So in the interest of the economic discussions (and maybe some help for Brad in coding the AI) I took a look at a simple challenge.

1) A Q10 planet. No crazy tiles or bonuses. I have put a very very bad picture of it so you can the configuration.
2) Start with 2.5 population (the default for a new colony ship). Also base Morale starts at 2.4 (meaning actually little expansion has occured).
3) Start with basic infrastructure (Xeno Farm, Colony Hospital, Entertainment Network, Basic Factory, Basic Lab). Assuming no racial bonuses or extras from technology. However, I did assume the 25% research bonus for "normal research".
The goal: Optimize the planet to generate the most cumulative research for 75 turns.
To do this required a lot of excel work, if anyone wants to see it let me know. I will put some math notes at the bottom of the post for those interested in that.
To read the work below it works like this:
A# (##)
A - letter representing the building (H = hospital, F = farm, M = factory, L = Lab, E = entertainment network)
# - There are 10 buildings built, this is when that building was built. Everything is a lab unless I specify otherwise.
## - The tile it was placed on (see the picture)
I went through a lot of configurations, but several patterns emerged:
Max the Slider to 100% social when building, until there is nothing left to build.
1) All labs: Kept slider to 100% until everything was finished: 2,424
2) All labs: Drop the slider by 10% after each building finishes: 1,160
3) All labs: 100% until 5 buildings done, than 50%: 2,280
4) All labs: 100% until 8 buildings done, than 50%: 2,413
I tried a number of non lab configs as well with adjustment of the slider, but in every case maxing the slider did the job. Which is great, as it made much of the rest of this so much easier!
Factories right? Wrong!
Most people would think 2-3 factories is the way to go. But those configurations performed badly.
1) M1(1): 2,584
2) M1(1), M2(5): 2,473
3) M1(2), M2(4), M3(5): 2,262 (note that this configuration gave the factories a good adjacency bonus, its just not worth it).
So the winner? Apparently scientists like fun too!
The lowly entertainment network did the job, by a big margin!
1) E1(1): 3,184 (23% better than a factory!)
Approval affects both production and growth, and this double whammy is actually very powerful, more than the factory.
How about farms? Scientists hate farming!
1) F8 (1): 2,507
I tried dual farms as well, and farms with hospitals, and hospitals by themselves. All of them did terrible, the single farm was the best, but not as a good as a factory or entertainment network.
It seems 75 turns is not enough to get the advantage of that larger population, or it may require a lot more base bonuses to make it happen.
What about Arson? No good!
I also tried configs where I would finish building, than destroy the initial buildings and build more labs. They did not do well.
1) E1 (1) -> destroyed, L11 (1): 2,532
2) E1 (1) -> destroyed, F11 (1): 2,597
In summary:
1) Approval is very important to colony development, and is first priority (this surprised me!)
2) If approval is maxxed, a single factory is optimal. No more than 1.
3) Farms and Hospitals are not optimal for a 75 return window...at least not without significant other bonuses.
I may try a few more with a strong hospital (and increased growth), as well as greater bonuses to see where population starts to win out, but for now there you have it!