The networking of intentional communities, once established, would allow a specialization scheme for greater efficiency and robustness of the network. Due to the four-color theorem we know that 4 or more specializations are required. I propose two sets of disjoint specializations: 
We also need to define a time parameter, T, which is the time taken to transport a bundle of resources from the adjacent node to this node. 
the demand over the time taken to transport the materials demanded, plus that demand times the probability of failure
the quantity of v* times the degree, plus the sum of the probabilities of failure times the demand, plus the time to replace the quantity demanded itself via the set {v*}
Keep in mind, both are derived from thought experiment and have no empirical support, but I think they’re general enough to suffice in terms of both shortage protection and efficiency, compared to any market mechanism.
Now, for the other typifying set, 
Each of these members is a type of specialization for an IC/VIAAC (henceforth referred to as a “vertex”) in the network.
F is a fabrication specialization. An F-class vertex would, at the least, have a multi-machine, CNC mill, or similar tool for making arbitrary, full-strength parts. The machine shop is the main production building in an F-class vert. All unspecialized vertices should probably start as an F-class, because an F-class could hypothetically provide an easy way to reach any other specialization. Early on, F-class vertices should produce construction equipment (earth rams, CEB presses, or contour crafting machines) in order to help the establishment of new vertices, since that allows for greater connectivity, robustness, efficiency, and specialization.
If there is an unspecialized vertex adjacent to an F-class (adjacent as in within a practical distance to transport resources, or especially large pieces of equipment), the F-class vertex can specialize to one of the other classes via a “molting” operation, where the production capabilities of the F-class vertex are used to create specialist equipment of another class, while simultaneously transferring its existing equipment to the unspecialized vertex. Assuming transportation is not impractical, this could be far more efficient than the unspecialized vertex trying to bootstrap a specialization.
All nodes should take care of their own food and energy needs, plus a surplus for its neighbors, following the demand replacement theorem. F-class vertices would be ideal in locations with renewable mechanical energy, such as on rivers or in areas with high wind. Wind and hydro energy can be made much more efficient by using them to directly drive mechanical devices, which is an ideal situation for machine tools. This does not necessarily require machine tools to be located at the base of a windmill or hydro plant, either; mechanical transmission of energy over a distance can be done more efficiently than converting it to electricity and then back to mechanical.
A mix of solar electric and thermal would be the logical next step for energy in an F-class locale. Many manufacturing processes require primarily thermal energy, meaning solar thermal is a much more efficient way to go about powering them. With both mechanical wind power and solar thermal power, it may even be possible to carry out manufacturing operations without using electricity at all.
F-class vertices should be very easy to build, since there is a great abundance of open-source machine tool designs, most of which can be built at extremely low cost (multimachine, various open-source CNC tools, RepRap, MetalicaRap, and subsequent tools become easier to build, such as brakes). The parts they can produce are fairly high cost on the market, especially when they are uncommon designs and not in truly huge quantities, but there is a high need for the products of machine tools. Thus, F-class vertices are the best choice for a first specialization in the presence of a capitalist society.
C-class vertices are the second specialization. C stands for “chemical”, and C-class verts produce chemicals, polymers, alloys, or other such products. The main production facilities for C-class vertices are refineries and chemical plants. C-class vertices are required to create S-class vertices unassisted (i.e. without monetary participation).
C-class verts are a good second choice, since they greatly advance the available technologies to the network, and provide many essential products that can carry a high cost or dependency on the market. They have a medium bootstrap cost, as there is some open-source information about C-class processes and many of them can use recycled materials (e.g. Filabot). C-class verts could produce molten salt batteries for energy storage.
T-class vertices produce textiles, which are also required for S-class (for PCB stock), and fulfill a great need for the VIAAC network. Though the need and usefulness of textiles is high, their relative market cost is low, hence their being ordered after C-class. T-class vertices could make solar updraft towers for energy production.
S-class vertices are essentially what brings the network to the economic level of a developed country. They are the ICs that produce semiconductor products, such as circuits, motors, batteries, and solar panels. They will make increased automation possible, and the first S-class will probably mark the point where the rate of development of the network achieves a slope greater than one (in other words, development shifts from linear progress to exponential progress). However, semiconductors currently have lots of dependencies and have a huge capital (production capability) cost. Photolithography is a very high-energy process, and silicon for semiconductors requires extremely high purity (nine nines, 99.9999999%), as well as the ability to produce PCB laminates (which requires textiles, copper, and the aforementioned silicon).
M-class communities would create bulk raw materials. Assuming a transitional context, the best choice for initial specialization would be a focus on recycling. These types of communities would be ideally located around mines, quarries, and large concentrations of junk. A junkyard full of cars would be perfect for supporting early development. Engines and wheel bearings can be cut up for use in multi-machines. Frames and timing chains may be useful for CNC mills. Tires are good for building rammed earth walls. An M-class community focused on scavenging would be a good first specialization in the case of a severe collapse.
A-class communities are for afforestation. This is a unique specialization in that it does not require an agricultural basis; it’s possible to start an A-class community based on hunting and gathering. All that’s required is an intermittent transportation corridor to other nodes, which can be nothing more than a terminal. In fact, band societies would probably lend particularly well to this specialization, since they have no special requirements for land use and no need to cut down trees.
Establishing a gift exchange post with existing hunter-gatherer tribes such as the Hadza may be a good way to reach out to them and allow them to help the effort to restore our environment, without forcing them to change their ways. Hunter-gatherer tribes have been increasingly victimized by industrial civilization, and it’s important to respect the idea that they have no desire to assimilate with our way of life. Band societies that exist today have been around longer than agriculture, demonstrating the extreme resilience.
These are not the only possibilities, nor definite ones. Experimentation will be necessary to develop smooth interaction between vertices. Fortunately, the localized community model lends well to experimental methods.
Transferics 