We’ve had enough of digital monopolies and surveillance capitalism. We want an alternative world that works for everyone, just like the original intention of the web and net.

We seek a world of open platforms and protocols with real choices of applications and services for people. We care about privacy, transparency and autonomy. Our tools and organisations should fundamentally be accountable and resilient.


Adam Ierymenko [LibreList] Re: [redecentralize] Thoughts on decentralization: "I want to believe." 2014-08-05 11:57:52 (6 years 8 mons 19:37:00 ago)
Oh definitely.

Homomorphic crypto could have a *lot* of uses. It opens up the potential for things like black box certificate authorities that could be distributed as open source software. The CA signs your key. With what? A key pair it generated internally that cannot *ever* be viewed by *anyone*. :)


On Aug 4, 2014, at 9:48 PM, Eric Mill <eric@konklone.com> wrote:

One line of research and technology that I personally find very exciting, and highly relevant to the idea of zero-knowledge centralization -- even though it's still some time off from being scalably useful -- is homomorphic encryption.

Homomorphic encryption is a technique where you take two inputs, encrypt them with a private key, hand them off to some other machine, have that machine perform a known computation *on the ciphertext*, and give you back the encrypted result, so you can decrypt it and get the answer. The machine that did the computation knows nothing about the inputs or the outputs -- it can only blindly operate on them.

While some techniques (like RSA) were partially homomorphic, what you need to make arbitrary homomorphic computation is a system that can do both multiplication and addition (together, these are Turing complete), and no system to do this was found for 40 years, until Craig Gentry's PhD thesis showed a working algorithm to do it.

The bad news it is many many orders of magnitude too slow to be useful -- and uses "lattice encryption", which requires very large private/public keys (like GBs). IBM has since scooped up Gentry, and made advances on the original scheme that have sped it up by a trillion times -- but it is still a trillion times too slow.

But, someday -- and maybe someday sooner than we think, as these things go -- maybe it will be feasible to have things like zero-knowledge search engines. Maybe low-level zero-knowledge tasks, like packet-switching or whatever, could be feasible much sooner.

It's something to watch!

-- Eric

On Mon, Aug 4, 2014 at 7:06 PM, Adam Ierymenko <adam.ierymenko@zerotier.com> wrote:
Not exactly, but close. CJDNS is a mesh protocol that creates a single L3 IPv6 network. ZeroTier One is a hybrid peer to peer protocol that creates virtual Ethernet networks (plural). ZeroTier is more like SDN for everyone, everywhere. (SDN is software defined networking, and refers to the creation of software defined virtual networks in data centers.)

I've been following CJDNS for a while. I know it's being used by several community meshnet projects. Anyone tried it? I admit I haven't yet, but I've heard it basically does work but not perfectly. I'm curious about how large it could scale though. I'll try it out at some point.

On Aug 3, 2014, at 5:21 AM, Jörg F. Wittenberger <Joerg.Wittenberge r@softeyes.net> wrote:

Sorry, this was supposed to be a private message.
(But hitting "reply" instead of "reply to list" sends it to the list anyways.)

One more question: am I correct to understand that zerotier serves essentially the same purpose as cjdns?



Am 03.08.2014 11:31, schrieb "Jörg F. Wittenberger":

I've got a question:
In this blog post you wrote:

> I designed the protocol to be capable of evolving toward a more decentralized design in the future without disrupting existing users, but that's where it stands today.