Preface

This is an attempt to give SESocial Engineering modders a tool to roughly compare different society effects for the purpose of better SE balancing. Thanks to all contributors on many other forums for expressing their opinions and sharing ideas. Such tool does not replace actual playing experience and testing because, fortunately, effect have very different effects and applicability to different strategies and play styles. However, it may help fixing too much imbalance in SE system.

In addition this article also analyses vanilla SE models using above comparison technique and suggests correction to balance models better, adjust effects range usage, adjust underused models appeal, adjust extra sharp effects, etc.

Effect weights idea

Effect weight is a relative effect value comparing to other effects. Definitely, their behavior is nonlinear both because of nonlinear effect scale and due to different game state, environment, and play style. Therefore, such comparison is nonlinear too and depends on many factors. All evaluations are bound to be very approximate and are applicable one on "in average" case. I also used multiple assumptions in below calculations. Feel free to correct me of propose your own. I will appreciate any input. Thank you.

Comparison method

Each effect is expressed in either minerals or energy difference to the player (average per base). Conversion ratio is 1 mineral = 2 energy. Then these differences are compared to each other to calculate relative SE weights.

Historical periods

I consider three historical periods for purpose of averaging. The very early part of the game is excluded since SE are not available there - nothing to compare. The very late part of the game is excluded as at that point game is either won or lost and no social engineering can help it.

Historical period average values
period	early	middle	late
approximate turns	25-50	50-100	100-200
base count	5	10	20
mineral intake	6	12	24
energy intake	4	12	36
base size	3	5	9
nutrients/square	2.0	2.5	3.0
commerce technologies	1	3	5
distance from HQ	3	6	12
economy allocation	0.3	0.3	0.3
labs allocation	0.7	0.6	0.5
psych allocation	0.0	0.1	0.2
minerals multiplier	1.0	1.5	2.0
energy multiplier	1.0	1.5	2.0

Effect result calculations

All calculation are done for normal map and all other medium values if not specified otherwise. This can be extrapolated to other map sizes.

ENERGY

This is an artificial effect introduced for the ease of further calculations and to establish base weight value. One step on this imaginary ENERGY effect scale changes raw city energy yield (including commerce) by 10%. So it works similar to other linear effects. Of course, it doesn't exist in a game as described but it is convenient to use it as a base value since many other effects can be expressed through it. I also pick this one as a base and not an Industry as energy can be converted to both Industry and Research - two important effects not easily comparable with each other.

ENERGY result = 0.1 * <effective energy intake>

ENERGY derivatives

From ENERGY weight we also can calculate its derivative weights: LAB/ECO/PSY. LAB is a RESEARCH. Other two do not exist in the game but it is still useful to calculate them to use as intermediary values for further computation. Obviously, all of them change proportionally when Energy change. We will compare each of them individually with Energy assuming we also can change allocation without penalty to channel all extra energy to one derivative only without modifying others.

Let's calculate how much ENERGY should increase to produce same result as LAB increase.

lab increase for +1 ENERGY = 0.1 * <effective energy intake> * <lab multiplier>
lab increase for +1 LAB    = 0.1 * <effective energy intake> * <lab allocation> * <lab multiplier>

Using the above we can express LAB result through ENERGY result.

LAB result = ENERGY result * <lab allocation>

Same way we get similar formulas for ECO and PSY:

ECO result = ENERGY result * <eco allocation>
PSY result = ENERGY result * <psy allocation>

A special thing to say about RESEARCH. Players tend to maximize labs as much as allocation penalty possibly allows. That is an indication that research advantage is actually much more valuable then mere economy to production contribution. Unfortunately, technology advantage is impossible to evaluate by itself and even less the discovery rate. That's why I tend to give RESEARCH additions factor of 2 on top of the above LAB result formula. Same bonus goes to ENERGY as this is also source of research. The more economy and labs combined player has the more they can divert to research - either directly or indirectly through economy->labs reallocation. In this regard it should be obvious that increasing energy intake by 10% is about twice as more valuable as just increasing labs by 10% because energy goes to both labs and economy.

INDUSTRY

INDUSTRY result = 0.1 * <mineral intake>

ECONOMY

Economy effect is drastically non linear. We can try to estimate each type of Economy change on its own and then summarize combination values for each step on ECONOMY scale. ECONOMY effect change types are: 1) +1 energy per base, 2) +1 energy per square, 3) +1 commerce rating.

ECONOMY (energy per base)

energy increase for +1 ENERGY            = 0.1 * <effective energy intake>
energy increase for +1 energy per base   = 1

+1 energy per base result = ENERGY result * 1 / (0.1 * <effective energy intake>)

ECONOMY (energy per square)

energy increase for +1 ENERGY            = 0.1 * <effective energy intake>
energy increase for +1 energy per square = <base size> + 1

+1 energy per square result = ENERGY result * (<base size> + 1) / (0.1 * <effective energy intake>)

ECONOMY (commerce rating)

Commerce rating is very difficult to estimate as it fluctuates greatly based on diplomacy. Besides, it has quite a complex formula. It would be easier just to estimate it as a portion of total energy intake and then assign it different historical period values based on play testing average.

energy increase for +1 ENERGY            = 0.1 * <effective energy intake>
energy increase for +1 commerce rating   = <effective energy intake> * <average commerce proportion> / (1 + <number of commerce technologies>)

+1 commerce rating result = ENERGY result * [<effective energy intake> * <average commerce proportion> / (1 + <number of commerce technologies>)] / [0.1 * <effective energy intake>] = [<average commerce proportion> / (1 + <number of commerce technologies>)] / 0.1

EFFICIENCY

EFFICIENCY effect is threefold. It decreases inefficiency, decreases number of b-drones (for positive ratings), and decreases penalty for uneven Eco/Res allocation. Let's review them one by one.

EFFICIENCY (inefficiency)

energy increase for +1 ENERGY     = 0.1 * <effective energy intake> = 0.1 * <energy intake> * (1 - <distance to HQ> / (64-((4-EFFICIENCY)*8)))
energy increase for +1 EFFICIENCY = <energy intake> * [min(1, <distance to HQ> / (64-((4-EFFICIENCY_from)*8))) - min(1, <distance to HQ> / (64-((4-EFFICIENCY_to)*8)))]

This is very nonlinear, cumbersome, and depends on the distance to HQ. I won't even try to devise a nice formula just calculate and list these values in table for illustration.

EFFICIENCY (inefficiency) weight
from	to	early game	mid game	late game
-4	-3	3.21	2.01	0.00
-3	-2	0.96	3.02	2.63
-2	-1	0.32	1.01	2.63
-1	0	0.16	0.50	1.31
0	1	0.10	0.30	0.79
1	2	0.06	0.20	0.53
2	3	0.05	0.14	0.38
3	4	0.03	0.11	0.28

As you can see the effect is very dependent on EFFICIENCY current rating itself and even stronger on distance to HQ. The effect is almost negligible for small empires and huge for large ones. The zero in the top right corner means that switching from -4 to -3 rating does not help at all. Still all energy is lost to inefficiency.

EFFICIENCY (inefficiency) WTP

WTP mod introduces two parts to inefficiency formula: the HQ influence and flat EFFICIENCY rating contribution. That reduces non-linearity. You can check inefficiency formula in WTP readme. Here I just show the same value table for illustration.

EFFICIENCY (inefficiency) WTP weight
from	to	early game	mid game	late game
-4	-3	0.71	1.21	1.88
-3	-2	0.43	1.21	1.88
-2	-1	0.00	1.21	1.88
-1	0	0.00	0.24	1.88
0	1	0.00	0.00	1.88
1	2	0.00	0.00	1.88
2	3	0.00	0.00	0.75
3	4	0.00	0.00	0.00

This looks better than vanilla formula not only because it is linear but also because every step on a scale matters. Of course, the bigger the empire the more it matters.

EFFICIENCY (b-drones)

We'll calculate EFFICIENCY b-drones weight by comparing its drone reduction effect to two other means to quell same number of drones: 1) psych increase, and 2) energy reserves increase to maintain pacifying facilities. Then we select whichever is cheaper at certain game period. Number of b-drones per base is calculated assuming normal size map on highest difficulty. Two most common pacifying facilities are Recreation Commons (40/1) and Hologram Theater (60/2). The average cost of these two (per drone) is 25 minerals + 0.75 energy/turn. That is roughly 1 energy per drone per turn.

Obviously, with equal psych and economy multipliers maintaining pacifying facilities is twice cheaper than allocating energy to psych. However, there are only few pacifying facilities. So we use psych method as the main one keeping in mind that early on EFFICIENCY (b-drones) weight could be even lower due to pacifying facilities option.

Quelling drones with psych:

drones decrease for +1 ENERGY     = 0.1 * <effective energy intake> * <psych multiplier> / 2
drones decrease for +1 EFFICIENCY = [<base count> / ((8-Difficulty)*(4+EFFICIENCY_from)/2) - 1] - [<base count> / ((8-Difficulty)*(4+EFFICIENCY_to)/2) - 1]

Again I am not going to convert it into formula. Here is the table values for highest difficulty.

EFFICIENCY (b-drones) weight
from	to	mid game	late game
0	1	0.72	0.97
1	2	0.36	0.69
2	3	0.12	0.51
3	4	0.00	0.38

EFFICIENCY (allocation penalty)

Finally, the penalty for unequal economy/research allocation. Let's take a smallest unequal distribution of 60%-40%. For that distribution you loose 2% for bigger allocation + 4% for smaller allocation per 20% difference per Efficiency level.

energy increase for +1 ENERGY     = 0.1 * <effective energy intake>
energy increase for +1 EFFICIENCY = 0.03 * <effective energy intake>

EFFICIENCY (allocation penalty) result = ENERGY result * [0.03 * <effective energy intake>] / [0.1 * <effective energy intake>] = ENERGY result * 0.3

GROWTH

GROWTH power lies not in immediate economical effect but in economical development acceleration instead. Roughly you'll have 14% bigger bases at any future point in time with 10% growth rate bonus. That gives you 14% more workers => 14% more of both mineral and energy yield.

GROWTH result = 1.4 * (INDUSTRY result + ENERGY result)

Special consideration about GROWTH is that it becomes proportionally less valuable as game progress as it leaves less time for GROWTH effect to manifest itself.

SUPPORT

SUPPORT translates to absolute minerals gain/loss. However, it is not linear and allows free units only if there are so many supported. Here is the table for different SUPPORT transitions.

minerals saved per SUPPORT rating
from	to	result
-4	-3	<number of supported units>
-3	-2	first unit is free
-2	-1	0, don't know how to express free minerals for new base
-1	0	second unit is free
0	1	third unit is free
1	2	fourth unit is free
2	3	all units beyond four and up to base size are free

The -4 to -3 transition is the most drastic one. The 2 to 3 transition is also drastic if there are much more than 4 supported units. The rest of steps are just about +1 mineral per base. Assuming the average +1 mineral per SUPPORT rating we can craft following formula.

minerals gain from +1 INDUSTRY  = 0.1 * <mineral intake>
minerals gain from +1 SUPPORT   = 1

SUPPORT result = INDUSTRY result * 1 / [0.1 * <mineral intake>]

POLICE

Police effect is similar to EFFICIENCY b-drones effect and to PSY effect. They all reduce number of drones. However, this is another nonlinear effect. The summary is in the table below.

drones reduced per POLICE rating
from	to	result
-5	-4	<number of outside units>
-4	-3	first outside unit does not generate drone
-3	-2	outside units beyond first do not generate drones
-2	-1	first police unit removes drones
-1	0	nerve stapling is available
0	1	second police unit removes drones
1	2	third police unit removes drones
2	3	up to three police units remove one more drone each

The progression is greatly depends on how many police units base has and how many outside units it supports. Up to mid game each step would probably cost one drone. Later on number of outside units may grow as well as non-lethal method ability may add one more drone quelling capacity to police units. So I'd estimate 1 drone per rating at game start that linearly grows to 2 toward late game and maybe then even to 3 with more outside units.

drones decrease for +1 ENERGY = 0.1 * <effective energy intake> * <psych multiplier> / 2
drones decrease for +1 POLICE = <POLICE game stage multiplier>

POLICE (b-drones) result = ENERGY result * <POLICE game stage multiplier> / [0.1 * <effective energy intake> * <psych multiplier> / 2]

MORALE

MORALE scale is not exactly linear but we can approximate it as such. It gives +6 morale levels per +8 ratings = 0.75 level per rating. Although, negative values are additionally halving morale facilities effect. So we probably can safely average it as 1 morale level per 1 SE rating.

The way to estimate its result is to realize that each morale level adds 1/8 to army strength or equivalently saves 1/8 of production.

minerals gained by +1 INDUSTRY = 0.1 * <mineral intake>
minerals saved by +1 MORALE = 1/8 * <mineral surplus> * <proportion of production spent on combat units>

MORALE result = INDUSTRY result * [1/8 * <mineral surplus> * <proportion of production spent on conventional combat units>] / [0.1 * <mineral intake>]

Mineral surplus is not the same as mineral intake since some of them are spent on support. Yet, at least after the very early game they tend to be close enough. If we assume <mineral surplus> = 0.8 * <mineral intake> on average over the course of the game the formula reduces to this nice equation:

MORALE result = INDUSTRY result * <proportion of production spent on conventional combat units>

PLANET

PLANET has many applications. Boosting native units combat odds effect is similar to conventional units morale. Therefore, it is worth that much. Another noticeable effects are ability to capture worms and impact on global warming. Ability to capture worms is very nice yet it dissolves with discovering Centauri Empathy when you can stamps them at production rate instead. The global warming part is significant without eco-damage containing facilities. However, players usually do eventually build them even if for other benefits. So the eco-damage reduction is a slight temporary addition. I don't remember a game where I had to fiddle with PLANET rating to contain global warming.

PLANET (psi combat) result = INDUSTRY result * [15% * <mineral surplus> * <proportion of production spent on psi units>] / [0.1 * <mineral intake>]

Counting other bonuses combined I would increase resulting PLANET result by factor of 2. This is pretty arbitrary number, though. Feel free to correct me.

PROBE

This is most difficult to compare effect among them. Probes morale can be evaluated same as other unit morale. However, the proportion of probes is small relative to other units. Therefore, the effect is small too. Another aspect of this SE is making subversion costlier. Which probably should result in less number of subverted bases. However, I don't know how to evaluate and and I honestly don't bother as it is IMHO the least useful effect of them all. Mostly because it makes difference so rare in the game and does not impact game outcome in general. The only exclusion is a Miriam ability to set +3 Probe rating at the beginning of the game when you don't have cover ops centers yet. And even then it makes no difference for her.

Effect weights

Next step is to build a ration of effect results compared to INDUSTRY, for example. I call this effect weight as this number could be directly plugged into SE model evaluation.

Effect weights normalized by INDUSTRY
effect	early game	mid game	late game
INDUSTRY	1.0	1.0	1.0
RESEARCH	0.6	0.5	0.4
ECONOMY (average)	5.5	2.0	0.9
GROWTH	2.0	1.4	0.7
EFFICIENCY (average and aggregated)	0.8	1.4	1.4
EFFICIENCY WTP (average and aggregated)	0.5	0.9	1.9
SUPPORT	2.1	0.9	0.4
POLICE	2.3	1.2	0.7
MORALE	0.4	0.5	0.6
PLANET	0.7	0.7	0.6
PROBE	0.4	0.4	0.4
TALENT	5.0	1.3	0.4

Let me reiterate that these numbers are not by all means exact. Different evaluation approach may yield slightly different results. Yet, this is enough to understand SE relative values and how they are changing with the course of the game. Fixed benefit effects like ECONOMY, SUPPORT are obviously strong at the beginning and decline as bases produce more resources themselves.

Society Models analyzis

Vanilla

The above devised tool allows us evaluate models and compare them to each other. Let's review vanilla models first. Only relevant game historical periods are shown.

Vanilla SE models weight
effect	early game	mid game	late game	average
Police State	7.0	1.5	-0.6	2.6
Democratic	1.5	3.8	3.4	2.9
Fundamentalist	-0.1	0.2	0.6	0.2
Free Market	-2.3	-4.2	-3.3	-3.3
Planned	3.4	1.1	-0.4	1.3
Green	-1.0	1.3	2.6	0.9
Power		0.7	0.1	0.4
Knowledge		2.5	2.5	2.5
Wealth		2.0	0.7	1.4
Cybernetic			2.8	2.8
Eudaimonic			4.1	4.1
Thought Control			2.0	2.0

Apparently, Fundamentalist is a complete waste of a SE slot, which is also corresponds to community opinion. Free Market is very difficult to evaluate due to both non-linear ECONOMY scale and negative POLICE effects those manifest themselves only during offensive campaigns. I'll leave it without comments. Power seems to be under-powered as well especially later in the game when SUPPORT value deteriorates.

WTP v.117

Here is the similar table for WTP SE choice for comparison purposes. It accounts for WTP inefficiency formula. Future society models are also shown for mid game as they can be discovered by that time in WTP.

WTP SE models weight
effect	early game	mid game	late game	average
Police State	6.9	1.7	-2.2	2.1
Democratic	-0.1	1.9	3.4	1.7
Fundamentalist	1.3	1.7	2.2	1.7
Free Market	0.2	-2.7	-2.4	-1.6
Planned	4.0	2.0	-1.4	1.5
Green	-4.6	-2.0	1.9	-1.6
Power		2.1	1.5	1.8
Knowledge		0.9	2.1	1.5
Wealth		1.8	1.1	1.4
Cybernetic		-1.9	2.5	0.3
Eudaimonic		4.4	-0.4	2.0
Thought Control		4.3	2.9	3.6

Fundamentalist and FM look better. Although Green somehow went down a little. I probably need to revisit it. Everything else looks on target.

WTP v.121

Effect weights are adjusted a little internally. Here I just list resulting SE models summary weights with explanations.

WTP SE models weight
effect	early game	mid game	late game	average
Police State	4.3	1.3	-1.8	1.2
Democratic	-0.8	2.0	3.6	1.6
Fundamentalist	1.7	1.7	1.9	1.8
Free Market	0.9	-1.9	-1.7	-0.9
Planned	2.3	3.1	0.9	2.1
Green	3.0	0.0	1.4	1.5
Power		2.6	1.4	2.0
Knowledge		0.3	2.5	1.4
Wealth		1.5	1.1	1.3
Cybernetic		0.3	2.9	1.6
Eudaimonic		2.3	-0.4	1.0
Thought Control		1.3	1.6	1.5

Social Engineering Mod

Contents