Computer chips have been two-dimensional for quite a few orders of magnitude change each in density and speed. I think you're trying to think of your em world scenario as driving change against resistance, rather than trying to guess the equilibrium between constraints, that is, the ways various unreasonable-seeming constraints might intrude qualitatively into your scenario.

Communications bottlenecks relative to computing speed might re-introduce the analog of significant distance, and produce communities that had to be more autonomous. I mean, more informationally self-sufficient. You would have to have copies of information to work with, and do work that would take a subjectively long time between, say, getting the contract and delivering the result. Or you would have to spend time guessing your market subjectively further out.

Expand full comment

this reminds me of accelerando where humans aren't so much exiled from the inner solar system as simply outbid for real estate.

Expand full comment


Here's a "cute" example of a combo auction that I think would help you sell the concept. The goal of intellectually marketing something like this is to make it relatable and neutralize the "weirdness."

With that said there are many different names for grandparents in Western culture: pappy, poppy, pop-pop, grandpa, gramps, grandma, gammy, mimi, nana, gigi, etc.

To prevent confusion each common set of grandparents should have different nicknames. Each kid should have a different nickname for their maternal and paternal grandparents. Each grandparent should have a consistent nickname for each of their grandkids, regardless of who their other grandparent is.

This forms a complex network of preferences and mutual exclusion. A combo auction could solve this problem by having each grandparent bid on their preferred nicknames ahead of time. To simplify the model the whole generation is all born at the same time, so all pairs of grandparents are known ahead of time.

Expand full comment

How does this compare to the more sandboxy MMORPGs?

Expand full comment

Why do you folks think that most human work in rich nations today is office work? It isn't a ruling class conspiracy, really.

Expand full comment

Very generally, trying to anticipate the requirements of somethinghalfway between a data center and a city "three to fifteen decades" outis really stretching anyone's foresight. This post is a good try.I think one can make a good case from fundamental thermodynamicsthat the power and cooling requirements will be there in some form.Everything else is uncertain on that timescale.

Perhaps Drexler/Merkle nanotechnology will make very local (~1 meter)recycling of atoms as pervasive then as electric power is now, andpaths for physical artifacts and wastes will be a smaller consideration.Perhaps the computing machinery will be so radiation-sensitive thatvarious radiation fluxes of city parcels will be a major consideration.

Who would have expected, in 1900, that uranium, and particularly U-235,would be an important material in the 21st century? Some similarresource flow might become unexpectedly (to us, now) important inthe next fifteen decades.

Expand full comment

I have to agree...

I'm not a neophyte, but this just seems plain nuts.

Unless the point is to please a ruling class...at which point it becomes very much little to do with combinatorial equality...

Just make sure the emulated brains are all efficient (a problem Jeffrey offers a solution to), something a combinatorial voting scheme seems to do less well than a hierarchical control structure...I'm not sure you fully appreciate what emulating brains on massive scales means...

Expand full comment

I do not understand the assertion that the most economically useful use of a emulated human brain is office work.

Expand full comment

Software API may not quite be the right consideration,but communications bandwidths with various flavorsof protocols and hardware layers could well be.

You have:communication distance from other particular residentsandline of sight to outside, or to specific parties

which are important metrics for some communicationstechnologies, but not for all of them today. If bandwidthsare high enough that format conversion is a substantial cost,then data pipes with different formats might be as hard tosubstitute as coal and propane are today - or, to useJohnicholas's example, as metric and imperial hardware are.

Expand full comment

Unless there is an interesting interaction between city land use choices and some software system's choice of API, I don't see much reason to include them both in the same system.

Expand full comment

I was thinking about coordination problems and growth of cities in a hardware store today.

Shapes have complements - a 1/4 inch nut is valuable because there are 1/4 inch bolts, and so on. After a while, the whole system, imperial vs. metric, is a coordination problem, just like deciding where a city is going to arise.

APIs and computer languages are coordination problems like cities and measurement systems. Your list focuses on locations, but APIs and programming languages might be just as important or more important.

The "Religious wars" of the net might make more sense if you translate them to fighting over which city will win out and dominate trade; it's pretty much the same reason wars have been fought between actual cities.

Expand full comment