[no players] We like to lurk lurkers lurking.

[Commodore]

Indeed...

Nope. Promote HA to F2 and they just bounce off regardless of the odds, ie what units are defending. Those that go splat don't matter, but get enough to bounce and the siege units disappear in a glorious cloud of smoke. The amount of units you need is proportional to the amount of siege in the stack, the total number of actual defenders is (almost) irrelevant. I think Jowy has, what, 6 cats? Don't need that many HA to flank those off. Need like 4 good retreats and they all die.

This. Horse archers are just lovely for flanking the ever-loving crap out of waddling siege stacks.

But...

All that said, taking a bet on what 2 frankly insane civ players do in a war is not all that smart. Might as well roll a dice and bet on if it rolls pi or e.

This. It's the flaccid force against the super-penetrable object!

[Bobchillingworth]

Goddamn dropbox replaced all the pictures I had in this post with ones from my SG. Suffice to say, it was a crushing representation of how siege flanking totes sucks.

[NobleHelium]

Krill always mentions a bunch of scenarios that don't generally apply to the game at hand, you should know that by now...

[Bobchillingworth]

Missed this:

Catwalk, if you read this after the game, sometimes you just have to step up and punch the bully in the nuts. If you pay him once, he's only going to come back for more.

Totally. Paying "protection money" rarely works out well in real life, let alone Civ, unless you enjoy extortionists gradually squeezing you dry. Trying to bribe Jowy, who plays Civ as a combat simulator and has never met a fight he didn't like, well...


The simplified economic mechanics of Civ make this scenario a classic vicious cycle. All that gold gifted away slows down Catwalk's ability to both defend himself & tech, while it concurrently props up Jowy's military, which means Jowy is able to come back around in another 10 turns to extort more gold, which means...

[Krill]

The amount of flanking damage done to catapults is dependent on the amount of damage the HA does against the defender. IIRC it is the amount of damage done in 1 round against the defender, but I'm not 100% certain on that; it may have changed since I originally found out. Which means that F2 becomes less useful the stronger the defender becomes, and in this instance, it's almost certainly better to just promote to C1/C2 to get similar odds for flanking damage.

[Krill]

I guess in the same way that a hypothetical, infinite army of any unit will eventually overwhelm anything.

Krill always mentions a bunch of scenarios that don't generally apply to the game at hand, you should know that by now...

[WilliamLP]

The odd thing about Bob's experiment to me was that the flanking damage depends on the defensive strength of the top defender, not on the siege unit itself! (I've read that that's how it worked - that it's just like a combat hit to the cat, but this is wrong.)

I tested it in WB myself, a HA does 26 HP of damage in flanking to a cat when attacking a cat as top defender, 17 HP of damage to a cat when attacking a spear as top defender.

Getting in the source code, demystifying flanking for myself. I'm reading the code quickly and it's convoluted with lots of dependent functions, but here's my understanding of how flanking works. (Some of what I see contrary to what I've read.)

- You go through all flankable siege in the stack.
- For whether or not a unit is elegible for flanking depends on a die roll per unit: "iAttackerKillOdds = iDefenderOdds * (100 - withdrawalProbability()) / 100;", and we test whether a roll is _greater_ than that. So a higher withdrawal chance means a higher chance to hit a particular unit with flanking, as does higher odds against the top defender. It's weird and non-intuitive but it's what the code says unless I'm missing something. What it does do though is give the Flanking promotion line non-obvious benefits in flanking, but only if there are smaller numbers of cats to flank in a stack.
- Then you pick from eligible units to flank, up to the max collateral hits for the attack. (For a HA, this is 6.)

(Also, to be honest, the code looks obviously bugged to me: it goes to a lot of trouble to get defensive values for the flanked unit, but never uses them.)

This solves a lot of mystery for me that I've seen in World Builder. In particular, how inconsistent the number of flanking hits. Yet, if the number of siege units is large enough, you will get 6 hits every time. If you are attacking a top defender at even odds, and there is a cat in the stack, you will flank it 60% of the time according to the above formula. If you have 6 cats which could be flanked, the expected value of flank hits will range from 3 to 4 or so.

This code also means a Cav will do the same flanking damage to a cat, cannon, or treb, and it doesn't depend on the siege unit's promotions at all.

The case of HAs vs 6 cats with an unlimited number of spears, with no promotions:
- Attacker strength value is 6, defender is 8.
- The damage done per hit to the spear is 20 * (6*3 + 8) / (6 + 8*3) = 17.33, rounded to 17. This is also the damage done by flanking to a cat. So six hits will kill.
- The probability that a particular cat is flanked is about 54%. (Use the above weird formula.)
- So the expected number of successful combats (win or withdraw) to kill a particular cat (if there are 6 or fewer) is 6 / (0.54) = about 11. (Note this doesn't mean that 11 hits will kill 50% of the time, but it should be near that.)

My conclusions:

- You're going to need a ton of HAs to actually kill any catapults with flanking, regardless of numbers in the stack. At 50% to win or withdraw vs a top defender it needs to be over 20, at least. More if there are more than 6 cats in the stack.
- I think the power of flanking is overstated, as Bob said. I understand that in previous patches it was much stronger, which is where the ideas of some of the vets probably come from.
- As Bob said, flanking is not a decisive factor in this game, until you reach a number of attackers that is unrealistic to have before Knights.
- And, as such, I don't understand why it needed to be nerfed in RBMod at all.

Check the code and test yourself if you don't believe me, but I've seen enough to convince myself.

[Commodore]

Secondary conclusion...Numidians can be reverted?

[Krill]

Really going to have this conversation tonight? OK then...

Flanking was implemented originally to remove the 95% catapult stacks that occurred in MP games, where players would build minimal cover units and then just mass catapults. Catapults don't receive collateral damage, so at the time the thoughts from teh majority were that there was no cost efficient way to counter that tactic. So the Devs in BtS introduced Flanking which is basically a method of applying collateral damage to a specific type of unit, and added to the HA because at the time the HA was thought to suck. As it turned out both of these thoughts were basically wrong, one specific counter to the massed cat stack was to mass swords, strength 6 units that cost the same and moved just as fast became an immoveable object and slaughtered cats that wandered into your culture. HA we a close second but better the longer the game went on due to the extra movement, but they cost 50 hammers compared to the 40 hammers of the cat (cost was increased in BtS).

History lesson about the evolution of tactics and strategy aside, I don't think those numbers are representative of either normal tech paths, stack compositions, or a sane overarching game strategy. If I see someone building up a huge stack of spears why would I decide to build HA? Why wouldn't I instead decide to build axes, or even mass swords? Swords eat every unit alive in that stack. Another method of dealing with it is the standard suicide catapult: after all, I'm not sure unlimited spear stacks exist. That lowers the number of HA needed to kill the cats. I'd also point out the more cats there are the more likely it is to actually hit one with Flanking damage so I don't think the amount of HA needed increases proportionally. And then I'm still trying to figure out what 6 cats and a spear stack is going to do to me in the first place...?

This entire debate started from when Bob stated that HA don't obsolete cats. The counter point was specifically that enough cats do, and honestly, 20 HA by T130 is a minimum that you will see. Jowy has 27 here for instance. In fact I even agree with you that with small stacks Flanking doesn't do much at all, I don't think I've ever disagreed with that. And as to why it got nerfed in RB mod, I'll just point at PBEM13. Those numbers where you talk about the minimum are easy to reach and make it too prohibitive to attack into through numbers.

[wetbandit]

FWIW, and it's poorly reported in my thread there, but in PB11, I was able to flank away about 12-15 cannon with 16 Cav in the late war I had with Molach. Molach's thread states that he lost 10 cannons, I think it was a little more, but it's in the ballpark. I believe I have some other pictures at home which may provide more specific information. To my memory, most fights the Cavs had were against Rifles, and the majority of cannons, if not all, were killed by flanking damage.