Industry, Growth, and Economy

[Chaos Theory]

After fiddling with Impaler's mod, I discovered some undocumented or unexpected SE effects, particularly regarding industry, growth, and economy:


Industry and Growth

In a standard game, your nutrient tanks and mineral boxes are 10 wide, before SE mods. Each + reduces the width by 1 and each - increases it by 1. What if you change the starting widths, by messing with alphax? Then each + and - represents a 10% shift, *not* a 1 column shift! The result is that the formula for the width of a nutrient tank or mineral box is
ceil(base * (1 - .1*rating))
Where ceil is the ceiling function,
base is the width of the box with 0 industry or growth,
and rating is of course the appropriate SE effect rating, bounded to be within [-2, 5] for growth, and [-3, 5] for industry.

Furthermore, on Transcend difficulty, the AI has boxes of base width
floor(.7*your base),
or 7 wide with an unaltered alphax. During the course of games, I have observed these base widths decline to 6 and even 5. The dates at which they change are not fixed and probably depend on how the AI is doing. This explains why Lal can have nutrient tanks 3 boxes wide (base width 6 and +6 growth or base width 5 and +4 growth).

Industry < -7 overflows and acts as if it were +5.


Economy

The effects of negative economy are not properly documented. It turns out that for an econ rating of less than -1, the effect is that each base tile loses (-rating - 1) energy. Base tiles can never give negative energy, but recycling tanks contribute to their production before energy is subtracted for low econ. Therefore, if base tiles produce 10 energy and recycling tanks add 10 energy, but you have -20 econ, those base tiles provide zero energy, or one energy with rec tanks. Headquarters and rivers seem to add after this step.


Research

After a little testing, I have observed that changes to the research rating via SE have NO direct effect on tech cost.

Oddly, when the AI was given the choice between -48 research and economy, or +32 research and economy, it usually picked -48 research and economy. When given the choice between -48 economy and industry, or +32 economy and industry, it always picked +32 economy and industry.


Talent

Yes, this is a SE category, but it doesn't appear anywhere except a few places in alphax. It was apparently abandoned in development, but it still functions. Government choices can contribute to Talent, but their contributions will not appear on the SE screen. Your talent rating has the following effect:

Talent = X, X > 0: X extra talents, added in the same way as Lal's talents
Talent = 0: nothing, of course
Talent = X, X < 0: X extra drones, added in the same way as Zak's. This also means that on transcend, Talent < -1 has no additional effect.

[Chaunk]

Re: Industry, Growth, and Economy

Originally posted by Chaos Theory
Research

After a little testing, I have observed that changes to the research rating via SE have NO direct effect on tech cost.

Oddly, when the AI was given the choice between -48 research and economy, or +32 research and economy, it usually picked -48 research and economy. When given the choice between -48 economy and industry, or +32 economy and industry, it always picked +32 economy and industry.

Sorry, could you explain that? I don't quite know what you're going on about... The rest is very interesting indeed....

[Maniac]

Re: Industry, Growth, and Economy

Originally posted by Chaos Theory
Industry and Growth

You can also notice those effects when playing on Small or Tiny maps, where respectively only 9 or 8 nuts/mins are needed to fill a row. This leads to many problems, for example the Hive doesn't get their +1 Gro/Ind boni, and for most factions Planned SE only gives +1 Growth instead of +2 Growth and +1 Industry.

[Chaos Theory]

For Chaunk:

To test the effects of extreme growth and industry, I altered the SE government choices so that Police State was -48 research and economy, and Eudaimonic was +32 research and economy (and the prereqs were removed), the AI usually picked Police State politics and None future society. When I replaced research with industry, the AI always picked Frontier politics and Eudaimonic future society.

[binTravkin]

This means that AI SE evaluation regarding RESEARCH is quite buggy, but if he sees any INDUSTRY benefit, it will go for it..

Have you tried setting pure +32 & -48 economy?
What does AI think about that?

[Impaler[WrG]]

Interesting, I guess their motivation to make every unit of Groth and Industry = 1/10 of the bar was to keep the relative value of these effects constant regardless of how large the total bar is. That would insulate the game Social Enginering settings from changes made to the slots in a row.

Unfortunatly this will likly screw up many of the changes I have been trying to make because thats adsactly WHAT I want to do, make rows 12 across and decresse the relative effects of each level of Industry and Growth for better balance.

Chaos how do you think I could best get around what Firaxis did and get a system of 1 growth/industry = 1 reduction in a row cost. At what point dose the equation round up and start deviating from strati 1:1 ratio?

The news on Economy is pleasing though as it means that their CAN be a meaningfull penalty to very low Econ raitings and this can be used for Balancing combinations that relust in low Econ ratings (particularly with the Hive)

[Googlie]

Re: Industry, Growth, and Economy

Originally posted by Chaos Theory
Research

After a little testing, I have observed that changes to the research rating via SE have NO direct effect on tech cost.

Correct. They impact your research rate (lab points generation), not the cost of the tech you are researching.

Prima's Guide gives a simplified version of the formula used:

Research cost=[11 + (DIFF x 4) + {(TECHS x 6) / 5} - (TURNS/8) - (MOSTTECHS/5)] x TECHS x WORLDSIZE

where:

DIFF = Citizen = 0 up to Transcend = 5
TECHS = # techs you have discovered as a player
MOSTTECHS = # techs discovered by any one player
TURNS = # turns already played
WORLDSIZE = 1 if standard, (.6 tiny; .8 small; 1.1 large and 1.6 huge)

Some additional latitude uis given the program to increase or reduce costs by between 2% to 15% depending on the relative strength/weakness of the faction in overall research points per turn.

Prima concludes by saying:

As you might be able to see, the greatest factor is how many techs you yourself have discovered

[Chaos Theory]

Impaler: These are some tables generated from the formula I derived:

Base Rating Width
11   -3    15
11   -2    14
11   -1    13
11    0    11
11   +1    10
11   +2     9
11   +3     8
11   +4     7
11   +5     6

12   -3    16
12   -2    15
12   -1    14
12    0    12
12   +1    11
12   +2    10
12   +3     9
12   +4     8
12   +5     6

13   -3    17
13   -2    16
13   -1    15
13    0    13
13   +1    12
13   +2    11
13   +3    10
13   +4     8
13   +5     7

14   -3    19
14   -2    17
14   -1    16
14    0    14
14   +1    13
14   +2    12
14   +3    10
14   +4     9
14   +5     7

As you can see, when base / 10 is not an integer, the change from a rating of -1 to 0 is always one larger than the change from a rating of 0 to +1.

Rather than changing the row width, if everyone starts with low industry and growth values, each + means less. For example, if everyone suffers a -2 industry penalty on top of other mods (yielding -3 for Santi, -1 for Yang), then mineral rows are 12 wide, minus your relative industry rating. The side effect is that the effective range of industry is now [-1, 7], but power's downside is too harsh anyway.


Googlie:
I checked the effect of Research because I still see people claim that your research rating affects your tech costs.


binTravkin:
Worth a try.

[Chaunk]

I'm fairly sure it does... As the believers I always have a massivly inflated tech cost at the start of the game. I remember reading an Acol post saying the extra tech cost for a negetive research bonus in the faction file was 2x for the first two techs, then 1/2 extra cost, 1/3, 1/4, 1/5 and so on. Thats for any negetive bonus. I've never checked that, cos it always felt about right.

For instance, in a recent 3x2, my factions are Miriam and Morgan. My starting tech costs were 32 & 16 IIRC, which is precicly what I'd have expected. During the game, I don't think it makes a difference, but at the start of the game from the faction.txt, it does I think.

Will run some tests though.

[binTravkin]

Yeah, I think that actual techcost is calculated in COMPLETELY different way:

I have seen WAY too many games where the player with the best labs score gets the techcost at least halved - there were times when I was running very close to my opponent & built a few new cities while he didnt expand.
What we got was a major techcost switch:
Prior to building those cities I had tech/9 turns, he - tech/5turns
After: I - tech/4turns, he - tech/7turns
And those bases were only size 3 towns with no lab facs in them

And I have expierenced techcost switches in that way very often.
At the lowest level too - Im Miriam having IIRC ~ 4bases, my opponent - Morgan having 3 bases

He had 40-50% more labs & he had 10+turns/tech & way more techs than me. I had 30+ turns/tech..

It was something around 2125..

[Chaos Theory]

I investigated the effects of factional Research mods on tech cost. To research your nth tech (ignoring any granted at the start of the game), any faction with any negative factional Research must pay n/n-1 (or 2 when n = 1) times as much as a faction with zero factional Research.
Any faction with positive factional Research must pay n-2/n-1 (or 1 when n <= 2) times as much as a faction with zero factional Research.

The effect of these factors is to give factions with positive factional Research a net ~.5-1 tech lead, before accounting for the actual magnitude of the Research bonus. Likewise, a faction with a negative factional Research suffers a ~.5-1 tech deficit before accounting for the direct effect of receiving fewer lab points.

[johndmuller]

I also frequently get the impression that the research penalties and bonuses seem to be exagerated by an impact on tech costs as well. It is, however, difficult to nail it down in a game, as people are often getting additional techs in the middle of researching stuff, so a simple snapshot may not reallly reflect the tech costs associated with the apparent number of techs at the moment.

I just invested some serious time checking through am archived game where I was pacted with 2 other human players from nearly the beginning. I was expecting that I would find evidence that the Drones were penalized in the tech cost department relative to their two 'normal' partners (Morgan and PK). Unfortunately, after the Drones' tech cost rocketed up to 70 due to exchanging techs with us other two, they got spooked and henceforth stayed generally behind in techs, making direct comparisons elusive. FYI, here is a table with selected snapshots of the tech cost situation (until I got tired of looking for evidence)
.

            Drones             Morgan             PKs
Year      #Techs / TCost     #Techs / TCost     #Techs / TCost
----      ------   -----     ------   -----     ------   -----
2102          1       ?         1       14          1       ?      Initial Turn, no pact
2103          1       ?         1       14          1       ?      Still not pacted
2104          1      28         2       33          2      33      All Pacted; M & PK traded techs
2105          2      49         2       33          2      33      Biogen from M to Ds
2109          2      46         2       31          2      31
2110          2      46         3       53          2      31      M's & PK's 1st Breakthrus (PK's later in turn)
2111          3      70         4       53          3      31      PK's New Tech distributed
2112          3      70         4       53          4      53      M's tech to PK's only
2115          3      70         4       53          4      53
2117          3      66         5      105          4      53      M's New Breakthru
2120          4      88         6      105          6     105      M's received tech being researched from PK
2124          5     119         6      105          6     105      D's First Breakthru
2125          5     119         6      105          6     105
2127          5     119         7      162          6     105      M's & PK's get Breakthrus (PK's later in turn)
2130          7     119         9      162          8     162
2132          8     229         9      162          8     162      D's had new breakthru
2135          8     229         9      162          8     162      PK's had new breakthru after M's turn
2136          8     229        11      353         10     298      M's had new breakthru

.

What this does show is that there is indeed some sort of initial tech cost penalty for the Drones, but I don't think it clearly shows any ongoing penalty, just some sort of decaying penalty at the beginning. It looks like it might be a x2 factor at the very beginning, a x1.5 penalty with 2 techs, a x1.333 with 3 techs, pretty much what Chaunk cited as being posted at ACOL. By the time the Drones got their first own discovery, freezing their costs, they were up to 5 techs, which still seemed to qualify them for a penalty (119 vx 105). When they got their next breakthru, they would be at 8 techs (for which there was no direct comparison as the other two went from 7 to 10 or 11 techs for their breakthrus - the Drones' 229 at 8 techs compared to PK's 298 at 10 and/or PK/Morgan's 162 at 7. It looks like there is still some sort of penalty going on there, but it is hard to say for sure. Perhaps I'll run some more numbers further on in the game.

It does certainly demonstrate the inadvisability of acquiring additional techs before researdhing your own first tech, as that seems to cause a recalculation of tech cost greatly to your disadvantage

[Commy]

Okay, I haven't really been to this site for a couple of months, and I really don't understand half the stuff you guys are saying, but the AI SE choices are based on what gives the most energy in the future. So basically, the AI will usually do planned economy in the beginning (usually) to increase growth so that in the future, the AI can do whatever that will give them a lot of energy then. Kind of confusing? Basically, the AI wants the maximum amount of energy produced in the entire game, not on the present turn. Using formulas and calculations, it makes the right decisions that will allow it to get the most energy in the future. The problem is, this way doesn't hold the AI factions to their idealogies and the AI doesn't know when the game ends, so they always plan for 200 years after they will have to "retire".


If any body has already said this I deeply apoligize, for I haven't read this entire thread and am not trying to steal others ideas.

[Chaos Theory]

Well, that doesn't explain why it picked a SE category with two -48s over one with two +32s, as I posted earlier. I still need to run some experiments to tease out the details of the AI's SE preferences.

[binTravkin]

Well, that doesn't explain why it picked a SE category with two -48s over one with two +32s, as I posted earlier. I still need to run some experiments to tease out the details of the AI's SE preferences.

Some kind of overflow bug?