Those are some nice plots. You might want to change the scaling on the cumulative research axis so it's easier to see what's happening earlier on in the game. Perhaps switch to square-root scaling.

Also, I wonder if there isn't something wrong with your calculations. Several of the 'special' techs seem to be below their base curves. While I might expect that on cost, on rating/unit I really don't see how that can be true.

Entropy Avatar wrote:

You might want to change the scaling on the cumulative research axis so it's easier to see what's happening earlier on in the game.

Yup, in the spreadsheet there are log-log plots of each of the graphs to make it easier to see what's going on in the beginning.

Safari Joe wrote:

Missile, beam and mass drivers scale differently for hull sizes which means the damage to size ratio drops drastically as you go up in hull sizes especially for beam and mass drivers. This could mess up your results somewhat.

Yikes. This would change things, to be sure. Can you elaborate on this? What's your source? Is this for weapons only or defenses too? It's the first I've heard of anything like this.
I'll be doing some testing soon on component size as a function hull size and miniturization level, and now I'll take a look at possible variations of this for beam/missle/mass driver, too. Any more information you have would help.

So if I understand #4 correctly, each component has a value, probably SizeMod, that alters how much space the component requires given hull size and miniaturization? That is, this varies on a per-component, not per-branch (beam, missile, mass driver), basis?

Is there a file somewhere in my GalCiv2 directory that contains these SizeMod values?

My next step is collect some in-game data and then reverse-engineer the underlying functions from that. Something like this SizeMod business could make that much more complicated... more information, please!

Hmm, yes, if you could figure out the component-size / hull-size variance for us I'd be quite happy. Less of my own effort to sate my curiousity.

Empyrean wruote:

The amount of space required isn't changed by miniaturization. Miniaturization just gives you more space on each hull, so that won't affect your calculations.

Starting a new game (v1.1 BETA-2A) in cheat mode and advancing instantly through the miniaturization techs gives me the following sizes for a hyperdrive on a cargo hull:
Hyperdrive Size, Cargo Hull Size, Miniaturization bonus, Miniaturization tech
11, 55, 0%, None
12, 60, 10%, Basic Miniaturization
12, 68, 25%, Enhanced Miniaturization
13, 77, 40%, Advanced Miniaturization
14, 85, 55%, Expert Miniaturization
15, 96, 75%, Ultimate Miniaturization
17, 110, 100%, Supreme Miniaturization

xFlukex wrote:

Yeah, if you use larger hulls then weapons/armor are a bit larger. It's weird you are suprised by this, because it is kinda obvious when you play GC2. Your source should be just playing the game and writing down the numbers you get, otherwise

Re-reading what was written earlier, I may have misunderstood what Safari Joe was saying. I know that hull size affects component size. I thought that Safari Joe was saying that beam weapon size varies with respect to hull size at a different rate than missile weapons vary with respect to hull size. If all Safari Joe and Alfonse were saying is that component size varies with hull size, then there's no problem. The graphs are still going to be relevant because the data will stay in the same relation to each other. (There's a slight complication to this caused by the discretization of the values; the values will change slightly when fractional values are rounded down to the nearest integer.)

It's too cluttered in the beginning to see what's happening. Oh yeah, and you forget the nano-ripper.

In the Excel file there are 8 plots; the four I mentioned in the orignal post, and four that are copies of these but plotted logarithmically. Maybe I should have mentioned this... I only posted those two because they are probably the two most important, and saving the graph as a jpeg, uploading it to an image host, and linking it in the post is a bit tedious to do for large numbers of plots. Just check out the .xls file.

The nano-ripper is there. It's not an ethically dependent tech (as far as I know? is the documentation wrong?), just a high damage/size, high cost/size weapon available with Singularity Driver IV. It's the tall, vertical (because two components are available with that tech), red spike fairly early on in the Damage / Size plot.

What exactly do you need to reverse-engineer? The way combat and ship building works is pretty darn simple.

Are attack rolls 0-based (like the devs say, but game never shows), 1-based, or 0-based but with a minimum of 1? How exactly is the "most dangerous ship' in a fleet chosen when being fired upon? If a 10-0-0 offense fires upon a 0-4-4 defense, is the defense int(sqrt(4) + sqrt(4)) = 4, int(sqrt(4)) + int(sqrt(4)) = 4, or int(sqrt(4+4)) = 3? If a component has size S on a small hull, what is the size of that component on a large hull with Expert Miniaturization and Hyperion Shrinkers?

And then, once these are answered, you can start to answer questions like "What is the probabilty that Fleet A will defeat Fleet B without losing a ship?" and so on.

The formula for component size is as follows:

ComponentEffectiveSize = INT(BaseComponentSize + (SizeMod / 100) * EffectiveHullSize)

Where

EffectiveHullSize = INT( BaseHullSize * (1 + TotalMiniaturizationBonus)); for example, 60 for a cargo hull (55) with 10% Miniaturization bonus.

BaseComponentSize is the no-hull size of a component listed in, for example, the Galactopedia;

SizeMod is an integer specific to each component, usually between 1 and about 10, in general starting high (5-8) in the technologically early components of a branch and decreasing down that branch to lower value (2-5), and listed only, as far as I know, in the GalCiv2\MetaverseData\English\GC2Types.xml. Thankfully Excel can import xml data...

I'll post a thread about this when I get a chance.

Yeah, I figured this out. The final formula in #13 takes the different rates of scaling into account.

xFlukex wrote:

Yeah you can roll a 0 for attack but it never displays it, I didn't believe it until I saw it myself.

Yeah, I don't think this actually happens. I just booted up a test game, took a A(1-0-0) D(0-0-0) ship against a A(3-0-0) D(0-0-0) ship, and each round I did exactly 1 damage, and each round the enemy ship did 1-3 damage. We alterenated firing with no interruptions. If it was 0 based attack and just didn't display it, it would appear that ships occaisionally attack twice in a row. This never happened. So it looks like it's 1-based or 0-based with a minimum of 1.

attack_rating/(defense_rating + hp)

I've seen this formula around, but never any source or validation...