We've set out to apply genetic algorithms on SpatialOS, as something that lacked efficiency in scale and innovation in "spatiality". This slowly turned into trying to simulate large populations of unique genetic individuals, moving around an environment, consuming and producing resources, reproducing when meeting each other. We've come across many issues, overcome most, but ultimately falls short of what we imagined.
The difficult part was SpatialOS' lack of documentation and complex layers of code infrastructure. We got through that too late, and although our ambitions were high, in the end we didn't manage to finalise our simulation. Our hope is that the project made sense from the point of view of Improbable and we could potentially lay the groundwork for an evolutionary simulation that innovates in its scale.