The Transformation of Archers

[Dominae]

/me is confused for the moment, but is glad Nathan supports the "Swordsmen are better than Archers" stance.

[nbarclay]

Looking at the situation from yet another angle, swordsmen suffer a loss of 13.26 shields per thirty shields of attacking units and kill an average of 11.16 shields per thirty shields of attackers. Archers suffer a loss of 19.77 shields per thirty shields of attacking units (since it takes one and a half archers to make up thirty shields of attacking units) and inflict a loss of 10.23 shields per thirty shields of attackers. But as I noted before, the stack effect can make the overall kill/loss ratio far more favorable to the player then the odds with undamaged units fighting each other would indicate.

[Theseus]

Hmmm... you have thus far left out (I think ):

* The cost of supporting 50% more units (Archers) for appr. the same initial attack strength.

* On the other hand, the benefit of sheer numerical superiority on defense... IOW, the opponent will only have x units, and thus, on counter-attack, one will have either y Swords minus x damaged, versus 1.5y Archers minus x damaged (albeit more damaged or dead). That excess number of Archers will be untouched.

Overall, an interesting thread. My own shorthand is that yes, Archers and all of their begets are more interesting tools, and that further, in the case of iron deprivation, it's less painful to devise an Archer based military strategy given the upgrade path.

[Arrian]

I used to hate Archers, perferring mass chariot to horse upgrades, or warrior to sword, or a mixture.

Then I came to love archers, when I discovered how horrible the AI is at ultra-early warfare (sending drips and drabs of archers at you with no defensive backup) and was shocked to find how often I could generate early great leaders and nab the Pyramids in like 2000bc (PTW, obviously).

Now I simply see them as tools of rather limited value. Yeah, zero-range bombard is new, but honestly it mostly helps the AI, not me. If I want bombard, I will build catapults (which are now VERY effective). Much better use for 20 shields. I may build an early archer or two for barb hunting, but that's usually it, unless I'm forced into it by AI attack or lack of iron (though if I have horsies, I may go with the 'ole chariot upgrade).

-Arrian

[bongo]

Here's a nice little tool:
http://www.zachriel.com/BattleCivulation.asp

Using this I set up two battles. All regular units, non-militaristic civs. 1000 trials.

The first is 100 swords vs 63* spears fortified on flatland. The attacker won 52% of the times with an average of 63 survivors. The defender had an average of 2 survivors.

The second is 150 archers vs the same 63 spearmen. The attackers won 99% of the times with an average of 85 survivors. The defender had an average of 0.02 survivors.

If you attack with archers you will spend more shields on lost units (1300 vs 1110) but the increased number of units makes it easier to achieve numerical superiority. So using archers will increase your changes of winning the battle but it will also increase the price of victory.

[nbarclay]

Thanks, Bongo.

I tried playing around some with that, and was totally shocked at the size of the difference in outcomes if both attacker and defender are regulars (as appears to have been the case in Bongo's tests) compared with if both were veterans. With both sides using veterans, a sample outcome for swords is victory 75% of the time with 66 survivors, while archers won 100% of the time with an average of 88 survivors. I'm guessing the difference has largely to do with the fact that regulars get more promotions when they win.

The real key, though, is how the balance of forces on each side affects things. With veteran vs. veteran, 120 swordsmen can achieve the same essentially 100% victory rate that anpit 150 archers are needed for (and the extra losses involved in ensuring victory are negligible). Thus, if you can build a few extra units (or obtain them through warrior upgrades), swordsmen can provide an enromous long-term cost advantage without sacrificing near-certainty of victory. In contrast, archers are at their best when you can't build enough units to make victory almost certain and you need the weight of numbers they can provide.

With veteran attackers versus regular defenders (as is common against AIs), 90 swordsmen against 63 spearmen won 92% of the time with an average of 69 survivors (21 units lost) while 135 archers won against the same 63 spearmen 100% of the time with 94 survivors (41 units lost). 130 archers (2600 shields) was enough to ensure victory, while 100 swordsmen (3000 shields) were needed to ensure victory (testing only multiples of five). Thus, as when both sides have equal experience, a higher up-front cost in swordsmen leads to significantly fewer losses down the road.

Nathan