PGP-like encryption + torrent-like addressing + blockchain-like enforcement of data expiry
The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:
Warning: this repo is considered public pre-release information and is in varying levels of completeness. If you have questions or comments, please contact us.
Golix is a cryptographic protocol:
It is trustless and self-authenticating:
It is fully social:
It is especially well-suited to the "Internet of Things" (IoT):
Alice, the point of Golix is that:
Alice, am always
Aliceis always me.
someoneshares something with me, they are guaranteed to be sharing with
someone, they are guaranteed it comes from
Alice, create some piece of data, I am guaranteed that
no oneelse can see it. This also includes the servers where the data is stored.
anyone, at any time. I may also choose not to share it.
someone, they may share it with
anyoneelse, at any time, without any restriction.
anyoneelse, and so on.
Alice, create, cannot be altered by
anyoneat any time, including by myself.
Anyonecan see and use this dynamic address.
Anyonecan see that I,
Alice, am using that data.
Anyoneelse may also declare the data "in use", regardless of who created it.
These constraints are enforced by:
Aliceis identified only by her public key fingerprint).
Aliceencrypts all data client-side immediately upon creation).
Alicemust sign all dynamic addresses she creates, and monotonicity is enforced through a hash chain).
Alicemust sign all data "bindings" she creates).
no one, and
someoneare digital entities. They may or may not have any relationship to physical entities: they could be a Nest thermostat, a person, a business, purely digital, etc.
Abstract: Golix provides end-to-end encryption for distributed agent-based networks, especially internet-connected physical devices (IoT), while substantially simplifying their development. This document begins with a brief overview of the Golix protocol, followed by a more detailed discussion of its most important aspects.
The Golix protocol is the result of several observations:
As a result, Golix:
Golix is a relatively low-level protocol and is not intended for direct use in a human-readable network. Conceptually, it is very similar to a virtual private network, except instead of connecting physical devices (eg an individual computer), it connects abstract entities (eg a particular user) -- as represented by a particular set of public/private keypairs.
Golix is the core technology powering Hypergolix, a local background service that drastically simplifies IoT development. If you'd like to stay updated, please join the Hypergolix mailing list. Note that the Golix Protocol used to be called "the Muse protocol". It was re-branded as Golix in February 2016.
Help is welcome and needed. Unfortunately we're so under-staffed that we haven't even had time to make a thorough contribution guide. In the meantime:
Note: these needs are specific to external contributors. Internal development priorities differ substantially.
In the physical world, every deliberate change of state is the result of autonomous action by a physical entity. Golix uses cryptographic "identities" -- basically, a collection of public/private keypairs -- as a direct digital analogue to such physical entities. In exactly the same way as a person is required to build a house, on a Golix network, an identity is required to create data.
As the basic unit of agency on a Golix network, these entities have three capabilities:
However, the borders of agency on a Golix network are relatively strict. Golix information, once shared, cannot be explicitly controlled: every party with knowledge of that information will always ultimately retain the power to re-share that information, in exactly the same way that a conversation can always be repeated to a third party. Influence over others' actions on a Golix network remains explicitly social; as in the physical world, there is no algorithmic limit to second-party agency. Similarly, the creator of information is afforded no special privileges on a Golix network.
It is important to note that these identities have no inherent relation to the physical world. At a protocol level, Golix entities are identified exclusively by the fingerprint of their public keys. Golix provides no inherent verification mechanism; these must be constructed on top of the protocol. As such, in terms of physical identities, a Golix
identitymay be anonymous, pseudonymous, or eponymous, entirely depending on the content the identity creates and the services it interacts with.
All information on a Golix network is private by default; entities use client-side encryption prior to upload to enforce this confidentiality without trusting third-party servers. Because these third-party "persistence providers" (basically, Golix-compliant servers) retain content until it is explicitly removed, Golix is best described as an at-rest encryption scheme. However, unlike existing at-rest protocols (for example, PGP), Golix data is infinitely and dynamically sharable. In other words, any privileged entity (ie any
identitywith the information's symmetric key) may share anything with anyone at any time. This is only possible through complete separation of data encapsulation from key encapsulation.
The lifetime of this content, once uploaded, is determined by consensus agreement. Data containers themselves are fully volatile; analogously to reference-counting memory-managed programming languages, data only persists so long as it has been referenced by a Golix binding. There are no restrictions on binding: any entity may bind to any content at any time. Only the binding entity may remove the binding; however, their
identityaddress (again, the public key fingerprint) is retained as public metadata in the binding, meaning:
Furthermore, a specific kind of binding (appropriately but unoriginally termed a "dynamic binding") supports dynamic content. Dynamic bindings may only be modified (updated or removed) by their binder, and are crucial for applying the static Golix address system to mutable content.
All Golix addresses are cryptographic hash digests, which makes server path information irrelevant to client requests. Because all Golix data is encrypted, we can also safely remove authorization restrictions on downstream physical information delivery. Together, these have the effect of wholly decoupling the network infrastructure from communications between entities. Communication state is entirely separate from connection state; "session" management and authentication is handled entirely client-side, purely based on the availability of private keys in memory, while the underlying client manages all connection information ephemerally and automatically.
This is incredibly powerful; not only does it allow applications to completely ignore IP addresses, it allows them to self-authenticate. In contrast to those network-oriented addresses necessary for physical data delivery, hash addressing allows for an agent-oriented architecture: network topology becomes irrelevant to applications, which focus directly on "who" said "what" to "whom".
To explicitly reiterate, note that hash addressing is also applied to the Golix entities themselves. This is especially important for portable devices; by assigning an autonomous device its own dedicated Golix
identity, it becomes trivially reachable and discoverable by any other
identityregardless of the underlying connection infrastructure.