 pairs of spurs from slaughtered knights after the battle
*1314: Bannockburn - Scots half-pikes (12 ft spears held in both hands, more maneuverable than 'true' pikes) massacre English knights
*1315: Morgarten - Swiss pikemen charge into and hack to pieces a force of knights.
In other words, Knights have at most about 200 years when they can romp and stomp over spearmen and swordsmen, before pikes (re)appear on the battlefield and start spitting knights like squabs on a stick.
*1375 - 1411 CE - The first Bombard appears in France battering down a city wall, then in 1411 the first cast iron Bombard firing a 400 kg stone shot appears, and the 'Hackbus' appears, the first effective hand gun with a trigger mechanism BUT it is primarily a 'wall weapon' not used in the field.
*1470 - 1475 CE - the 'hackbuss' or 'hook gun' gets a shoulder stock and a matchlock trigger mechanism, becoming the Arquebus, the first effective gunpowder firearm for infantry in the field.
*1480 CE - the 'true cannon' appears in France: cast bronze, with trunnions and a trailed carriage for easy moving and aiming. It remains essentially the same for the next 350 years.
*1493 CE - 'Colunelas' or Columns appear in Spain, combining arquebusiers, pikemen, some halberdiers and crossbowmen - the first 'pike and shot' formations.
*1520 CE - the 'Musket', a heavier arquebus that can penetrate a knight's plate armor, appears in Europe
*1530 CE - Spanish Colunelas are combined into Tercios, each with a 50-50% mix of harquebuses or muskets and pikemen.

In other words, no matter which 'effective' date you use, there really is no such historical unit as a separate unit of Musketmen: virtually as soon as the arquebus or musket appeared on the battlefield, it was combined with pikemen into some kind of 'Pike and Shot' formation.

Next, a look at Initial and Maintenance Costs:
1. ANY mounted unit is at least 3 - 4 times more costly to form: horses, especially horses big enough to haul chariots or carry men and armor, are expensive to breed and feed. A knight's heavy 'destrier' or war horse could cost as much as all of his weapons and armor combined! In addition, men must be trained to ride the horses, in addition to the cost of equipping them with weapons and training them how to use those weapons.
2. The Initial Cost of some later weapons is actually Less than the earlier ones. For example, the early arquebus/musket cost about 1/3 as much as a breastplate, helmet and steel-pointed pike cost, so Initial Cost of a musket-armed unit was actually lower than the earlier pike or sword-armed unit!

Maintenance Costs:
1. Swords take lots of practice: ANY sword-armed unit will have higher maintenance costs than a pike or spear-armed unit, where the training is mostly close-order drill rather than weapons practice.
2. Cavalry requires constant training for both men and horses, and feeding of both men and horses, so any mounted unit will have a Maintenance Cost 2 - 3 times higher than a similarly-armed unit of infantry, even accounting for the fact that the mounted units are generally smaller than infantry ones (Example: a Napoleonic (Industrial Era) Cavalry Brigade was 1500 - 2000 men and horses, an infantry brigade was 3500 - 5000 men).
3. All 'gunpowder' units, of course, have an on-going Maintenance Cost for powder and shot - ammunition, which has to be manufactured and supplied, unlike pre-gunpowder units which for the most part did not require constant resupply of weapons.

(Tentative) Conclusions:
1. The balance between Swordsmen and Spearmen can be addressed by higher Maintenance and Initial Costs for the Swordsmen - it is much more expensive to maintain a bunch of swordsmen than spearmen - just ask the Romans.
2. Pikemen should be much, much cheaper to Produce and Maintain than either Swordsmen or Knights, and much more effective versus Knights than they are now.
3. There is no such thing as a separate Musketman Unit: the first gunpowder weapons on the battlefield will be combined arms units like Pike and Shot, slightly preceded (1480 - 1530 CE) by the first Field Cannon, and preceded by a century or more (1375 - 1530 CE) by Bombards.
As an Aside, correctly, Bombards should be Support Units. They took hours to load and aim, so had no effect on Units, only Walls and Cities, which they flattened.

 

[acluewithout]

@Sostratus Excellent analysis and write-up. Seriously good stuff.

A few thoughts from me:

• From your analysis, you can see why a Swordsmen rush is so effective. Not only are your Swordsmen one ear ahead of Warriors, but Warriors are a little underpowered meaning there's an even bigger melee strength difference. Combine this with how easy it is to get to the Swordsmen tech - build three to four warriors, use these to kill three barbs (Eureka for Bronze), and then if you have iron, build an iron mine (Eureka for Ironworking), then finish up Iron Working tech and upgrade your initial warriors (who are likely now promoted anyway).
  
  
• Units which deviate negatively from the trend seem to be units without resource requirements - specifically Anti-Cav, Ranged, Warrior and Infantry, Helicopters and Cavalry (this last is a special case, because they do technically require Horses, but Horses are visible from the ancient era and used for Horsemen, so this is a kind of 'light' resource requirement). I assume this is intentional, i.e. players that have secured strategic resources gain an advantage.
  
  
• I think part of the Anti-Cav weakness may also reflect Anti-Cav being a defensive unit (their promotions certainly suggest they are more defensive). This is because the game needs to balance out the light edge to defending units have, because they can fortify, heal faster, and be reinforced more easily.
  
  
• You excellent analysis of course doesn't yet look at science cost, promotions or oligarchy bonuses. Looking at science, I've posted elsewhere how players may want to bee-line down the 'district' tech tree to get campuses, commercial hubs, industrial zones and universities, rather than go down the offensive unit and defensive infrastructure branches (respectively, archer-horsemen-knight, and mining-bronze-x bows, walls etc). If you're making that bee-line, then Pikes represent a quick and more powerful unit you can grab that doesn't side track you into researching more extensive military tech.
  
  
• Great in principle, but it breaks down for a tonne of reasons. The first problem is that, although Anti-Cav are a little underpowered (which I think is okay for a defensive unit), Pikes are flat out over priced, and that is a real problem. Weaker units are not inferior if they are priced correctly; but they do become inferior if overpriced.
  
  
• The second problem is that it's way, way, too easy to research and get to Swords and Knights, so those units just aren't enough of a detour to make the extra science a problem. This is partly to do with the tech trees and eurekas, and the fact Swords and Knights both share the same strategic resource (meaning that getting resources for both units is overall easier) but is more fundamentally because you can unlock the Campus so early and then spam them, allowing you to push all the science. Combine that with the fact that the Commercial Hub and Industrial Zone aren't actually that strong, and Bee-lining the district branch rather than military just becomes a weaker and weaker strategy.
  
  
• On the second and third issues - promotions and oligarchy - I think this is tricky to account for. Probably you should just take the average of each units first two promotions, because it's really only the first promotion that's 'easy to get'. You'd probably have to make some assumptions about what eras a player can realistically benefit from oligarch bonuses. My guess though is that promotions and oligarchy probably result in melee being much stronger than your table suggests, particularly now they can maybe get +1 movement.

Possible rebalancing;

• I'm not a fan of any solutions which radically change how units or combat works, or how the tech tree works. But I think a few tweaks could really help.
  
  
• Anti-Cav need to be cheaper. I'm really surprised the Spring Patch didn't address that, but regardless, it seems like something ripe for a community patch.

• I think Military Tactics can stay, and could even stay a leaf tech (although it would be better being a pre-requisite for something). But the tech needs to have more value. That could be another unit, a general unit or yield buff (eg how Military Tradition unlocks flanking and support bonuses) and or making unlocking Military Tactics and or building or using Pikemen a Eureka for some other key technology.
  
  
• Ideally, Campuses should be harder to get, but I don't see any way to achieve that without doing serious damage to the tech tree. But buffing Commercial Hubs and Industrial Zones (or their related techs) might help a little. Perhaps making the Eureka for Campuses a little more tricky would help too.
  
  
• Heavy Cav, and Knights specifically, need to be harder to get. I think part of solution may be to exclude Heavy Cav from %production cards and from the Professional Army cards, so players are forced to invest more to get these units. Likewise, Heavy Cav should maybe have +1 maintenance over contemporary units, making you invest more in maintaining them. Putting Knights on a leaf tech would also help, although that might take some serious re-jigging of the tech tree, as would making the Eureka for Stirrups more difficult to get (it's quite easy to rush Feudalism because you can capture rather than build farms). A possible Eureka for Stirrups or Feudalism might be to build at least 2 or 3 Medieval Walls, or indeed have built x number of Pikes. Lastly, a more radical solution would be to only allow Heavy Cav to be built in cities with Encampments and or walls (that's my least favourite solution).
  
  
• Not so much a balance issue, but I wonder if infantry should require oil. I get why they don't, because from the modern era onwards the game sort of starts relaxing some strategic resource requirements, but I feel like infantry are a slightly too early point to start that relaxing.

 

[Trav'ling Canuck]

Boiling down these past few posts, here are my three take aways:

• Anti Cav should cost less to produce and cost 50% less maintenance than other units of the same era
• Light Cavalry should cost 50% more maintenance and Heavy Cavalry 100% more maintenance than other units of the same era
• either eliminate leaf techs entirely, or find a way to put Knights and Swordsmen on them; neither massed Swordsmen nor massed Heavy Cavalry were universally used globally, and should require as much of a science detour as Pikes if you want to grab them

 

[Temppu]

Boiling down these past few posts, here are my three take aways:

• Anti Cav should cost less to produce and cost 50% less maintenance than other units of the same era
• Light Cavalry should cost 50% more maintenance and Heavy Cavalry 100% more maintenance than other units of the same era
• either eliminate leaf techs entirely, or find a way to put Knights and Swordsmen on them; neither massed Swordsmen nor massed Heavy Cavalry were universally used globally, and should require as much of a science detour as Pikes if you want to grab them

These takeaways are pure gold. Even more so with the +1 movement Commando, Id like to see very costly (Production and Research-wise) and mobile offensive Melee and Cavalry classes, and slow but very very cheap defensive Anti-Cavs.

The also strong Ranged units havent been much discussed here yet. I feel Ranged units should be more terrain dependent. Being able to shoot from Forest is somewhat strange. Im also thinking reducing range to 1 in open terrain and only being able to shoot with range 2 from Hills would be a nice balance and tactical tweak.

 

[acluewithout]

...The also strong Ranged units havent been much discussed here yet. I feel Ranged units should be more terrain dependent. Being able to shoot from Forest is somewhat strange. Im also thinking reducing range to 1 in open terrain and only being able to shoot with range 2 from Hills would be a nice balance and tactical tweak.

That’s a cool idea, although the AI might struggle.

I’m not sure ranges are as strong as all that on higher difficulties. They’re actually quite squishy.

 

[kb27787]

That’s a cool idea, although the AI might struggle.

I’m not sure ranges are as strong as all that on higher difficulties. They’re actually quite squishy.

Nerf cavalry that hard and you will see nothing but ranged units I guarantee you, since their main counter is gone. People stll will not build pikes due to them getting shredded by crossbows before they can even get close.
Ranged does not need toughness once they hit critical mass. Esp in a world without 4 move cavalry.

 

[acluewithout]

Well, yes, they’re “strong” if you just spam them. I try to keep my armies lean, and find they’re okay if you have two or three and some melee. If you make a tonne it’s a different story.

I agree Anti cav are weak on Offence: slow, lack real punch, and no promotions to defend against ranged. But I think they’re actually pretty good on defence (save for cost). I’ve posted about this before somewhere... If they were cheaper, and MT was more attractive, I think people would be more likely to build them (at least if they’re not playing aggressively).

 

[Trav'ling Canuck]

Boiling down these past few posts, here are my three take aways:

• Anti Cav should cost less to produce and cost 50% less maintenance than other units of the same era
• Light Cavalry should cost 50% more maintenance and Heavy Cavalry 100% more maintenance than other units of the same era
• either eliminate leaf techs entirely, or find a way to put Knights and Swordsmen on them; neither massed Swordsmen nor massed Heavy Cavalry were universally used globally, and should require as much of a science detour as Pikes if you want to grab them

I've suggested this elsewhere, but in light of some of the comments, let me add to the above:

• All Ranged units should have an attack range of 1, but receive a free attack against the first unit to melee attack them each turn. This would better reflect their historic use as primarily defensive units and make the longer range Bombard units more important for supporting city attacks.

Also to some above comments, I don't think what I've suggested is much of a nerf to cavalry. The extra maintenance is more of a nuisance factor to encourage them to be part of your army, not your entire army.

 

[tedhebert]

@Sostratus , awesome post !!!

Lots of really fascinating info to analyse and digest in here... Thanks

 

[tedhebert]

@Sostratus

• Units which deviate negatively from the trend seem to be units without resource requirements - specifically Anti-Cav, Ranged, Warrior and Infantry, Helicopters and Cavalry (this last is a special case, because they do technically require Horses, but Horses are visible from the ancient era and used for Horsemen, so this is a kind of 'light' resource requirement). I assume this is intentional, i.e. players that have secured strategic resources gain an advantage.

Man... I was SO certain that helicopters needed Aluminum... prepared this message to kindly chide you on this, then just to make sure checked wikia... sometimes we think we have all the details in our heads, but ;-)

 

[Turrdy]

So does that mean I'd need in average 11 knights to kill 1 tank?

And Efficiency isn't very clear to me. Your other post states:

Efficiency is the Combat effectiveness vs the production cost ratio; Combat efficiency * (Cost_New/Cost_Old).

Which makes no sense to me, since that formula would suggest that the efficiency should always be higher than effectivness since (Cost_New/Cost_Old) is always greater than 1 - because modern units don't get cheaper, I suppose.

 

[Sostratus]

And Efficiency isn't very clear to me. Your other post states:

That is a typo in the old post. It's Combat effectiveness divided by the production ratio. Great question though!

What "Efficiency" captures is the idea that upgraded units are better fighters, but, they cost more. So I needed a metric which could say if the extra strength is worth the extra production cost.
Quick example, let's compare an Infantry to an Infantry Army. The army's strength went up by 17, meaning it's 4x as effective on the field as a single infantry (it deals twice as much damage and only takes half.) However, it costs 3 times what an infantry does to make an infantry army. (430 vs 1290, without the military academy.)
So it's "efficiency" score is:
Efficiency = Combat Effectiveness / (CostNew/CostOld)
Efficiency = 4 / (1290/430)
Efficiency = 1.33

What that means is that if Player A puts his production into infantry units, and Player B puts his production into infantry army units, Player B's forces will be 1.33 time as effective per production invested.
At the end of the day, armed forces are fielded in Civ6 by building them, so efficiently using production is what it's all about.

So does that mean I'd need in average 11 knights to kill 1 tank?

Normally what it means is that if we send 1 tank and 1 knight to go fight an unending stream of the same faceless unit, the tank would deal 11ish times the damage the knight would until it is killed.
This breaks down when the difference in strength exceeds 30. A combat strength difference of 30 or more is set to be a one hit kill- so you're actually going to need a ton of knights to kill a tank.
However, for smaller differences, your idea is correct. (We have to ignore healing and that wounded units have reduced strength bit.) You may not need 2.05 swordsmen to kill a single khevsur, but you'll need about 2.05X swordsmen to kill X khevsur.

 

[UWHabs]

That is a typo in the old post. It's Combat effectiveness divided by the production ratio. Great question though!

What "Efficiency" captures is the idea that upgraded units are better fighters, but, they cost more. So I needed a metric which could say if the extra strength is worth the extra production cost.
Quick example, let's compare an Infantry to an Infantry Army. The army's strength went up by 17, meaning it's 4x as effective on the field as a single infantry (it deals twice as much damage and only takes half.) However, it costs 3 times what an infantry does to make an infantry army. (430 vs 1290, without the military academy.)
So it's "efficiency" score is:
Efficiency = Combat Effectiveness / (CostNew/CostOld)
Efficiency = 4 / (1290/430)
Efficiency = 1.33

What that means is that if Player A puts his production into infantry units, and Player B puts his production into infantry army units, Player B's forces will be 1.33 time as effective per production invested.
At the end of the day, armed forces are fielded in Civ6 by building them, so efficiently using production is what it's all about.



Normally what it means is that if we send 1 tank and 1 knight to go fight an unending stream of the same faceless unit, the tank would deal 11ish times the damage the knight would until it is killed.
This breaks down when the difference in strength exceeds 30. A combat strength difference of 30 or more is set to be a one hit kill- so you're actually going to need a ton of knights to kill a tank.
However, for smaller differences, your idea is correct. (We have to ignore healing and that wounded units have reduced strength bit.) You may not need 2.05 swordsmen to kill a single khevsur, but you'll need about 2.05X swordsmen to kill X khevsur.

And I would imagine that this is meant only as a guide for comparison, not as a true final value, since we're not taking into account what policy cards may affect things, maintenance costs, hard build vs upgrade cost, etc...

I read it more as saying that it's worth it to rush to field cannons/artillery/tanks, because they're generally speaking a larger upgrade for a cheaper cost than their earlier counterparts. But obviously maybe not worth it if I have to hard-build them, or the eurekas don't work out, etc...