The ai constantly tweaks the sliders. Every turn.

 

Agreed, I would love to see a journal post explaining how the Stardock economy researchers came up with this current system and why it is cool, awesome, cutting edge, etc..  Every other 4X game has production / research, etc. that is essentially "produce and forget" rather than having to build the building AND allocate to it. While this gives some flexibility, it creates a lot of micromanagement, and you then feel stupid for not exploiting its weaknesses by having all manufacturing worlds, which end up giving more research in the end!!  Also, the allocation mechnism produces another complicated point that the AI struggles with.  I guess for now, we can all rest easy that Skynet is still a ways off in the if advanced AI can't figure out how to maximize sliders, then we wont have plasma rifle wielding terminators on our doorstep all that soon...

the easiest way would be a simple a change of the current system, i.e.

to

 

i.e. you no longer run raw production through two modifier pipelines, thus preventing double dipping and weird outcomes as described in reply 15 (page 1)

 

edit: i still would like to see some (ingame) documentation of the diminishing returns of population on raw production, since that was changed since the blog post that introduced the economic system.

 

My investigation is that this must be very subtle change if it actually do it at all. It is quite easy to see when you have the AI play your own race and when you spend some time studying other AI planets with FOW turned off.

 

I have yet to see the slider set to anything but 33/33/33... although I have not studied the RAW numbers so perhaps I should. In a nutshell the AI always seem to be very centered and I have never seen the AI use close to 100% in any area ever so far in my observations. That is unless what I see is completely different from what is actually happen, but I doubt it so far when I consider the overall efficiency of AI colonies.

 

This seem very easy to check and I will...

 

**EDIT**

The AI do seem to manipulate the sliders when you look at the RAW numbers and in the beginning of a game it is about 60/40 production/science and then about 50/50 for social/military production... certainly better than a constant 33 split but not too effective. The AI also seem to prefer building shipyards on almost every system and do so very early in their development cycle, sort of wasting time and resources. But I suppose the AI is a continual work to perfect...

@tesb

joeball123 explained it well, thank to him.

*

My suggestion for the functioning of Projects:

I would prefer a solution where the different projects are completely disassociated from the manufacturing production setting (the planet production "slider" setting (the "wheel") for % of manufacturing), and act as independent bonuses.

They should give flat and % bonuses in a way of complementarity, flat bonus for the improvements with % bonus, % bonus for the improvements with flat bonus (or even % and flat bonuses at the same time in some cases), and, perhaps (but not fundamentally needed), then be modified by the Population on the planet (more population, greater bonus).

Example: (all numbers are for explanation only)

Research Project: +2 points (flat bonus to act with the % from the improvements (but can be flat and % bonus at the same time)) of research ("similar" to raw production) x Population/10

For 10B Population, with 100 % bonus from improvements:

Research Project: (2 x 100 %) x (10 / 10) = 2

--> 2 points of "final research points", not modified by the planet production setting (the "wheel"), thus completely independent.

For 100B Population, with 100 % bonus from improvements:

Research Project: (2 x 100 %) x (100 / 10) = 20

For 100B Population, with 1000 % bonus from improvements:

Research Project: (2 x 1000 %) x (100 / 10) = 200

So, the Research Project is only influenced by the Research improvements built on the planet, and perhaps also by the Population (to make it different for each planet, and more fun to use).

And the same for the others types of projects.

And, when there is no improvement and no ship to be built, the manufacturing points generated (the "wheel" setting) are directly redirected to Research and Wealth (with some penalties), in consistency with the player's settings for Research and Wealth (the player's setting is conserved and readjusted when there is again improvement or ship to be built).

I saw the changelog on the latest version changed the system, yet chengelog documentation doesn't say how the new system works.

Can anyone tell me how the new system works now in game?

 

Judging by the scope of the latest changes (tech specializations, production model, trade resource stacking etc) I would say that Brad is having a great nerf party given that he can't make the AI follow his original game design. [no offense meant]

 

 

Without even thinking about it very hard, that's soo meh, you'd have to cut my arm before selecting a project like that on any of my planets.

So we're back to building 100% specialized worlds and rush buying labs and markets everywhere else, cause manufacturing + other isn't worth it?

Ahhhh, this should be so simple. Don't do percentages and don't do raw production. Just have the research and economy projects convert manufacturing to research or credits at a loss. So, say, you get 3 research for each 5 manufacturing. This would not be amplified by any research or economy modifiers. This way, you'd be better off never running these on research or economy specialist worlds, but it gives you something to do with excess manufacturing if you don't want to build more ships.

i tend to agree, in the end conversions complicate THIS economic system (where allocation is done via sliders) too much. i'll give feedback on the new system once played with it for a while.

Having checked the numbers shown in the game, it appears that the output O of a chosen project's type is given by

O = x*P*(1 + Bx + 0.05*ceil(s*(1 - x - y)*(1 + Bm)*P/10))

where x is the fraction of planetary production dedicated to the project's output type, Bx is the planet's bonus to the project's output type, s is the fraction of manufacturing dedicated to social production, y is the fraction of planetary production not dedicated to either manufacturing or the project's output type, Bm is the planet's bonus to manufacturing production, P is the planet's base production, and ceil(X) is the function which returns the smallest integer N such that N >= X. The ceiling function is used because the projects appear to only give bonuses which are integer multiples of 5% and appear to round up to the nearest integer when determining which multiple is used.

This means that if you can get a bonus of r*100% from the project by sacrificing (1 - x)*100% of the planet's production to manufacturing and dedicating the remaining x*100% of the planet's prodcution to the chosen output type, the project is beneficial so long as the planet's bonus to the chosen production type is less than Bmax, where

Bmax = [(x*(1 + r) - 1) / (1 - x)]*100%