Age of Empires 2 is the weirdest game of rock, paper, scissors you’ll ever play

Age of Empires 2 has proven to be an extremely resilient game; despite being originally released before the current millennium (in late 1999) and having been essentially abandoned by Microsoft for nearly a decade, it remained popular, and, after two remakes (in 2013 and 2020), it is perhaps more popular than it has ever been.

There are a host of reasons for this: the game manages to work well as a casual game and as a competitive one: its medieval setting helps make the game feel intuitive and compelling to the kinds of nerds that play RTSes, it has an absolutely ridiculous number of different modes and settings, and — what I want to talk about — its gameplay is mind-boggling deep.

Like pretty much every strategy game that is about winning fights, Age of Empires 2 (“AoE2” from here on) is based on a complicated rock-paper-scissors-esque web of interactions. Every unit is good against some units and bad against others. But AoE2 has remained fresh and compelling in no small part because its complicated web of rock, paper, scissors, interactions comes with many asterisks and caveats.

Rock Beats Scissors? It Depends

If you ask a high level AoE2 player about whether some particular unit is a good choice, the most common answer you get will be “It depends“.

This is not because high level players are fiercely guarding trade secrets that give them an advantage over other players. How a matchup between different units in AoE2 can be extremely finicky depending on a host of factors — numbers, upgrades, terrain, how well you control them –, and you often won’t know whether you’ll win an engagement until you’ve seen dozens or even hundreds of similar ones.

What makes fights so difficult to predict?

There are a lot of small factors determining how an engagement goes: small changes in upgrades, terrain, how you control units, and the size of the armies can radically change how well a fight goes for you. The soft counter system of AoE2, while easy to learn, is very hard to master for two reasons:

1. Unit matchups in AoE2 can be susceptible to what I’ll call “counter-flipping”: a unit that counters another unit in one situation might be countered by that unit in a different situation despite nothing about the two units themselves changing between the situations.
2. Unit matchups in AoE2 are “finicky”: seemingly small changes in inputs can lead to massive changes in outcomes.

How the scale of the fight affects who wins is one of my favorite aspects of AoE2 — a fight between 5 knights and 10 crossbows might turn out completely differently from a fight between 20 knights and 40 crossbows. Simply doubling the size of each army might change which side wins.

How can that be?

Units in AoE2 aren’t all engaging the same way. Some are melee, others are ranged, some have longer range than others, and others can damage multiple units at once.

A common AoE2 fight involves melee units on one side (like knights) that need to be adjacent to units to attack them and ranged units (like crossbowmen) that can attack units from within a set distance on the other. As the two armies approach each other, the ranged units get to shoot first as the melee units close the gap.

One side getting to strike first leads to some interesting possibilities.

Let’s consider a fight between some knights and some crossbowmen. With some of the basic upgrades, a knight has 120 hitpoints and does 10 damage to a crossbowman, while a crossbowman has 35 hitpoints and does 5 damage to a knight. (I’ve picked these numbers because they are convenient to illustrate my point.) The knight seems to have a huge advantage, but the crossbowman has two things going for it — it’s much cheaper than the knight and it can hit the knight once or twice before the knight can hit it. This will mean that how well knights and crossbowmen do against each other depends enormously on how many units are involved in a fight.

Counter-Flipping: Knights vs Crossbowmen in theory

Let’s look at a theoretical, pen-and-paper version of a Knights vs Crossbowmen fight. Knights will charge in and each attack a different crossbowman, one-by-one. Crossbowmen will all attack focus their fire on one knight until it dies, then move on to the next one, without kiting backwards away from the knights.

If we send one knight (120 hp, 10 damage) at two crossbowmen (70 total HP, 10 total damage), the fight goes about as you might expect, with the much beefier knight winning.

The single knight prevails with a third of its health left. Its massively higher health combines with the fact that after killing the first crossbowman it takes less damage each second to give it the victory.

However, if we scale this fight up to eight knights fighting sixteen crossbowman, the battle is a virtual tie, ending with one knight alive with a mere 5 hitpoints. Again, the knights are outnumbered 2 to 1, but now that numbers disadvantage balances out their superior individual stats.

Make the battle larger and the crossbowman win. Twenty-four crossbowman should beat twelve knights with about 18% of their hitpoints remaining; the relationship has flipped. We didn’t have to change anything about the units or how they fought; simply making the battle bigger radically changed the outcome.

Do knights beat crossbowmen? It depends.

Trying out Knights vs Crossbows in-game

Trying out these fights in-game by replicating the assumptions of the theoretical analysis as much as possible (the crossbows target knights one-by-one and don’t hit and run, the knights charge in directly and spread out among the crossbowmen), we get results that are pretty similar to the theoretical pen-and-paper approach:

Crossbowmen	Knights	Final Xbow HP%	Final Knight HP%
2	1	0%	37.5%
16	8	0%	3.8%
24	12	19.2%	0%

The results differ a bit the simplified theoretical analysis — knights in-game attack slightly faster than crossbows (without the Thumb Ring upgrade that increases their fire rate) but can bump into each other and lose time before reaching their next target, and in-game crossbow projectiles can miss. But the general principle is clear: all else equal, scale up a fight and ranged units do better; scale down a fight and melee units do better.

Why Ranged Units Do Better Against Melee Units in Bigger Battles

Ranged units have two major advantages over melee units:

1. They get to attack first (obviously)
2. They can focus their damage on a single unit to kill it as fast as possible.

Combine these two factors, and large groups of ranged units get to thin out melee units before they close the gap whereas small groups of ranged units don’t. If we look at how much damage each group of units can do over the span of the fight, we can see this principle in action.

In a smaller fight, the crossbows aren’t able to thin the knights down fast enough to make up for the discrepancy in their overall stats.

In a larger fight, the crossbows can thin out the knights quickly. Even though the fights might look similar (some knights charging against twice as many crossbows), the scaling power of ranged units work in their favor, as they are able to kill knights before they can engage. The crossbowman never feel the full damage potential of the knight army because they’ve already cut down the potential damage that the knight army could do before it even reaches them (at time=3).

Despite their far lower total health, the crossbowmen are able to focus down their opponents fast enough to neutralize the initial discrepancy, as long as the fight is big enough. A group of 24 crossbowmen can kill a knight in one volley and negate the huge statistical advantage that the knights have (over 70% more initial HP!).

Matchups between ranged and melee units flipping as the size of the fight changes isn’t unique to AoE2. But what makes AoE2 particularly compelling is that when (and even whether) a matchup flips can vary wildly based on a host of small, sometimes subtle factors.

Unit matchups in AoE2 are finicky. That’s good.

How a fight will go depends on a host of factors — the units involved, their upgrades, how they’re controlled, the terrain, and even the size of the fight (as we just saw). Small changes in input can produce huge differences in outcome, so what will work against what your opponent is doing might not always be the same thing. Pair this with the fact that there are thousands of possible combinations of units, upgrades, and bonuses, and the skill ceiling just for understanding how each matchup should play out becomes very high.

Upgrades in AoE2 work by changing how much damage a unit does or takes by a flat amount. This means that fights can go extremely differently depending on which upgrades each combatant has. In my discussion so far, I’ve been assuming knights take 5 damage from a crossbow shot, but upgrades could let them take a mere 3. Dropping the crossbowmen’s damage output by 40% makes a huge difference in a fight; 24 crossbowmen doing 3 damage each to 12 knights will get absolutely wrecked, leaving the knights with over half of their HP after an open fight. Knowing what upgrades to prioritize and when you do and don’t need to wait for them before taking a fight is a huge part of mastering the game. Countless counter-flipping scenarios come as consequences of one or two upgrades completely changing how one unit does against another.

“Micro“, how units are controlled, can completely flip fights. Ranged units that hit and run can do much better against units that need to close in to attack than ranged units that just stand in place. Units can dodge projectiles, focus on thinning out weak units quickly, or blunder into an opponent’s army without attacking, all with radically different consequences. At very high levels of play, a unit that typically counters another might not just because expert players are too good at dodging its attacks.

Terrain can also flip fights. Units on elevated ground take less damage from and do more damage to units below them, meaning an army on top of a hill can win fights it otherwise wouldn’t. If ranged units can squeeze into a narrow gap, melee units that try to attack them will not all be able to engage at once. A group of 20 crossbows that can squeeze themselves into a narrow chokepoint before a set of 20 knights closes in might find themselves fighting a stream of 2 knights at a time, because most of the knight army is stuck outside the chokepoint.

All this means is that unit matchups in AoE2 can be very finicky; small changes in input can result in massive changes in output. Fights that might look similar when they begin can look entirely different after they end due to a difference of one upgrade or one player maneuvering into an advantageous chokepoint. This creates a very high skill ceiling for choosing when and when not to take a fight. An expert player is going to be better than a merely very good player at quickly assessing a situation (the numbers, the terrain, the upgrades on both sides, how well they control their army compared to their opponent) and getting a good estimate of whether charging forward or backing up is the better choice.

Picking good fights is just one of many challenging aspects of a well-designed strategy game, and AoE2’s soft counter system and enormous number of consequential differences among units make it a great example of how making fights more difficult to predict and giving players tools to flip a disadvantageous situation in their favor can keep a game fresh for decades.

For more on games

The Objectively Best Pokemon Type, According to Game Theory