The first solution I came up with for the unit production problem was fairly simple: it was a structure that I called the egg hill, which (as it's name implies) was a hill-like structure made of creep that could produce up to a hundred eggs per minute. Each egg was produced from a gap in the structure (Trypophobics would shudder at the sight of it), which would then be picked up by a specially made drone that was covered in tentacles and could carry up to 25 eggs at a time. The Drone would then plant the egg somewhere else in the creep, and from the egg a cocoon would sprout. Then an hour later, a Zerg unit would be spawned from that cocoon.
At first I had to manually designate which unit each egg would spawn, but that got mind numbingly boring very fast, so I handed over the task to two overlords and gave them set rates for each unit I wanted produced each hour.
That should have been my first clue that my design was flawed.
No longer having to worry about unit designations (at least, I thought that was the case), I turned my mind to other matters. By now I had eight of my overlords in orbit of the planet, and one of them had discovered lights on the night side of the planet, indicating that there may be a civilization on this world.
I was still in the process of creating a complete mental map of the world, and was micromanaging resource throughput, when I decided to check in on unit production.
What I found was a complete mess: the creep surrounding my hive was littered with cocoons, many of which had wilted because the creep was having a hard time distributing nutrients to each one. There was too much demand and not enough supply. Then there was the fact that my drones were having to crawl over cocoons or even smash them because I had failed to set up proper supply lanes. Plus, the Lodestones were starting to over heat because the extractors were pulling out more psionic energy than they could handle to keep up with demand.
The Overlords I had assigned to do unit designations simply hadn't done their job very well, as apparently they weren't very mentally flexible. The unit rates were off, and the cocoons that had hatched, produced Zerglings or Drones, but not enough Overlords. And without Overlords, other Zerg units were at a dangerous risk of going feral.
I gave a mental sigh. Back to the drawing board it was.
I gave the order to the drones to halt mineral and unit production, and then had them uproot all the extra cocoons from the creep and started tossing them into Gizzards. I then ordered the Nutrient Storage Tank to flush what was left of it's contents into the creep so that the creep could repair itself. Finally, I ordered the creep to reabsorb the egg hill.
When all that was done, I executed a kill switch in some the extra Units I didn't have enough overlords for, and had their bodies tossed into the Gizzards.
Then I ordered mineral production to continue, and for unit production to start again the original way.
Now, to think: How could I learn from this experience, and not repeat the same mistakes?
Well, I thought the biggest problem was how the cocoons got littered everywhere on the creep, and how the creep had a hard time distributing resources. Now, that problem could be solved by being careful with how, when, and where cocoons were placed, and by making a dedicated resource funnel network, but I had the feeling that it wouldn't solve the fundamental problem.
I had plans to spread out and form new Hive Clusters in the future, and I didn't want to have to waste time micromanaging unit production. I wanted my unit production to be self managing, and I had the feeling that would be much harder if all my cocoons were spread out across the creep. For one thing it would increase travel time for drones carrying eggs to their cocoon sites, and any intelligence I created to handle the problem would have to adapt to new terrain as the Zerg Swarm came across it, which would be a logistical nightmare for a race that was constantly on the move.
I was starting to understand why the Overmind had never bothered to improve it's unit production system.
What I really wanted was a structure that could handle it's logistics internally, automatically, without having to worry about how the surrounding terrain might change those logistics.
So, I decided, lets go back to basics. The Zerg Unit Production System was fairly simple. A large, upgrade-able structure called a Hatchery was the source of Zerg Units. It made units by producing a large grub like unit called a Larva. When the larva was made, it would plop down onto the creep and start eating it like food, and when it had eaten enough, it could turn into whatever zerg unit the Overmind or Cerebate required of it, then go about it's business.
This was as logistically simple as you could get, trading off production efficiency for logistic simplicity, such that the Cerebate in control of that particular brood didn't have to worry about where, when, and how his units will get produced, allowing them to focus on other, more important things, such as whatever larger task they were engaged in.
So the question was, how could I improve upon this system without sacrificing logistic simplicity?
On a whim, I decided to read the genetic sequence of the zerg Hatchery to get an idea of how it worked, and noticed several areas that could be improved right away. The main thing was that the Hatchery doubled as a resource consumption structure as well as a unit production structure. Taking out the resource consumption parts would make a lot more room for unit production parts, and would improve unit production by a factor of 2, allowing the Hatchery to produce 18 Larva per hour, instead of just 6.
Another part was the way that larva was produced. Larva were created in womb like sacks, and it took them two hours form. At any given time there was 18 Larva being baked in the oven, with 6 outputted per hour. With the new system that number would rise to 54, with 18 outputted per hour. I could improve this by restructuring the genetic sequence to make eggs instead, and a lot of the more inefficient systems meant to support the womb like sacs would be removed, allowing me to produce 100 eggs per hour, or even more.
But then I would run into the cocoon space problem, again. Ugg, I felt like screaming. Mentally, of course. My mouths couldn't scream. Who knew logistical problems could be so hard?
Clearly, I was approaching the problem from the wrong direction. With that in mind, I decided to cast my mind into an overlord who was floating a kilometer high above my hive cluster. I had been looking at things from the bottom up, maybe the top down view would change how I looked at things.