What is the difference between firepower and attack points?

[SlowThinker]

quote: Originally posted by finbar on 04-09-2001 08:27 PM I'm just wondering what causes the difference, ST. Is there room for fortuitousness in this kind of computer software?

Of course, there is.
The civ2 program counts a chance to win (7/16 for the attacking trireme, 9/16 for the defending one) and then the random generator chooses the winner of the "round" and so on.

[finbar]

Well, there you go. That explains everything. Thanks for that.

------------------
Founder, ACS Pedantry Institute
Founder, ACS Gourmet Recipe Exchange
Horse and Hydey Wrangler
Mono Rules!
#33984591

[Rufus T. Firefly]

If I remember the earlier discussion (the one the Marquis alludes to) correctly, it was suggested that the true strength in combat should be measured by multiplying together Attack Strength and Fire Power for the attacker, Defense Strength and Hit Points for the defender. Thus, the settler has a de facto defense of 2 (1DSx2HP) while the warrior has a de facto attack of 1 (1DSx1FP). Hence, in general, it should take at least two warriors attacking in succession to kill a settler, and even then they only have a 50-50 chance. I also supect that nukes are a special case, since by this formula a vet mech. infantry fortified on a mountain in a city with SAM should be able to withstand a nuclear strike (6DS x 3HP x 1.5vet x 3mountain x 1.5fortified x 2sam battery = 243) -- unless SAM's don't protect against nukes and terrain doesn't matter against air attacks; in that case total defense is only 40 (rounded down). Is that correct?

------------------
Dig trenches, with our men being killed off like flies? There isn't time to dig trenches. We'll have to buy them ready made. Here, run out and get some trenches.
-- Rufus T. Firefly, the original rush-builder
[This message has been edited by Rufus T. Firefly (edited April 10, 2001).]

[Scouse Gits]

Forgive me if I am misremembering, but ...
was there not some discussion that suggested some evidence that attackers gained a partial terrain advantage -- IIRC this point was not completely resolved, but there seemed to be quite a lot of circumstantial support for the theory

------------------
Scouse Git[1] -- git1@scousers.net

"Staring at your screen in horror and disbelief when you open a saved game is one of the fun things of a succession game " - Hueij
"The Great Library must be built!"
"A short cut has to be challenging,
were it not so it would be 'the way'." - Paul Craven

[Thue]

Here is how combat is calculated in freeciv. That should be civ 2 compatible: http://www.freeciv.org/lxr/source/se...s.c?v=cvs#L517

WOuldn't it be nicer to not have firepower? Couldn't about the same result be reached by increasing the attack rating of the unit? That would make everything simpler, yet not really limit you.
Hasn't ctp and SMAC done away with the firepower?

[East Street Trader]

Trireme to trireme is the combat I hate most. Mainly, I suppose because early on the trireme itself represents a hefty investment and all the more so if it has caravans on board. Not only are a whole load of shields at stake but also the time and effort of getting them to where they are and the prospective benefit their arrival will bring.

But also because I'm sure I lose more than my fair share (a view held by every one of us I confidently predict ).

Anyway, for a time I thought there was an edge to the attacker, then the defender, then I agreed with myself that it was in the lap of the Gods but the A1 probably had some edge or other.

For the time being I'll act on Slow Thinker's tip and say that the defender has a small edge (when I don't go for the Lighthouse, that is).

Can I just see if I've now got the overall picture? If I did the necessary arithmtic I could work out four things about a round. The chance for each of the two units to win that round; the damage which the defender will suffer if he loses; and the damage the attacker will suffer if he loses (or is the damage to the loser of the round the same for both?). Damage accumulates until the first one to run out of points hits the canvas. The victor is left standing with whatever damage he accumulated, round by round, before the knock out.

I'm no closer to understanding the "pouf" aspect to many ironclad's victories but maybe if I did the arithmetic (and compared the arithmetic to that produced in other combats) the numbers would tell a story.

By the way, gave up on a game last night. I was trying a new strat and it was going really well (well enough to keep me at the machine until the early hours happily micromanaging every city every turn). A neighbour had sent a phalanx and a chariot to sit in woods next to one of my plump heartland cities. They did not attack. I manoevred a catapult to the city, demanded their withdrawal and got a declaration of war. Smugly I attacked with the catapult, no doubt about the outcome in my head. With horror I saw my catapult creamed. But at least the bastard defending phalanx had been rendered vulnerable so with an angry, "I'll teach you" I attacked with one of my two defenders - only to see the undamaged chariot come out of the stack and successfully defend. Now in a fury I threw my second defender at the bastards but he hardly added to the chariots wounds. And there I sat contemplating the undefended city. Fury unabated I smashed down on the retire key and went to bed - there to contemplate the certainty of a cat victory had I just sneak attacked; the fact that it would have taken little energy to take a second cat along, just for insurance; just how easy it would have been to keep the neighbour sweet. And, fury abated, just how short a time it would have taken to accumulate the cash to bribe the city back.

[SlowThinker]

quote: Originally posted by Rufus T. Firefly it was suggested that the true strength in combat should be measured by multiplying together Attack Strength and Fire Power for the attacker, Defense Strength and Hit Points for the defender.

quote: Originally posted by Thue WOuldn't it be nicer to not have firepower? Couldn't about the same result be reached by increasing the attack rating of the unit?

IMHO Rufus T. Firefly, Thue and the FreeCiv algorithm are not right. See http://www.apolyton.net/forums/Forum1/HTML/001761.html (Modifiers for Attack/Defense), start reading by my post (posted January 27, 2001 06:12).

BTW, IMHO the product
"hit points" * "firepower"
affects the result of the combat
and the quotient
"hit points" / "firepower"
lowers accidental behaviour of units.

I think there are another mistakes in the freeciv program:

quote: 380 after a battle this routine is called to decide whether or not the unit 381 should become a veteran, if unit isn't already. 382 there is a 50/50% chance for it to happend, (100% if player got SUNTZU)

IMHO 50/50% is not true for defender, see East Street Trader's post (Modifiers for Attack/Defense, posted February 08, 2001 10:50).

Thue, how do you share the knowledge about Civ for the need of FreeCiv? You should have some "Great Library"...
Don't you hide it from us?

[Marquis de Sodaq]

SlowThinker is correct, the HP and FP figure in to reduce the chance of the absurd result. It doesn't eliminate it entirely; I'm sure we've all seen two barbarian knights die trying to kill an unfortified caravan on grassland. The attack:defence ratio determines who lands each punch, but clearly some units are stronger than others. If a warrior and armor both had the same HP/FP, the absurd warrior win could happen far more regularly than it can with this difference built in.

That said, EST rings a bell. In a trireme vs trireme combat, I always lose! Does everybody have the problem of even battles always seeming to go the AI's way? I generally don't even consider attacking unless the attack:defence ratio is at least 2:1.

Those damn incompetent shipbuilders... gotta teach 'em how to make caravels...

------------------
"There is no fortress impregnable to an ass laden with gold."
-Philip of Macedon

[Guest]

After all these discussions I simply can't resist the temptation anymore. This thread is filling up with all kinds of wild theories which we don't need because we are told the facts.

OK folks, ready to be reformed regarding how combats are resolved under this game? Then have a look in the official Civ2 manual!

---------------------------------------------------------------------------
Civ2 manual page 105:

CALCULATING THE WINNER

Combat in Civilization II is essentially like a rapid-fire boxing match. Units fight one-on-one in rounds, with damage equal to the firepower of the winner being subtracted from the hit points of the loser of each round. When one unit loses all its hit points, it is destroyed. If the loser is defending a stack of units and they are not inside a Fortress or a city, the whole stack is destroyed.

The important factors in combat are the attack and defense strengths of the
combatants as well as their hit points and firepower; the presence of veteran units on either side; the terrain occupied by the defender; and any defensive improvements in the square. In addition to considering all of these factors, combat also includes an element of chance. Imagine that sometimes, a unit just gets lucky. We don’t want to drag you through lots of heavy arithmetic for each combination of factors, but the calculations for each round of combat can be boiled down to a simple comparison.

The total modified attack and defense factors are combined and the probability of either side winning is approximately the ratio of each side’s factor compared to this total. For example, if an Elephant (attack factor 4) attacks a Phalanx (defense factor 2), the total of the factors is 6 (4 + 2). The Elephant has about a 66 percent chance of winning (4 out of 6) and the Phalanx about a 33 percent chance (2 out of 6).

Both the Elephant and the Phalanx have ten hit points and a firepower of one, so the battle goes between ten and nineteen rounds, until one or the other unit is reduced to zero hit points. It is possible for one opponent to win every round and take no damage at all, and it is possible for the opponents to trade damage for damage until even the eventual winner is badly beaten up. Most combats fall somewhere in the middle.

ADDING IN ADJUSTMENTS

How do those adjustments for veteran status and terrain and so on work? They’re added into each factor they affect before the total is determined. For instance, if both units are veterans, each gets a 50 percent bonus to attack and defense, giving the Elephant an attack factor of 6 (4 + 2) and the Phalanx a defense factor of 3 (2 + 1). Of course, modifying each unit’s factors also changes the total: Instead of 6, it is 9 (the total of each modified factor, 6 + 3). Now the odds are close to 6 out of 9 for the Elephant and about 3 out of 9 for the Phalanx.

If both are veterans and the Phalanx is behind City Walls (which triples a unit’s defense factor, making the veteran Phalanx a 9), the odds are about 6 out of 15 for the Elephant and close to 9 out of 15 for the Phalanx. Though the adjustments change the odds of each unit winning a single round, they have no affect on the total number of rounds or on the amount of damage inflicted.
[This message has been edited by Marko Polo (edited April 10, 2001).]

[SlowThinker]

Thue,
How can I use freeciv-dev@freeciv.org. Is freeciv a listserver? Shall I post "help" there?

quote: Originally posted by Marko Polo The total modified attack and defense factors are combined and the probability of either side winning is approximately the ratio of each side’s factor compared to this total. For example, if an Elephant (attack factor 4) attacks a Phalanx (defense factor 2), the total of the factors is 6 (4 + 2). The Elephant has about a 66 percent chance of winning (4 out of 6) and the Phalanx about a 33 percent chance (2 out of 6).

IMO the Civ 2 manual lies
In fact, for this case, Elephant's battle factor is
4 + (4 - 2 "bonus for the stronger unit") - 1/8 "malus for the attacker"
and Phalanx's battle factor is
2 + 1/8 "bonus for the defender"

The elephant's chance to win is (5+7/8) / 8 (=0.73)

I suppose it was tested and proved very well.

------------------------------
Do you see that the absence of the Great Library causes confusion in basics of Civ?
All knowledge stored at Apolyton must be sorted!

[This message has been edited by SlowThinker (edited April 10, 2001).]
[This message has been edited by SlowThinker (edited April 10, 2001).]

[Guest]

quote: Originally posted by SlowThinker on 04-10-2001 02:50 PM IMO the Civ 2 manual lie In fact, for this case, Elephant's battle factor is 4 + (4 - 2 "bonus for the stronger unit") - 1/8 malus for the attacker and Phalanx's battle factor is 2 + 1/8 bonus for the defender The elephant's chance to win is (5+7/8) / 8 (=0.73) I suppose it was tested and proved very well.

Sorry, I don't understand this at all. 'Bonus for the stronger unit'? Where is this theory coming from? Can someone confirm this?

[SlowThinker]

See the "Modifiers for Attack/Defense" forum, link is in my post here (posted April 10, 2001 09:48)

{combat formula}
{}{SlowThinker}{end2}

------------------------------
This is a post with keywords. See a thread The Great Library: a hierarchical structure" thread.

[This message has been edited by SlowThinker (edited April 15, 2001).]

[Marquis de Sodaq]

quote: Originally posted by Marko Polo on 04-10-2001 01:08 PM This thread is filling up with all kinds of wild theories which we don't need because we are told the facts. OK folks, ready to be reformed regarding how combats are resolved under this game? Then have a look in the official Civ2 manual!

HEAR YE! Yes, this thread and others are filling up with wild theories, speculation, and propogation of previously debunked hooey! See my notice of the impending Great Library posting of Combat Modifiers in the Strategy forum - I've been condensing threads into GL-usable entries - the topics of defense modifiers, hit points, and firepower are all going to be included. Condensing is the right word, too. Out of literally hundreds of posts, enough final info is available to fill a page or two.

To quote the often-proven-inaccurate manual, "We don’t want to drag you through lots of heavy arithmetic for each combination of factors, but the calculations for each round of combat can be boiled down to a simple comparison." In other words, the formula they give is a SIMPLIFIED VERSION, not the actual one used! Yet that is what people seem to be striving for in all these threads.

There, I've vented. This thread's topic has been resolved before, you'll see it soon in the GL...

[East Street Trader]

I compulsively attack barbs (in deity you really can't let them attack you) and the 1 -v- 1 battles probably work out fairly. I certainly win some close ones with warriors or phalanxes attacking one defence point barb archers.

For some reason I get a feeling that phalanxes do just a little better than warriors but the feeling is faint and I have no real faith that it truly is so.

I also find that I have to accept equal combat more often at sea, not just with triremes but ironclad against ironclad and (less often these days) destroyer -v- destroyer. I have no great sense that anything other than chance determines these battles.

I am not good at guessing whether my opponent is a vet or not. I have speculated that the more warlike civs may well build barracks or shipyards but have no evidence on that. By the time my spies are taking a peak into cities, ancient barrack will have been sold.

Since I made that post speculating that the heat of the battle might have an influence on defenders becoming vets I have watched out in games (in a desultory way) and still slightly incline to the notion. There are certainly exceptions, so some dice roll mechanism is at work, but I think there is an edge given to a stout defender.

Ironclads definately get made up more frequently than any other attacker. An easy victory may well be enough. I saw one become a vet after instantly destroying a poor caravan on open downland a day or two back (and no, I didn't have Sun Tsu's).

[Thue]

SlowThinker: The fact that combats gets less random with a higher "hit points" / "firepower" quotient needs not be explicitly build in. You can statistically prove that taking the average of many event will lie close to the mean value with a greater probability. So no error there.

The place to discuss freeciv code is on freeciv-dev@freeciv.org