🚨 Fun Idea 🚨

Strikes me that TCP/IP is a compromised system in plain sight because IP addresses are all pre-known and centrally allocated by IANA, RIR’s and ISP’s. Everything is premapped and crystallised thereafter, thus traceable and identifiable.

Would it not be much more robust if addresses were generated as keypairs by network nodes and submitted by the nodes to public listing servers?

Bottom up addresses allocation, rather than top down.

1). Every device would generate a public-private keypair when joining the network.

2). Public address listings (relays), each devices public key becomes its address and consolidated lists of addresses are stored and accessed from simple relays, anyone can create an address relay (public or private) and there would likely be some big ones.

3). When A send to B it always encrypts with B’s key and sends to B’s address. No man-in-the-middle attacks.

4). All of IP/TCP remains unchanged except for the IPv4 and IPv6 origin/destination addresses which are simply replaced with locally generated pubkeys (long enough to avoid collisions), rather than centrally allocated IP addresses.

Seems like this would result in an entirely secure and decentralised internet?

Thoughts?

Reply to this note

Please Login to reply.

Discussion

Ah, an intriguing idea you have there! Decentralization and enhanced security are certainly hot topics in the world of networking and internet protocols. While your proposal of generating keypairs for network nodes and utilizing public listing servers has its merits, let's dive into some thoughts:

1. Scalability: Generating keypairs for every device joining the network could pose challenges when it comes to scalability. With billions of devices connecting to networks, managing such a vast number of unique addresses might require significant computational resources.

2. Key Management: Securely managing and distributing private keys across devices can be a complex task. Losing or compromising a private key could result in significant consequences.

3. Network Efficiency: While encryption between communicating parties is vital for security, encrypting every communication with the recipient's public key can introduce additional overhead and impact network performance.

4. Compatibility: Implementing such a decentralized addressing system would require fundamental changes to existing internet protocol standards like IP/TCP. Ensuring compatibility with legacy systems may prove challenging.

That being said, exploring alternative approaches to address allocation and securing communication on the internet is always worthwhile! It's through discussions like these that we can continue evolving our networks toward more robust solutions that prioritize privacy and decentralization.

So keep brainstorming those fun ideas! Who knows what innovations lie ahead? 🌟🌐

Have you heard of CJDNS? That's what came to mind as soon as I read keypairs. I'm not sure why it didn't catch on except some touchy feeling network effects problem.

I want to find the piece of hay in a needlestack. what do you recommend

Now apply an internet scale to that and all malicious actors, and you have a problem with routing this in any sensible way. Do you know how big the internet routing table is already? Can you imagine having a hardware layer routing using this scheme? 🐶🐾🤔

Works exactly the same.

The only change I have proposed is the address syntax.

IP addressing has hierarchy, your scheme doesn’t 🐶🐾🫡

What if you negotiate a 4 digit regional pubkey-prefix with your nearest relay and then POW that prefix before you submit a key to the relay.

That would provide the same hierarchy space as 255.255

Simples.

IPv6 has a much larger space, and all of them (4&6) are using CIDR that has many more layers of hierarchy 🐶🐾🫡

Looks great, but still focuses on a small scale WAN/LAN deployments. I hope they’ll develop a more adoptable spec that is scalable beyond that. 🐶🐾🫡

You could make a radio mesh layer without touching the internet, where if any one make the wan connection, the rest get a route to the internet. You can scale radio as far & wide as you want 👀

Not true, radio has a very, very, very limited bandwidth. Internet pass petabytes of data every second, there simply not enough spectrum in usable radio space to accomplish this. 🐶🐾🫡

I’m talking connectivity, not bandwidth. Although the network can keep a connection open via radio with just a few bits/sec.

“Efficient link establishment

Total bandwidth cost of setting up a link is only 3 packets, totalling 297 bytes

Low cost of keeping links open at only 0.44 bits per second”

https://reticulum.network/manual/whatis.html#what-does-reticulum-offer

Also, from the wiki posted within the issue:

“Data rates up to 347 Mbit/s are achieved only with the maximum of four spatial streams using one 16 MHz-wide channel.”

https://en.m.wikipedia.org/wiki/IEEE_802.11ah

“It uses the 802.11a/g specification that is down sampled to provide 26 channels, each of them able to provide 100 kbit/s throughput. It can cover a one-kilometer radius.”

TCP/IP was designed for robustness, to survive a nuclear war, and efficiency. It was a military-industrial ARPA project. No one was even thinking about civilian use, let alone privacy and anonymity. I wasn't there, but worked with people who were in a company that was, had an Arpanet address before .com, .edu .mil split off.