I wouldn't be surprised if earth cities end up under domes someday. It would offer protection from big storms (including snowstorms which often shut cities down in NE America), extreme temperatures, falling debris, attack, and most importantly, these $%^ mosquitos!
Efficient Cities:
Yes. Domed cities control atmosphere inside the city, controls pollution and recycles important gasses. Also protects the inhabitants and structures from radiation and other incoming stuff, not least the biological hazards from outside, and from inside out.
Building cities on pillars, making dome cities like big mushrooms, makes air ways easier on high biological worlds or high spine worlds.
Realty is precious, so they will pack it in 3D, much like Manhattan today, but there will also be lots of interconnecting subways and highways as tunnels between the structures inside. It is purely mathematical efficiency. Less space, more efficient.
Psychology affect cities:
Humans rather live on widespan surfaces, but then we have to move up into orbit or stratosphere, because there surface space is endless or almost endless. But more expensive.
Living on the actual surface of a world may prove hazardous unless it has been fully terraformed, or is Earth. But even then you are subject to hazards such as storms, biohazards, old tech pollution and waste, earthquakes and temperature differences of the seasons. Living near Tjernobyl or u-shima requires these type of domed cities even today, nobody can live there for thousands of years, and the number of such "places" will grown over time. Just 100 years and already we have 2 such high radiation areas on Earth. Think what it will be like in 1000 years from now?
Social Constructs:
Owning and living on a surface of a fully terraformed world on a big ranch is only for the most wealthy, or those with enough money that they do no longer have to work anything except their lands ( using Semi-Sentient Machines as workforce or likeminded people ). Owning and living on a surface of a non-terraformed world however, can be super cheap... because it is awful.
Over the thousands of years with genetic engineering running its course. Humans of different worlds will develop different traits. Such as red skin, green skin, yellow skin, thicker skin, bigger eyes, smaller eyes, bigger ears, longer limbs, smaller noses, an extra eye.... etc etc... all because engineered evolution prefers that type on that world. This will create new social constructs and divide people into worlds.
The Inevitable truth about space colonization:
Differentiation is something wonderful, but will perhaps be frowned upon by the "originals" that strive to keep humanity uniform. Differentiation also applies to mental abilities and mental strive, meaning conflicts, communication and love of things will be different among different human world races.
When millions of years pass, human races will evolve so separate that even with genetic control, there will be difficulty interbreeding, causing humans to form separate species of extreme variation.
Technology and artificial biology will happen, and with that the real speed of evolution begins.
Simulating these cities and their social communities:
Will be a task that may be considered imaginative to say the least. Artists, coders and designers have free hands, as long as it is all interconnected using common sense. Making procedural cities that just look like 21th century communities in the 33th century could make sense for low tech worlds, but the span between the societies in tech and abilities will be huge. Explorers should be able to explore different world community cities as vigorously as any space exploration ever could provide. The differences will be notable. The unknown cities will pop and get lost almost faster than we can explore.
A procedural construction require: origins, genetic traits, social structures, social function, economy and environmental states. And that just to begin the foundation of what any such cities would look like.
What is its purpose, function, progress and fate? All these questions must be within the simulation. Adding several simulation layers to simulate history of such simulations is also preferred. This because we want to read the history of the community and its cities, not just go there, look at a bland structure heap and realize it has no depth. History and current events makes depth. And it can all be done with simple numbers in math space. Nothing have to be saved to memory. It can be procedurally regenerated by simplex functions over and over as needed. All we need is a seed, in the form of time, place and starting state. It can then be procedurally generated forward in time, or backwards in time for its history. The marker just moves along the timeline and the procedural "living" thing, that is the community/city, changes and moves as it goes.
Hierarchical Procedural Universes:
If two cities or communities happens to "grow" in over each other in space and in time, the simplex functions just adds together, making a slightly more simplex math space (meaning not necessarily a complex) . If a community seeds a new world, its simplex functions just moves along with it. Simplex functions are like DNA for math space. Making it all hierarchical in nature, a parent simplex function of the universe can control how sub simplex functions moves and progress over the histoverse ( universes over time, as in the history of all the universes ). What the most parent simplex function is, or how it came to be as it is, that is up to the creator to decide.
Player interaction:
Player interaction is the only thing that alters the simplex in any way that would require some sort of memory. These events should be packed into historical packs, that is preferred rare but can be approximated as continuous and intact over math space. A player can shoot a building in a city, causing it to "repair" its simplex by adding overlay simplex functions to simulate that behavior. Just as a meteor strike causes damage to a planet surface, but a player induced meteor strike that causes damage to a planetary surface require some form of "mending" the natural simplex, or require a memory of that event.
This can be solved by letting the natural simplex fabric repair itself over time so that player interaction becomes insignificant, yet plays occasionally a role for the history of the universe. But those occasions will always require a memory block containing that important event and how it altered the simplex. I can not deny that. But letting all other insignificant events "fade" away using less and less factoring in of that injected simplex will keep a balance to the number of simplex that interconnect and produce the final history of the universe.
In layman terms: If a player destroys a planet, it will most likely reform over a loooooong time in history, but until it is fully reformed, the significance of that event have to be recorded and will influence the rest of the local universe.
If a player destroys a bug, that is easy to replace and has no significance in more than a day or two, tops.
Conclusion:
So there you have it lads and lasses, I created the entire universe for you in concept. Get busy.
Frontier Developments should be advised.