Way back in the days of MSRUS with Sasamisa I came up with several ideas that were this in a nutshell.
My premise was simple: if I owned land, a house, a well, enough solar panels or whatever I would basically have no bills minus property tax which is another matter entirely.
Except I would own no method by which internet access was free or permanent.Obviously unacceptable I began making my own internet to understand the technology enough to replicate/supplant/improve it. Along the way I ran into the Invisible Internet Protocol (I2P) and link layer 2/3 level topological protocols which replaced things like TCP/IP.
A router is nothing more than a computer with just enough (and sometimes not enough) CPU power to run a basic TCP/IP stack and several network interface cards. The NICs have unique identifiers that are layer 2: machine allocation code, MAC addresses. These are universally unique and handed out by a clearing house type of system. (Each manufacturer gets a range, basically.)
The TCP/IP stack is a set of handlers that decode raw data streams into the "packet" concept with which we're familiar. It says that a packet is so long, constructed in such a way and how to handle the data. A protocol is just a method of handling the data, there is nothing more complicated than that at this level.
TCP/IP is a "mature" protocol and with IP v6 the system actually becomes even simpler. I would recommend using this as a basis for standard "web" traffic with UDP or something else for other types of internet traffic.
At this point I need to clarify: the internet is the network of computers itself whereas the web is the inter-connected structure of sites. "Deep web" is a bullshit term used by idiots who will eventually all get caught with illicit material on their computer because they don't know what the fuck they're talking about. If it's connected to this "web" it's part of that web. There are at least seventeen entire webs that operate almost entirely separate of each other with no, only one-way or very few links between them.
Everything else is just disconnected internet
services. A
server is a physical computer that hosts a
service which is provided by a
dæmon (link is to the Wikipedia etymology of the word) and the web is no exception.
So this is the basic list of starting points you'll need to be able to remake the internet/web given a post-apocalyptic scenario where the computers work but the zombies ate the wires. Seriously.
This is where I'm at: economics is supply and demand or put another way the representation of limited resources in values determined by scarcity. If you want a "free as in thought" internet you have it: IP geolocation is very simple to mitigate.
If you want a "free as in beer" internet then you need to make it un-scarce. Abundance past the point of excess and into unlimited. This may never happen as data consumption follows Moore's law and the population of devices will undoubtedly increase over time. Eventually you won't have harddrives you'll just "save" a bit of data to a looping TCP/IP route with a null TTL.
The Forkheads Satellite Network would be a system of server in low Earth orbit (barely above the atmosphere) arranged in such a way as to make the nearest satellite no farther than an acceptable ping time for terrestrial internet connections. Current satellite internet connections make use of satellites in
geostationary orbits around the equator:
35,786 kilometers (22,236 miles) away. Compared to the 160 kilometers (99 miles) of a
low Earth orbit the round trip at the speed of light is obviously shorter (lower ping times).
The power required for such a WiFi based satellite in LEO is much less than the commercial satellite providers utilizing K-band or Kₔ-band primary uplinks; further, the use of technologies like the
Optimized Link State Routing protocol (OLSR) and a WiFi
wireless distribution system would allow simple inter-connectivity for end users of the network and simple connections to the existing internet for primary terrestrial bases.
This is of course in addition to the design of satellites: I advocate solid state components for obvious reasons with the two exceptions being the orientation gyroscope and the possibility of a reactionless drive or ion thruster depending on how far the engineering comes along by deployment time. The main body will need to be composed of a woven carbon fiber chassis framework with enclosing panels of the same material. The avionics cluster will need to have self-resetting overload protection and full electromagnetic shielding.
The server portion will need a multi-path heat dissipation system allowing heat to be radiated into space along whichever path leads away from the sun. The capabilities of the server will need to exceed that of the network by at least the projection of the useful lifespan of the satellite which given our insanity will be enormous. I wouldn't be surprised to see recommendations for yottaflop capable CPUs and with maxed out ZFS drives. This will also need significant electrical/magnetic isolation and it's own internal power source; ZPR.
The communications portion will be a separate portion than the server and have a cluster of "spot beam" type antennas to isolate signals from narrow wedges of the surface of the Earth (or Moon, Mars, whatever) to allow better SNRs and lower transmit amplitudes. This portion will also have lateral antennas to communicate with neighboring satellites in the network allowing routing of data above the atmosphere so the data travels on the satellite network for as long as possible before needing to reach ground.
If the server is on the satellite network then the end user may never have to communicate with a ground station at all. Servers which connect directly to the satellite network allow the end user to connect solely through the satellite network and have their data follow a constantly changing path which still respects client identity: the server will recognize the client even when the satellites change position.
Satellites utilizing WiFi allow every current WiFi device to access the network natively needing only a view of the sky. WiFi frequencies penetrate clouds and such much better than higher frequency traditional satellite bands. WiFi has all the technologies to make seamless transfer of data streams from access point to access point and the OLSR allows the satellites to reroute the data mid-stream without interruption. This would be enough internet for mobile users to enjoy reasonable speeds; fixed-location stations would have a much more capable antenna and so much higher bandwidth without any preferential scheduling by the satellites. A fixed location station might also act as a downlink/uplink/bridge increasing overall network bandwidth.
Data will flow over the network using a modification of the I2P structure where it is encrypted twice, end-to-end, with only destination and entry point IPs plaintext. This ensures that no middle man can access the data, that no middle man may alter the data (without destroying it) and that only the intended dæmon/client on the intended server/terminal.
Access will of course be free to use and free from filtering with reasonable constraints on hosting in the network due to limits of capacity and the portion dedicated to our efforts to retain all data forever.