Fuzzy math and game development

A trip inside the sausage factory of numbers

Published on March 6, 2006 By Draginol In GalCiv Journals

World of Warcraft is one of my favorite games. I was in the beta from the start of it and when released, re-did my Paladin. And it was good. I didn't really pay much attention to all the numbers. I knew that doing X was better than doing Y. I didn't really know how much better it was, I just knew it was from the description.

My Paladin today doesn't resemble too much my Paladin of that initial launch (let alone the beta). A lot of that is from player feedback. Third party tools combined with players putting all the stats together resulted eliminating a lot of the "fuzzier" mechanics. Things like "Procs" got dissected and analyzed at levels I couldn't even imagine. That's not necessarily a bad thing mind you, it just means the game has such large appeal that it's attracted both the casual gamers and the ones who want to understand the ins and out.

The first Galactic Civilizations was all fuzzy math. Even I would have a hard time explaining with precision how morale worked or how production worked in terms of putting together a formula. In GalCiv I, your planet quality was central to everything. Various planetary improvements, morale, and bunches of other attributes got in there to do all kinds of multiplications to the various numbers. The order of some of these mattered since there'd be an addition here, a square root there. The system was designed essentially that building improvement X was better for production than improvement Y. The numbers, in essence, were all relative to one another. Someone looking for an entertainment network to make their morale go up by 15% would be sorely disappointed.

For the sequel, I wanted to dispense with as much of that as possible. A factory would build X production units. A research center would produce Y research units. Period. How much of its capacity was used depended on how fully funded the building was.

Entering the sausage factory

Things get murky when you start dealing with civilization ability points. One might argue that if I have a planet where my approval rating is 50% and I have a morale ability of 10% then my approval rating should be 55% (50*1.1). And if I build an entertainment network whose job is to improve morale by 20% that my approval rating would then be 50*1.3 = 65%.

What about production? If I have a factory that produces 10 units of production and it's all on social production and my social production ability is 20%, then my social production should be 12? Sure. But should the player be charged the 12 units? Or should it be 10 units with the other 2 production as bonus? That's the way it is on research.

Speaking of which, if I have 10 units of research being produced and my research ability is 50%, shouldn't my research by 15? No argument from me. On the surface, that's how it should be. And indeed, often that's how things start out. Then you have human beings playing your game and all those good designs go out the window. Probably the biggest reason for that in this particular game is from the mining resources.

There are research, military, economic, influence, and morale resources. To keep the game from being too complex to the casual user, starbase modules that mine these resoruces are the same no matter what type you are building on. That is, A mining barracks adds say 10% to your ability regardless of whether it's a morale resource or a weapons resource. The problem is, a 10% bonus to research is hugely different than a 10% bonus to morale which is a huge difference form a 10% bonus to weapons. I mean, heck, if I have a ship with 8 attack, I won't even get an additional point.

And we're just getting warmed up. Should a 10% bonus to your morale ability increase it by 10%? Or should it add 10 points to it? That is, if my morale ability is 10% and I mine a morale resoruce for 10%, should my morale ability be 11%? Or should it add 10% to it and make it 20%? We add it because otherwise, any semblance of balance could go out the window (create a civ with a 80% morale ability natively and then these percent multipliers would get crazy).

Has your head exploded yet? Because it just keeps getting better. In order to have some semblance of balance, we mess around with the ability values in order for them not to get out of whack.

That morale ability? fCivABilityFactor = pow(fCivABilityFactor,0.80f);

At release, Your Civ Ability at morale was just that. But it turned out on a large galaxy you could have several morale resources cranked up to over 100 points each. So suddenly you could have 100% taxes and 100% morale. Oops. So it was changed .95, then .9, then .7, then back upu to .8. That's the sausage factory that's game development. Where all your nice clean, elegant mechanics start to get murky.

Your research ability? It's chopped in half for the same reason.

Government waste

And what about "wasted" social production? In GalCiv I, military and social production was wasted even if you weren't building anything.

For GalCiv II, we decided to eliminate that. If you weren't building a ship or an improvement, you weren't charged for that production. That makes sense. Except, well, it turned out that players couldn't control their economy if social production was handled that way.

What happened is that say you're playing on a really large galaxy with 100 colonies and your economy is producing 5,000 net revenue per turn. Your planets have no improvements being built. Then an alien offers to trade you xeno factories. You take it. Then suddenly your net revenue goes to -2,000 per turn. Huh? What happened? All your planets started upgrading their factories and all that social production started to get charged for again. Yikes! Worse, it would gradually come back down as those improvements were completed all without the player doing anything.

Given that there's people who find the economic system in GalCiv II to be complex (what? separate tax and spend sliders?) having massively changing net revenue without user intervention would have put them over the top. So we ended up charging for social production. Which, is probably more realistic anyway and requires the player to put a little bit of effort into making their government more efficient. Still, it's not ideal because it has to be rationalized.

One of the ideas we had was to have social production go towards approval rating. But it's the same problem. Players see their approval alter by moving the spending slider and it's just another complexity.

Another idea was to have social production be added to military production on a given planet if there's nothing else to build. This is possibly more doable. And if there's no ship being built, it would still be spent. But at least that way, there's some benefit.

You have the power

Just like with World of Warcraft, games, especially statistics laden games like Galactic Civilizations are designed to evolve. We'll listen to what you have to say and together we'll keep improving the system. But never think there's a "best" system. There is only, at best, a system that annoys fewer people than the alternatives.

Comments (Page 7)

Alfonse

on Mar 08, 2006

How does it feel that your 60 minute-to-write speech got owned by 60 seconds of rebuttal?

Pretty good, considering that:

1: The 60 second "rebuttal" clearly addressed nothing in the first half of the piece.

2: The presence or absence of the auto updating feature was ultimately a minor point in the second half of the piece. The fact that the justification for social waste is rooted in a feature that can be turned off doesn't matter; indeed, the fact that the primary justification can be turned off is an argument in favor of my point.

Perhaps if you had read the piece rather than copied it, you would have understood those two things.

So, how does it feel to have your 30 seconds of fun pwned by an actual analysis of what's going on in the thread?

JamesACG

on Mar 08, 2006

Weighing two choices with uncertain results and picking the best takes luck. If you are informed of what effect the choices have you can make a strategic decision of which one will benefit you the most.

The point is determing what the best course of action is based on uncertain information, like the decisions people day in and day out. Sure, it's a little "luck" (whatever that means). But it's a matter of skill to determine the best ways to reduce your chances at bad outcomes and increase your chances of good outcomes.

I believe that global economy planning is to complex. It affects so many areas on so many planets that it is impossible to estimate wheather what you are about to do is good or not.

Nor should it be possible. I'm not saying I want the game to be an extremely realistic simulation. But I don't want a game dumbed down to 1+1=2 and that's all there is to it. I want it to at least be vaguely comparable to a real economy. There we determine the model that best fits what we (I say we, as in myself and other economists) know, rather than a full out accurate input -> output. There's just no game in that. At that point there is always a known optimum strategy. Civ II, well we know how this all works, so we'll just produce as many cities as we can fit on the map - there game end, I win.

I don't play games to repeat the same winning strategy time in and time out. I'd much rather look at each situation as it comes and determine the best course based on the information I have at the moment and the fundamentals I understand. I don't necessarily want to know exactly what the results will be next turn. I want to come into next turn, assess whether or not my choice was in fact optimum, and deal with things from that point.

Scumbag509

on Mar 09, 2006

One might argue that the social spending "wastage" is realistic.

I don't care if it's realistic. It's not fun. Even the Most Annoying TBS Thing of Late - Civ3's pollution - had the benefit of being understandable. GC2's economy is neither. It's unfun.

You'd think the Devs would understand that the bulk of the negative comments have come from seasoned strategy games. *We know best*.

Acleveralias

on Mar 09, 2006

Any changes to the system would have to fall under these guidelines:

1) No dramatic shifts in net revenue that aren't user controlled. That's why we have the social spending as-is. To prevent a scenaroi where smeone ends up going from +3000 per turn to -2000 per turn because they traded some tech with an AI.

2) No increase in micro management.

3) Does not involve totally redoing the economic system.

OK, here is a suggestion that fits those guidelines:
Get rid of the "spending" slider. Have three spending sliders instead of four. (absolute spending in military, social, and tech as percentages of maximum capacity)

Benefits from the newbie's perspective: It isn't obvious to many newbies that the three spending sliders are relative to each other. They may not realize that putting all three all the way on the right is the same as putting all three of them all the way on the left. If the sliders represent ABSOLUTE spending, it is easier to understand; if it all the way to the left I am not spending anything, all the way to the right I am maximizing spending on that sector of the economy. I don't inadvertantly spend less on my military by moving my research slider. Simple.

From the advanced player's perspective: MORE CONTROL. I can max out research and social spending at the same time. (expensive, but should be possible for brief periods). Percentages mean something. When my slider says "60%", it means that I am using 60 percent of my maximum capacity. I can change just the one facet of spending I am interested in changing without messing up my carefully calculated spending levels elsewhere. Total transparent spending control.

From the developer's perspective: Involves less "fuzzy math" than the current system. Eliminates a layer of obscure percentage times percentage muliplication. Eliminates the need for locking relative sliders. Cleaner interface. Less micromanagement (just move one slider instead of two or three to accomplish the same task) Does not cause massive swings in spending. Does not require a massive overhaul.

win - win.

P.S. Also please please don't let my social spending go to waste!

Luminae

on Mar 09, 2006

Just read through the thread, and wow, it's basically proved my entire theory about most strategy gamers being anal-retentive control freaks.

Personally, I don't crunch numbers, and I don't particularly want to crunch numbers. It's not really that big of a deal if I'm losing 3% production or a tiny bit of money somewhere in my gigantic empire goes to waste. I just play the game and have fun.

Playing a game just to have fun, imagine that. What a crazy idea!

zp00k

on Mar 09, 2006

Having your social spending not go to waste would make the game more fun, what a crazy idea!

Camp Freddie

on Mar 09, 2006

FROGBOY: For example, my pet peeve is that I can't see which flasks, shields, and hammers are from my ability and which are from my factories. I want them to be color-coded.

Please do this! Colour code stuff for:

Factory/lab production
Race bonuses
Starbase bonuses

As a second level of detail, could you shade the shields/beakers/hammers in the various categories to show if they need money to pay for them or if they are a free bonus (I know that *some* beakers are free with the race tech bonus, dunno about other production/bonus types). i.e. bright green beakers could show the bonus that are paid for, while less bright beakers are your freebies. Or maybe make the free beakers have a sparkley effect to show they are special. I'm sure your art guys could make something that looks good!

I don't need numbers and formulae everywhere, but I would like to see what my buildings are doing. I can't remember the exact production stats for each planet, so I've got no idea how much extra cash I got by building that trade center.
Tooltips on production like you have for planetary morale would be useful too, but colour coding makes it much easier to 'read'.

Regarding wasted social, the upgrade notification could be good - and so could the millitary divert idea.
I guess the notification could be tricky to code, and it would need an extra button on the governor UI to activate auto-upgrade at a later date if you choose 'no' at the initial prompt (there's no way in hell I want to manually upgrade 20+ planets!).
The problems I see with the military diversion of unspent social is that it would cause a budget fluctuation with non-shipbuilding colonies (which could be pretty big) and it would cause a big unanounced change in starship build times whenever you're given a new building tech.

Von Uber

on Mar 09, 2006

I agree, some idea of where you flasks etc are coming from would be very useful.

Martimus

on Mar 09, 2006

I love the idea of adding unused social production on military production. It would be a nice benefit to have.

100

Zartax

on Mar 09, 2006

Please notice I wrote "estimate", not predict. If the system is to complex to estimate the result you are just driving blindly. If you carefully read through my suggestion again and really think about it you will find that:

A) It's MORE realistic than the current model. How many countries set a global spendingbias of ex. research that ALL states must follow? Not many. More common is that they set a separate bias for every state depending on that state's possibility to use that budget. Having a separate set of spending sliders on each planet would resemble that.

If all planets would have their own set of sliders, we would exchange complexity for depth. Depth is something desirable while complexity is not. The number of combinations to adjust your planets now are limited since you have to control all at once. If you can control each planet separatly you will get a larger amount of combinations to set your planet (adding depth and divercity) while you loose the complex and overdynamic way the interstellar economy system had (brainmelting complexity). Please try to see the difference between comlexity and depth.

Ex: Chess is a simple, non-complex game that has lots of depth and absolutely no random elements.

C) Is the game about figuring out how to get what you want from your economic model OR how to best use your economic model to beat the opponent? Currently, the economic model is difficult to learn (to comlex) and impossible to master. If they change it to what I suggest it'll be easy to learn and difficult to master. It's not making it easier to win, just more easy to get a grip of and understand.

I don't play games to repeat the same winning strategy time in and time out. I'd much rather look at each situation as it comes and determine the best course based on the information I have at the moment and the fundamentals I understand. I don't necessarily want to know exactly what the results will be next turn. I want to come into next turn, assess whether or not my choice was in fact optimum, and deal with things from that point.

Se the example above. You know exactly what moving that queen will do, but there is no "winning strategy", is there? You know what your units will do, which is good, cause it'll enable planning. What you don't know is what your opponent will do, which adds diversity and challange.

101

JamesACG

on Mar 09, 2006

I suppose we should clarify, technically estimating and predicting are barely different - both words implying some ability for being wrong.

What I don't like, is a game like Civ IV where if I wanted I could know exactly everything going on (short of whether the spearman will beat the tank) without any uncertainty.

Now having a system that allows you to make an estimate on the results without knowing exactly what they'll really come out to, is accurate and enjoyable.

The specifics of how it does, doesn't, or should run the economy, I'm not terribly interested in getting into right now. I'm simply arguing against the idea that everything should a clear-cut formula that given an exact input it produces an exact output (in fact, I think there should always be some sort of random element within a formula in the game engine so at no point is there an exact input -> output result).

I think the suggestion you made is fine, although having independent sliders on every planet of you're vast empire would be a pain.

What I originally jumped into the argument about was the idea that knowing exactly what everything would do somehow makes for a great game, which I disagree.

And that ultimately comes down to this issue:

See the example above. You know exactly what moving that queen will do, but there is no "winning strategy", is there? You know what your units will do, which is good, cause it'll enable planning. What you don't know is what your opponent will do, which adds diversity and challange.

As I think I said previously, I never enjoyed the cut & dry aspect of chess. Of course, there are advantages to having no random element - it produces a purely skill based game. That's also why you won't see me jumping into a game with a world class player - there's no chance to win. And it's also why a computer can actually beat (on occasion) a world class player - there are ultimately a finite number of paths a game can play out. Sure, there's a whole heck of a lot of them.

Typically speaking in chess, a player, if he wishes to win, is always required to make the optimum play at any given point. There are no "Hail Mary" plays in Chess. The underdog rarely wins.

There was a time when I thought a game like chess was the only way a game should be made. Purely skill and no die roll (everyone hates playing a table top board game, executing a perfect strategy and having the dice turn against you). But ultimately I realized not only is history filled with examples of those situations happening, but that ultimately it makes the game more fun for everyone. It reduces the edge skill produces, but as long as the game doesn't become solely based on the random outcome (e.g. the card game of War), in the end skill still produces an edge.

102

Simplicity2

on Mar 09, 2006

Okay, I can present a simple solution which eliminates the entire problem, and can be summed up in one sentence.

MAKE SOCIAL PRODUCTION FREE.

Right now it costs 1 bc to convert to an industry point into social/military/research. I'm suggesting that you make it only cost money to convert into military or research.

Pros:

1) If social production is free, you aren't really wasting anything by having unspent social production. You're simply not putting more money into research or military, which is a perfectly valid option.

2) When a new building upgrade becomes available, every planet can start building it without any impact on the economy.

Cons:

1) Requires some economic rebalance. Perhaps make research cost 2 bcs? Or just lower benefits of trade/economic buildings?

Justification:

You're building social infrastructure. You make back any money you put in.

103

amanasleep

on Mar 09, 2006

Honestly, the wated social production doesn't bother me. The problems are:

1. The "focus" buttons are too inexact a tool. Should be sliders.
2. The econ sliders themselves are counterintuitive. Research should have it's own independant slider. Social and military should have their own sliders, but they can still draw from the same production capacity.
3. Each planet should have its own production and research sliders. That way your best planet is not penalized by the economic priorities of the rest of the empire (implementation of this might obviate the need for the focus buttons/sliders).

If you have money and capacity, you should be able to use it. In the current system, you can conceivably be at 100% spending, with a large positive income, and still have a large amount of unused capacity. The only way to make use of it in the current system is to force buy it for many times it's actual value.

Ultimately, the system has been balanced for the current econ setup, and changing it as outlined above would doubtless make it many times more efficient, forcing an intensive rebalancing of the game (and recoding of the AI). I understand if it can't be done.

The transparency issue, on the other hand, is entirely withing SD's control and should be addressed. This problem extends to every aspect of gameplay, and Stardock I feel is obligated to publish this information and adequately explain it to the players. Whatever your system is, however it works, it results in a playable game, so without making a single change to the system you could instantaneously make the game easier to play and more intuitive simply by publishing a complete listing of all relevant game mechanics, including.

1. Exactly how the economy is calculated, and how bonuses are applied.
2. How all other bonuses are applied.
3. A glossary defining all terms used in the game, and what characteristics, technologies, bonuses, or abilities they refer to. This is critical! There are many game terms that are inconsistant or totally mysterious. This is completely unacceptable. Why should a player encounter a reference to a bonus in the game and then have nowhere to go to find out what the bonus refers to? Unbelievable. There is no excuse for inadequate documentation of major gameplay-affecting terms.
4. Exactly how combat works, complete with formulae, how all bonuses combine, and relevant gameplay examples that give real results. The manual's examples are especially laughable, with typically coy SD answers like "suffice it to say, this will be close". That example should not have been close anyway, according to the manual itself! Of course, since it is clear that combat does not actually work as the manual explains...

Stardock has created a complex game the underlying mechanics of which are balanced enough as is to work and be enjoyable (with some exceptions). In any event, I am happy to play the game as designed and with whatever tweaks the designers see fit to include. But the lack of documentation on how to play the game, and the incredible amount of inaccuracies in the documentation provided are the principal reasons I am leaning towards not playing it any more until the devs get it together to provide it. Correct, adequate documentation is the bare minimum that a game company is expected to provide, and Stardock fails to meet that requirement.

This documentation should be added in game in a single "GalCivilopedia" or provided in a single document available for download. I see a few things like weapons and armor tables becoming available and this is good, but everything needs to be in one place, and still does not explain underlying game mechanics and modifiers, whach are even more critical.

104

warui1978

on Mar 09, 2006

Actually I like the system as it is. The way the system works now forces me to use the FOCUS button / micro my money / spend money on upgrading buildings in low production planets. Right now I have to create a balance so that all my planets finish upgrading at the same time.

But if I have to make a suggestion, I would suggest that unused social production be used to augment influence? I mean it sorta makes sense right? instead of building infrastructure, build culture that's my 2 cents worth but I really like the way it is right now..

105

Luhh

on Mar 09, 2006

This is not a solution. There are still waste. This only gets rid of the penalties, but you lose out on production as long
as :

1) military / social / research are on the same sliders.
2) tiles on planets aren't used because factories are idle.

Making social production free could be a good idea - I proposed it myself earlier in this thread - but with a different approach. Too bad these forums are as bugged as the game so I can't edit my previous post. Maybe I'll rewrite it later to make it more understandable as some things may seem unclear.

Welcome Guest! Please take the time to register with us.
There are many great features available to you once you register, including:

Richer content, access to many features that are disabled for guests like commenting on the forums.
Access to a great community, with a massive database of many, many areas of interest.
Access to contests & subscription offers like exclusive emails.
It's simple, and FREE!