Comment on The Sticker Network

BenRayfield Tue, Aug 24, 2010
Each Sticker should be a kind of Lisp Machine because that is the best way to avoid scope creep. There is no electronic device, except a watch or a timer etc, that is simpler than a Lisp machine. Your cell phone is much more complex. It is impractical to put up new stickers each time the network routing behaviors need to change, and they will need to change because there will always be hackers and we will need to find ways to make it harder to hack.

You say most people do not know Lisp. Thats ok, because most people do not know how to create software drivers for their network routers either. Most people will simply buy a USB stick for their computer and it will just work. They don't need to know Lisp. But I need it to run Lisp because the CPU and memory parts will be much cheaper and simpler that way.

There are well known ways to send secret messages in public, so adding encryption into the hardware is not necessary. If we added encryption to the hardware, for example at 128 bits, then those who are against the network (like Internet2 which is against freedom on the internet so people will have to pay monopoly prices to use their new infrastructure) could pay to create a new law that makes the maximum legal encryption be 127 bits (or maybe 64 if they like powers of 2). Any way we go about this, theres a lot of businesses and politicians that will put lots of effort and resources into making us fail, so they can make more money exploiting people. They just haven't noticed it yet. If encryption is in the hardware, I would leave the project and start my own project without encryption. Encryption can be added on top of public communication, as it is done in the existing Internet.

If somebody wants to "sniff traffic", there will be no way for them to find specific traffic, because nobody and no sticker and no computer will know what path any specific packet will go unless it chooses to use a very simple type of address.

The variable size addresses I proposed are to keep it simple. The existing Internet has lots of different ways of sending sequences of bits because they started with inflexible requirements. They are stuck on IP4 even though IP6 has been available for over 10 years and would solve many problems. I will only help with a project that is flexible enough to avoid all existing problems in the Internet.

If you think stickers are not the best way to do this, you're in the wrong thread. "Sticker" is in the title. As I wrote above, there should be more than 1 kind of device. You need a more complex device to connect to your computer to "bridge" The Sticker Network and the existing Internet. We agree on that.

Where to put the Stickers? How about on your car? On the back of street signs (It is my legal position that we have the Constitutional right to do this in USA, but it will have to be proven in court), on the side of businesses which you get permission from, in trash cans, tied on a string and hung on the branches of trees, or lots of other places. Or maybe we could invent some kind of gun that shoots them into the side of trees from far away, and shoot them as we're driving along the highway. Not all kinds of "drive by shootings" are bad.

The designs I proposed are to keep it simple, cheap, and flexible. If we copy what is normally done, it will get complex really fast. What is normally done is usually to serve those in power instead of because it works or is efficient. Keep it simple and cheap and a turing-complete amount of flexible or I'll find somebody else to do it that way with me.

Most importantly, The Sticker Network is about an internet that has no central authority (peer to peer), and putting encryption in the hardware creates a way for a central authority to control it. If you can control it through your encryption codes, you are that central authority, and as normally happens when anyone has a little authority, the government threatens them indirectly and takes some of that authority, resulting in a government controlled Sticker Network.

If we build it as I proposed, with the variable-size addresses, then such addresses would support simple numbers (like IP4) and things like including an encryption algorithm (written in Lambda Calculus) in the address, because addresses can be simple numbers or numbers that represent a string of Lisp code and some numbers for it to operate on. I proposed the simplest possible design that is flexible enough to do everything you said and everything we will ever need to do. The addresses are Turing Complete, but the data is whatever the computers (connected through USB sticks) send through the network. It can put time limits on running the code in the addresses. With that much flexibility (which costs us very little since its just Lambda Calculus), every internet protocol that exists can be implemented in an address. The difference between what me and you proposed is no Sticker would be forced to accept any algorithm or packet, so you could only get them to do encryption if large parts of The Sticker Network wanted to accept your algorithm. Its a decentralized democracy that way.