We take the Internet for granted, not realizing that such a global, decentralized system is a rare thing. Protocols, rightly, get credit, but they alone are insufficient. TCP/IP did not create the Internet. The Internet is not just a set of protocols, but rather a real thing. People and organizations created the Internet by hooking real hardware and communication lines together. To understand the importance of this, we need to understand what's necessary to create social systems like the Internet.
Social systems that are enduring, scalable, and generative require coherence among participants. Coherence allows us to manage complexity. Coherence is necessary for any group of people to cooperate. The coherence necessary to create the Internet came in part from standards, but more from the actions of people who created organizations, established those standards, ran services, and set up exchange points.
Coherence enables a group of people to operate with one mind about some set of ideas, processes, and outcomes. We only know of a few ways of creating coherence in social systems: tribes, institutions, markets, and networks1. Startups, for example, work as tribes. When there's only a small set of people, a strong leader can clearly communicate ideas and put incentives—and disincentives—in place. Coherence is the result. As companies grow, they become institutions that rely on rules and bureasucracy to establish coherence. While a strong leader is important, institutions are more about the organization than the personalities involved. Tribes and institutions are centralized--someone or some organization is making it all happen. More to the point, institutions rely on hierarchy to achive coherence.
Markets are decentralized—specifically they are heterarchical rather than hierarchical. A set of rules, perhaps evolved over time through countless interactions, govern interactions and market participants are incented by market forces driven by economic opportunity to abide by the rules. Competition among private interests (hopefully behaving fairly and freely) allows multiple players with their own agendas to process complex transactions around a diverse set of interests. Markets don't displace institutions completely but do help institutions process transactional exchanges.
Networks are also decentralized. The rules of interaction are set in protocol. But, as I said earlier, protocol alone is not enough. Defining a protocol doesn't string cable or set up routers. There's something more to it. One form of organization doesn't usually supplant the previous, but augments it. The Internet is the result of a mix of institutional, market-driven, and network-enabled forces. The Internet has endured and functions because these forces, whether by design or luck, are sufficient to create the coherence necessary to turn the idea of a global, public decentralized communications system into a real network that routes packets from place to place.
My interest in coherence is driven by efforts to create a global, public, decentralized identity system that we call Sovrin. Sovrin is an identity network that, like the Internet, achives coherence through a combination of institutional, market-driven, and network-enabled forces. I think of Sovrin as the Internet for Identity. The Sovrin network is not owned by anyone, but, like the Internet, is created through the cooperation of many people and organizations. Sovrin is seeking to create a global, decentralized, identity network that is open to all. That's a little bit different thing that creating a standard or a protocol.
When I say Sovrin is "public" I mean that we intend for it to be a public good that anyone can use so long as they adhere to the protocols, just like the Internet. Public spaces require coherence. Coherence in Sovrin springs from the ledger, the protocols, the trust framework, standards, and market incentives. These gives Sovrin greater coherence than a protocol alone can achieve.
The ledger creates a decentralized service where various participants can exchange important information without the need for a central authority. But like the Internet, the ledger requires multiple organizations to commit real resources. The Sovrin ledger is the network. The ledger provides a global registry for identifiers. This has several significant advantages over DNS as a registry. DNS-based identifiers aren't permanently owned. This can present real security concerns. With a ledger, identifiers can be persistent and non-reusable. Second, DNS-based solutions require that every site that wants to run registry functions has to install and maintain software for that purpose. This incents centralization of the directories. With a ledger-based solution, a global network does this without special effort on the part of participants.
Standards like those being developed for Decentralized Identifier (DID) and the verifiable credentials are important for interoperability, but more so specifications and standards on how to communicate with the ledger and the peer-to-peer communication of agents are critical to portability and the rise of a functional ecosystem of software that isn't controlled by a single vendor.
Market incentives have traditionally been difficult to incorporate into a networked ecosystem, but the rise of tokens is a recent development that promises to change this. Token-based ledgers come with incentives to participate and thus are self-sustaining in ways that DNS-based identifier registries can't be. This also alleviates power imbalances where large companies have an advantage relative to individuals. As Chris Dixon says "Token networks ...[align] network participants to work together toward a common goal—the growth of the network and the appreciation of the token." In short, tokens provide a means for establishing coherence.
Someone asked me recently how Sovrin differed from an identity protocol like OpenID Connect. The factors that give rise to coherence are the primary difference between a protocol like OpenID Connect and something like Sovrin. Simply put, Sovrin is a network and OpenID Connect is not. This is where I think we've largely gone wrong in building decentralized systems in the past. We've shied away from putting a stake in the ground and building a thing. Instead we've defined protocols and allowed the network effects to accrue to centralized companies who can establish the coherence necessary to create a real thing. My strong belief is that real networks are necessary for enduring decentralized systems.
In Decentralized Governance in Sovrin, I described how the Sovrin Network is governed. The centerpiece of that discussion is the Sovrin Trust Framework. The trust framework serves as the constitution for Sovrin, laying out the principles upon which Sovrin is governed and the specific requirements for various players in the Sovrin Ecosystem.
In A Universal Trust Framework, I say “a trust framework provides the structure necessary to leap between the known and unknown.” The idea is that online we often lack the necessary context to reduce the risk around the decisions we make. A trust framework defines that context using agreement, process, and technology so that people can make decisions with significantly less risk.
In this post, I want to talk about how organizations can use the Sovrin Network as a trustworthy infrastructure for creating businesses or processes that require strangers to trust one another. This happens by building domain-specific trust frameworks on top of Sovrin.
Exchanging Claims and Proofs
Let’s begin by reviewing Sovrin-based credentials and how they are used.
Issuers issue and identity owners accept Sovrin credentials that are based on emerging Verifiable Claims standards. Identity owners store these credentials in their digital wallet. A credential consists of a set of claims. Identity owners present proofs based on those credentials to verifiers. Proofs can be about any assertions in the credentials the identity owner holds.
The preceding figure shows how this might work in a specific use case. Suppose your employer is using Sovrin. They issue a Sovrin credential that includes claims about your employment status and current salary. The credential includes a reference to a credential definition (on the Sovrin ledger) that contains your employer’s public decentralized identifier, or DID, (also on the Sovrin ledger) and a reference to a schema (also on the Sovrin ledger) that describes the claims in the credential. The credential is signed by your employer, using their public DID. You store that credential in your Sovrin wallet.
Later, you apply to your bank for a loan. Before the bank processes the loan, you need to prove to them that you’re employed and that you make at least $75,000 per year. Using your Sovrin wallet1, you are able to prove what the bank needs to know without revealing your actual salary or any other claims that might be in the credential your employer provided. The proof process also ensures, through countersigning, that the bank sees these proofs based on their DID for you even though the credential was issued to you by your employer based on the DID they have for you.2
Of course, your wallet also holds other credentials from other issuers. For example, you might have a credential from your bank that you can use to prove your account information to your employer or any other party.
The beauty of this approach is that verifiers never need to contact issuers to verify the assertions or understand how they’re formatted. Anything they need to use to verify the credential is on the ledger. This results in immense benefits for scalability and reliability3.
What Sovrin Promises
The preceding example credential exchange shows how the credential-supported proof provides trustworthy attributes to the bank. But it’s worth teasing that apart and determining why the attributes can be trusted. Let’s start with what Sovrin is promising with respect to the credential that is being issued and the proof that is based on it.
Sovrin’s promise is based on the Sovrin Trust Framework. As I pointed out in Decentralized Governance in Sovrin, the framework governs the operation of the network and the code that it runs on.
Sovrin promises the following:
The DIDs can be resolved using the Sovrin ledger and the resulting DID document that can contain a public key and an endpoint associated with that DID.
The schema can be retrieved from the Sovrin ledger and hasn’t been tampered with.
The definition for the credential can can retrieved from the ledger and hasn’t been tampered with.
The credential can be validated as not having been tampered with.
The credential was given to the identity owner with whom the employer has a relationship.
The proof is about the identity owner with whom the bank has a relationship.
The identity owner who presents the proof is the same person to whom the credential was issued4.
The credential has not been revoked by the issuer.
The Sovrin Trust Framework is a general-purpose or universal trust framework that is meant to provide a set of universal features that form the foundation for trust transactions. The preceding list of Sovrin promises are all part of that trust framework. They are shored up by legal documents, business processes, and technology.
What Sovrin Does Not Promise
The promises made by Sovrin are important and foundational. But what the bank really wants to know is “are you employed?” That requires promises that Sovrin can’t make:
The identity owner is a real person.
The employer is a real business.
The bank is a real bank.
The identity owner is employed by the employer.
The salary figure comes from the employer.
If Sovrin doesn’t keep its promises, then the bank can’t trust anything. But even if Sovrin runs perfectly, the bank is still short of the assurances it needs to make decisions. That’s where domain-specific trust frameworks come in.
Domain-Specific Trust Frameworks
A domain-specific trust framework works on top of the Sovrin trust framework and contains the technology, business processes, and legal agreements necessary to trust the content of the claim. For example, in the scenario above, a domain-specific trust framework would enable the bank to trust the proof that you’re employed.
Let’s walk through each of the promises Sovrin can’t make and discuss who can make them and how.
You Are a Real Person
The bank knows you’re a real person because they have a relationship with you, memorialized by the exchange of Sovrin DIDs. As part of their new customer process, they performed a Know Your Customer (KYC) process as mandated by law. They associate the results of that with the DID you gave them to represent you. So when you contact them with that DID, they know the account belongs to a real person.
Your Employer is a Real Business
Knowing the employer is a real business is a little harder. Even if the bank knows the employer by name, how do they know that the credential was issued from that entity. As I pointed out in Sovrin Web of Trust, there are several options:
The employer could publish their public DID (the one associated with the claim definition that underlies the employment verification) on their Web site or some other well known place so that it is easy to verify. This method takes you out of the Sovrin system and into the hierarchical public key infrastructure since you’d verify the website using a standard SSL certificate from a traditional certificate authority.
They could offer supporting claims that show they are a legitimate business. These claims would come from what are called trust anchors in Sovrin, trusted entities who can prove who they are based on others who vouch for them. For example, the government, acting as a trust anchor, could provide claims to registered businesses. The Government of British Columbia is piloting this right now.
The preceding two steps could be used together in a hybrid verification. InfoCert, that largest European cerificate authority, is already a Sovrin Steward and could issue a Sovrin credential to go along with the TLS certificate the issue to the employer. This Sovrin credential would be backed up by the same process they use to verify the business as part of issuing the TLS certificate.
Businesses often have credentials from others that are part of their web of trust—think of all the “Chamber of Commerce” decals you see on the windows of local shops. The employer would have credentials from numerous government agencies, business partners, and trade associations that could all be used to back up the credentials they issue.
The Bank is a Real Bank
The person knows the bank is a real bank for the same reason the bank knows the person is real: the DID exchange. One of the important and often overlooked benefits of a DID exchange is that both sides can authenticate the other. Mutually authenticated connections are the default in Sovrin and one of its most powerful features. This is in contrast to TLS, which has been a one-sided affair in practice.
In our example, the bank is verifying the credential, so it obviously knows that it’s a real bank, but the identity owner want to know it's real when they open an account. If they did it physically, then they know of themselves, but if they enrolled online, they might have questions. The bank could be validated in the same way as the employer. For example, in the US, financial institutions are certified by different government institutions depending on whether they’re national banks, state-chartered banks, or credit unions. In any case, these certifying agencies could issue credentials to the institution that could be checked by the customer (using their wallet) as part of the enrollment process or at any other time.
You are Employed by the Employer and Have a Certain Salary
The credential itself is making the claim that you are employed by the employer and that the salary is a certain amount. Sovrin links all this up, but the bank would need to validate the claim schema and definition and understand what it means. This validation process would be internal to the bank.
In Sovrin, the schema is a document that is written to the ledger. Think of it as the description of what fields the schema contains and what the fields mean. The schema is readable and understandable by anyone—they are public documents.
Schemas can be reused. Ideally the schema is not specific to the employer, but is used by many employers for employment verification. An industry trade group or a large vendor of ERP systems might, for example, author the schema with inputs from many parties. Doing so would ensure that a wide range of organizations understand and recognize it. The schema becomes a standard, de facto or de jure, for employment verification.
I believe that there will be a relatively rapid agreement on schemas, reinforced by trust frameworks, in many fields. But this does not preclude anyone from creating any schema they want whenever they want for their own purposes.
Claim schema are automatically versioned. When a new schema is created, the original schema doesn't go away, so credentials issued with the old schema can still be used. Credentials issued with the new schema refer to that new schema and not the old one.
The claim definition references a claim schema and is written to the ledger by a specific entity. In our example, the employer writes a claim definition that references the schema and includes its public DID.
Depending on the use case, an industry group might do more than merely publish the schema. They might also create standards around how the schema is to be used. These standards would be enforced by legal contract and required business processes with compliance attested to by claims from the standards group to any organization creating claim definitions that use the schema. That way claim inspectors could know that the claim was created in accordance with an accepted process, increasing trust in the claim. How would they know? By means of a Sovrin credential from the standards group, of course.
Sovrin’s Role in Subject Specific Trust Frameworks
A domain-specific trust framework is a collection of policies, legal agreements and technologies that provides the context for claims in a given domain. Sovrin Foundation provides a structure and supporting systems for groups defining trust frameworks.
Sovrin works as a globally decentralized dictionary for schema definitions, with no central authority dictating what schemas, definitions, and claims can or can't be issued. This makes it uniquely flexible. Anyone can issue credentials about anything, enabling the "long tail" of use cases to be easily addressed.
Sovrin Foundation provides the basis for the Sovrin web of trust. As discussed in the Sovrin Trust Framework, Sovrin Trustees provide the foundation for this web of trust. They select Stewards who validate transactions on the Sovrin ledger5. Stewards can vouch for others. Anyone using the Sovrin network is part of this general web of trust.
Beyond this general web of trust, domain-specific trust frameworks can determine their own web of trust. This already happens in the physical world. For example, colleges and universities in the US are accredited by independent organizations. Brigham Young University, where I work, is accredited by the Northwest Commission on Colleges and Universities. They already have a process for accrediting member organizations. They could simply rely on that process and any attendant legal agreements to issue and revoke Sovrin credentials that reflect their determinations concerning the accredidation status of member institutions.
Other uses of Sovrin may not have a physical world counterpart. For example, suppose you wanted to start a decentralized version of AirBnB. One of the roles that AirBnB plays is identifying and attesting to certain attributes for renters and hosts. A decentralized AirBnB may want to establish a process for vetting renters and hosts and issue and revoke claims in accordance with that process. This decentralized AirBnB might define and include claim insurance for both renters and hosts in the event the process breaks down. The processes that define how vetting works, the legal agreements that hold parties to the processes, and the provisions for recourse and restitution via insurance are all part of the domain-specific trust framework for this ecosystem.
For anyone wishing to build domain-specific trust frameworks for an existing or planned business, Sovrin Foundation can help with technical advice about schema and claim definitions as well advice on how to go about creating and governing the legal agreements and business processes necessary to create environments where trust abounds.
A wallet in Sovrin is an agent that holds credentials on behalf of the identity owner and speaks the Sovrin agent-to-agent protocol.
Your employer and bank have a different DID for you to prevent correlation.
Specifically, issuers do not need to maintain always-on infrastructure to answer questions about credential validity and credential transactions are done peer-to-peer.
Technically it could also be presented by the identity owner's guardian if the identity owner is in a guardian relationship. But the Trust Framework considers that to be the same as the identity owner presenting it themself
When Stewards "validate" a transaction, they are merely achieving consensus with other Stewards about the validity of the transaction, not it's content. Stewards aren't making judgments about the assertions inside of any transaction on Sovrin.
Late last year CryptoKitties burst into the blockchain world. If you haven't been paying attention, CryptoKitties is a Web site that uses a browser-based wallet (MetaMask) to sell (for Ether) little virtual kitties. Once you have a kittie, you can breed it with others, to create new kitties. Each one is a unique individual created with some genetic algorithm. Some Gen 0 or Gen 1 kitties have sold for ridiculous amounts of money. If you were around in the 90's when the Web was taking off, think Beanie Babies meets Blockchain and you'll get the idea1.
Each kittie is disiguishable from all the rest and has a unique identity and existence
Each kittie is owned by someone. Specifically, it is controlled by whoever has the private keys associated with the address that the kittie is tied to.
Because of these properties, a CryptoKitty can be remixed and used in other applications. Just today, I heard that they are being sold on OpenBazaar for Bitcoin. Someone could create a game that uses the kitties that were part of the CryptoKitty game as characters in a new game. Because they are not just a line in some centralized company's database, they really do belong to the people who bought them and who can use them for anything they like.
Why is this important? You can go on Amazon and "buy" a digital copy of a movie or song. But do you really own it? Do you have a unique copy that is controlled only by you? Not likely. In the wake of Louis C.K's sexual miscoduct allegations, many services cut their ties to the comedian. I'm not defending his actions in any way, but people who "bought" that content didn't own it, because someone else's decisions took the content away from them.
That can't happen with a CryptoKittie (almost, keep reading). The token representing the kittie is mine and under my control. That's the idea of self-sovereignty. You control something without others being able to take the asset away from you. As I wrote in On Sovereignty:
Sovereignty is about relationships and boundaries. When we say a nation is sovereign, we mean that it can act as a peer to other sovereign states, not that it can do whatever it wants. Sovereignty defines a boundary, within which the sovereign has complete control and outside of which the sovereign relates to others within established rules and norms.
The border, in this case, is defined by Ethereum, the ERC-721 specification, the smart contract, and my private keys in my wallet. The only thing that can take my CryptoKitty away from me is Ethereum going away. They are a non-fungible asset like art and other collectible. They can be traded and they can be repurposed, but only with my consent.
The Fly in the Ointment
What I wrote above is almost true. The structure of individual ownership is all there2. There is one problem with CryptoKitties as a model of self-sovereignty: the CryptoKittie smart contract has a "pause" function that can be executed by certain addresses. This is probably a protection against bugs—no one wants to be the next theDAO—but it does provide someone besides the owner with a way to shut it all down.
I have no idea who that someone is and can't hold them responsible for their behavior—I'd guess it's someone connected with CryptoKitties. Whoever has control of these special addresses could shutdown the entire enterprise. I do not believe, based on the contract structure, that they could take away individual kitties; it's an all or nothing proposition. Since they charged money for setting this up, there's likely some contract law that could be used for recourse. Nevertheless, we need more discussion of formalized governance in decentralized systems. How do we balance the need for security (the kill switch) with the rights of the people who own kitties?
Next time you see me in person, I've got a funny story about Beanie Babies and iMall.
Addresses that pay for kitties are linked to those kitties in the smart contract. My understanding is that if the smart contract that controls CryptoKitties were updated, the old contract is still there and would control the kitties created by it. Because the contract is public, owners can always interact with their tokens.
The Sovrin Foundation is excited to announce that we have hired of Nathan George as our Chief Technology Officer. Nathan was previously Chief Architect at Evernym, Inc. He has been instrumental in maintaining the Hyperledger open-source Project Indy, which is sponsored by the Sovrin Foundation. Nathan comes with a wealth of experience that will help Sovrin thrive and reach its full potential.
I’m very excited to have Nathan join the foundation. The Sovrin Foundation is much more than an advocacy organization for self-sovereign identity. As I wrote in Decentralized Governance in the Sovrin Foundation, the foundation exists to administer the Sovrin Trust Framework and a significant aspect of that entails designing and implementing protocols, managing Project Indy, and supporting the Sovrin Stewards in their operation of the network nodes. These tasks are technical and having a full-time technology executive focused on this is critical to the success of the network, and the foundation itself.
The Sovrin Foundation technology goals, on which Nathan will have primary focus, include:
Planning and following a roadmap that outlines Sovrin development
Implementing a Sovrin RFC and Improvement Project process
Supporting steward onboarding and operation
Managing the release process that prepares code for Stewards
Monitoring network operation and reliability
Impacting design of the various components of the Sovrin ecosystem
Supporting feature development through bounties and other developer incentive programs
Participating in standards processes that impact Sovrin
Hiring a CTO represents a big step along the path to self sustainability for the Sovrin network. A huge part of being self sustaining is technical independence. Having a CTO in the Sovrin Foundation will allow us to drive many technical developments and coordinate our open source projects from within the foundation.
Marc Hulty defines governance as "the processes of interaction and decision-making among the actors involved in a collective problem that lead to the creation, reinforcement, or reproduction of social norms and institutions." From this we can conclude that everything gets governed, the question is whether governance is ad hoc or formal, explicit or implicit.
One of the ironies of decentralized systems is that they require better governance than most centralized systems. Centralized systems are often governed in an ad hoc way because the central point of control can easily tell all participants what to do. Decentralized systems, on the other hand, must coordinate across multiple parties, all acting independently in their own self-interest. This means that the rules of engagement and interaction must be spelled out and agreed to ahead of time, with incentives, disincentives, consequences, processes, and procedures made clear.
The Internet is an excellent example of this. All the computers that connect to the Internet, along with those that route messages between nodes, follow a set of procedures determined ahead of time that cannot be ignored by participants without loss of functionality. These procedures are called protocols and they are embodied in the code that participants in the network—both edge nodes and routers—execute.
Blockchain systems are no different. Their interactions are wholly or partially governed by protocols embodied in code. But these protocols are not emergent properties of the blockchain, rather they are designed by humans who write the code and agreed to by the humans who execute it. Without the design and agreement of humans, the blockchain does not function.
The Sovrin blockchain is a good example. There are two primary components of how Sovrin operates: the design of the protocols and the operation of the ledger. We can ask governance questions about each.
Sovrin's code is open source—specifically it is the Hyperledger Indy project under the umbrella of the Linux Foundation. This means developers from around the world collaborate to design and build the code that makes Sovrin work. Their decisions are governed in the way of most open source projects: rough consensus and running code with pull requests accepted by a core set of developers who manage the code base. Code embodies the rules that make up the Sovrin protocol. Since the protocol is a critical component of Sovrin governance, how decisions are made about the code is a key component of ensuring Sovrin is governed well.
But with the Sovrin blockchain there is another kind of governance that is not embodied in the code. Sovrin is built on a public permissioned blockchain. This means that anyone can use the Sovrin ledger, but consensus is achieved by a known set of validator nodes. Validators perform a similar function on a permissioned ledger than miners perform on a permissionless ledger—ensuring that each copy of the ledger records the same data in the same order1. Since Sovrin’s validators are part of a known group, we should also examine how this group of validators is selected, organized, and governed.
Decisions about how code is architected, which code is run by validator nodes, and how those nodes operate are made in accordance with the Sovrin Trust Framework, a document that specifies how the Sovrin Network is governed. The Trust Framework is Sovrin’s constitution, defining the operation and organization of the network, and is where the final authority for the Sovrin Network lies. In fact, Sovrin can be said to have a constitutional governance model. The Sovrin Foundation is an international, non-profit organization that represents the collective interests of all Sovrin identity owners. Sovrin Foundation exists to administer the Trust Framework. The Sovrin Foundation clearly and openly identifies the actors in decisions and the processes they use to reach those decisions and ensures they are made consistent with the trust framework.
In the constructive sense of the word, governance can be based only on clarity of shared intent and trust in expected behavior, heavily seasoned with common sense, tolerance, and caring for others as fellow human beings. This is not to say that contracts, laws, and regulations do not serve a purpose. Rather it is to point out that they can never achieve the mechanistic certainty and control we crave. Rules and regulations, laws and contracts, can never replace clarity of shared purpose and clear, deeply held principles about conduct in pursuit of that purpose.
In the case of Sovrin, that shared purpose and deeply held principles are contained in the Sovrin Trust Framework. The Trust Framework lays out the purpose of Sovrin as providing a global public utility for self-sovereign identity that adheres to a set of thirteen principles2:
Independence and Self-Sovereignty
Web of Trust
Security by Design
Privacy by Design
Identity for All
Collective Best Interest
Sovrin embraced these principles to meet the needs of identity owners and protect their data, privacy, and autonomy. These principles guide both the technical development of the Sovrin code and the operational rules for stewards, the trusted organizations who operate validator nodes and connect to the ledger in a transparent way.
Some have pointed to the permissioned structure and the existence of the foundation as a reason for distrusting whether it is truly a decentralized public network. To those people I say, “Look at the Trust Framework and tell me where it falls short.” Often this question is just a knee-jerk reaction to how Sovrin splits governance between code and agreement. There’s nothing magic about code when it comes to governance. It can automate things but it has no special properties when it comes to resisting accumulation of power or autocratic tendencies. Even in permissionless blockchains, change is proposed by the developers and accepted (or not) by miners. These are people coming to agreement or not based on the principles they hold dear.
Sovrin’s Trust Framework makes those principles explicit by providing a documented public agreement (consensus) about them. The process for amending Sovrin’s Trust Framework is also public. The stewards running validator nodes agree (or not) by accepting or rejecting changes. This is the same process we (and other public blockchains) use for managing code changes. It ensures that Sovrin continues to be public and open to all. There is no point of centralization in Sovrin because no single entity can force other participants to bend to that entity’s will. Sovrin is technically, politically, and geographically decentralized to achieve diffuse trust, which enables diffuse control.
Others might wonder how the Trust Framework governs the code that Sovrin runs. While it’s true that Sovrin code is open source and anyone can make contributions, those contributions still have to be accepted by the maintainers of Project Indy to be incorporated in the codebase. Further, the Indy code is not automatically transferred to Sovrin validators to run, but is subject to review by Sovrin’s Technical Governance board and the Stewards themselves.
Those reviews ensure that only code consistent with the Trust Framework is run by validator nodes. These reviews also put appropriate pressure on Project Indy contributors and maintainers to adhere to Sovrin Trust Framework principles, since not doing so could lead to a fork of the project (although not the ledger), with Sovrin developing its own branch of the Indy codebase to protect its interests. Recent events in the blockchain space show that this is not inconceivable, but it would be disruptive and costly to the ultimate success of Project Indy. This neatly aligns incentives between Project Indy and Sovrin. Make no mistake; Sovrin will take decisive action if our principles are threatened.
Participation in Sovrin is open to all. As I wrote in Is Sovrin Decentralized?: anyone can create identifiers. Sovrin does not have "identity providers" because they are entirely unnecessary; identity owners are empowered to carry their own claims. Furthermore, any organization who will publicly agree to the Trust Framework and meets its requirements is welcome to become a Sovrin Steward.
This is the ultimate point: no single entity owns or controls Sovrin, not even the Sovrin Foundation. The Sovrin network is a global public utility that we all own, collectively, just like we all own the Internet. And the governance model of the Trust Framework, along with governance of the open source Hyperledger Indy code run by all Sovrin stewards, has been carefully designed to reflect this. All of this is captured in the Sovrin Trust Framework which is developed and published under an open public process available to all.
The public and open nature of Sovrin supports an unprecedented level of autonomy, privacy, security, and control by the people and organizations using Sovrin. We welcome you and your organization to join us in this important work.
There are many reasons for using a permissioned model, as spelled out in the recent Sovrin Foundation white paper, but one of the most important is cost. Simply put, if the cost of using a ledger can be kept very low, then it will enable people to use a different cryptographic identifier for each relationship, thus reducing correlation risk and increasing privacy. Any blockchain-based identity system must be carefully architected to take advantage of the blockchain without sacrificing the privacy of the participants in the system. If writing transactions to the blockchain is expensive, as it typically is in permissionless ledgers, people are more likely to reuse identifiers (or hashes) and increase the risk of unwanted correlations.
See Section 2 of the Trust Framework for a detailed discussion of these principles.
Thanks to Steve Fulling, Drummond Reed, Timothy Ruff, Jason Law, and Daniel Hardman for helpful discussions on this topic that improved the post and made it more clear.
I'm very pleased to announce that the Sovrin whitepaper is now available. The whitepaper pulls together in one place detailed information about why Sovrin exists, what Sovrin is, and how it will impact nearly every aspect of your online life. Here's the abstract:
Digital identity is one of the oldest and hardest problems on the Internet. There is
still no way to use digital credentials to prove our online identity the same way we do
in the offline world. This is finally changing. First, the World Wide Web Consortium is
standardizing the format of digitally-signed credentials. Secondly, public blockchains can
provide decentralized registration and discovery of the public keys needed to verify digital
signatures. These two steps pave the way to establish a global public utility for self-sovereign
identity—lifetime portable digital identity that does not depend on any central authority
and can never be taken away.
The Sovrin Network has been designed exclusively for this
purpose, including governance (the Sovrin Foundation and the Sovrin Trust Framework),
scalability (validator and observer nodes and state proofs), and accessibility (minimal cost
and maximum availability). Most importantly, Sovrin implements Privacy by Design on a
global scale, including pairwise pseudonymous identifiers, peer-to-peer private agents,
and selective disclosure of personal data using zero-knowledge proof cryptography.
The emergence of this infrastructure can transform at least four major markets: identity and
access management, cybersecurity, RegTech, and data integration. To provide economic
incentives for credential issuers, owners, and verifiers, the Sovrin protocol will incorporate
a digital token designed expressly for privacy-preserving value exchange. The Sovrin token
should enable a global marketplace for digital credentials of all types and value levels
together with ancillary markets for digital credential insurance and permissioned first party
data (direct from the customer).
As you can see, Sovrin will incorporate a native token for exchanging value in identity transactions. We're confident that a protocol token for Sovrin will enable use cases that would be unrealizable without it and drive the network effects for Sovrin adoption.
The whitepaper has been a long time coming. We wanted to get it right and make it as clear and understandable as possible. I'm grateful for my co-managing editor Drummond Reed, a member of the Sovrin Board of Trustees and chair of the Trust Framework working group for his diligent efforts in making this a reality. Countless others participated in discussions, made comments, or proofread various versions of the document. And a special thanks to Monique Heileson for her fine work on graphic design, layout, and illustration.
Internet identity has become synonymous with authentication and that's too bad. Identity in real life is much richer, flexibly and conveniently solving all kinds of thorny problems. Now with Sovrin, we can bring those rich identity transactions online. This paper shows how that happens and why it will impact every sector of the Internet in significant ways. I hope you'll spend some time reading it.
Picos send an receive messages over channels. Each channel has a non-correlatable identifier, called an ECI. Because picos can have as many channels as they like, you can use them to prevent correlation of the pico's identity without the pico's participation.
When two picos exchange ECIs to create a relationship, we call that a subscription. Wrangler, the pico operating system, supports creating and using subscriptions. Subscriptions allow picos to use peer-to-peer, graph-based interaction patterns. From a given pico's perspective, it has an inbound channel to receive messages (the Rx channel) and an outbound channel to transmit messages (the Tx channel).
Sean George completed a project this past Fall semester that made modifications to the channel identifier code in the Pico engine to use DIDs as the channel identifier. Because a DID is derived from an associated private key, that means that each channel also has a public-private key pair. Sean's work uses the channel keys to support signing and encrypting channel messages (using Diffie-Helman key exchange).
When a subscription is created between two picos, each pico stores the DID and public key of the incoming Rx channel. Having the public key for the other pico in the subscription allows each pico to securely message the other. There is no way to access the private keys from within KRL to protect them from unauthorized access.
Future work will focus on making the use of these functions easier and more automatic. We will also be working on integrating the Sovrin network with the pico engine. Once the DIDs are registered on the Sovrin ledger, the ledger will be used to verify the public key and outside systems will be able to make use of these capabilities without storing the public key, so long as they know the DID.
Andy Tobin has a great presentation that describes five problems of Internet identity. Our claim is that self-sovereign identity, and Sovrin in particular, solve these five problems:
The Proximity Problem—The proximity problem is as old as the familiar cartoon with the caption "On the Internet, nobody knows you're a dog." Because we're not interacting with people physically, our traditional means of knowing who we're dealing with are useless. In their place we've substituted username-password-based authentication schemes. The result is that people's identity information is replicated in multiple identity silos around the Internet.
The Scale Problem—Digital identity currently relies on hubs of identity information. We login using Facebook or Google—huge "identity providers." But for every place that uses one of these big identity providers, there are dozens that will never be part of the social login system. Many businesses are leery of giving up control of their customer information to another business that might decide next week to change things up. I don't think it's any accident that this is the same concern that was holding back online services in the days of CompuServe.
The Flexibility Problem—Many of the so-called "identity solutions" in play today are limited by fixed schema or attribute sets. For example, GOV.UK Verify is a univeral identity assurance system for UK citizens but has a limited data set. And it's unlikely that they could reasonably expand whatever schema they have to cover all use cases, even if they were inclined to do so.
The Consent Problem—And the data in these thousands of identity silos is often shared with others without consent. Sometimes this is done in service of the subject, but often it's done in service of the bottom line of the organization who controls the silo.
The Sovrin Architecture
Sovrin has a unique architecture that addresses these five identity problems. Sovrin is designed to discourage correlation, minimize disclosure, and promote security. Sovrin's architecture is decentralized so that these benefits are available to all. This is achieved through the careful combination of several important technologies:
Decentralized Identifiers (DIDs)—DIDs are identifiers intended for self-sovereign, verifiable digital identities. Sovrin uses DIDs in a manner that is pairwise and psedonymous. That is, each relationship is given a new, opaque DID by default to prevent correlation. DIDs point to DID Documents that contain public keys and service endpoints and are thus the means of locating the place the identifier can be used and providing the keys to use it.
Verifiable Claims—Verifiable claims are the digital equivalent of the various third-party credentials we all carry around in our wallets. These credentials have several important properties:
The format and content of the credential is determined by the issuer, not some central authority.
Anyone can issue whatever credentials they like.
Anyone can choose to accept whatever credentials suit their purposes
The credentials say who they're about (using a DID)
The credentials say who issued them (using a DID)
The credentials are packaged in a way that makes them tamper-evident
The claims can be verified by anyone without any kind of technical integration to or business arrangement with the issuer.
Zero-Knowledge Proofs—Zero knowledge proofs (ZKP) allow a person to prove things about themselves, based on verifiable claims, without having to reveal the claim itself. This reduces the amount of data given out by a person. For example, a ZKP can just reveal that the holder of the claim is over 18 without revealing the date of birth or even their age. ZKPs also provide support for non-correlation by proving the claim is about the identity owner without revealing the identifier that the claim issuer has for the person.
Agents—Sovrin's architecture supports independent software agents to hold and process claims as well as to perform identity transactions on the identity owner's behalf. These agents interoperate directly with each other as peers. Sovrin specifies the protocols that agents use so that agents from different vendors can work together and to support substitutability.
Distributed Ledger—A distributed ledger provides a place where decentralized artifacts like DIDs, verifiable claims, and proofs can be anchored. When agents create or resolve DIDs, they are interacting with the ledger. When an agent creates a claim or a proof from a claim, the various parts of the claim are referenced on the ledger. Without a ledger, agents would need a central repository of some sort to resolve DIDs. The ledger enables decentralized identity by doing away with the need for a central authority.
Handling the Five Problems of Identity
The architecture of Sovrin is designed to solve the five problems of identity.
DIDs and verifiable claims solve the proximity problem by giving people the means to prove information about themselves at a distance.
Agents and the ledger ensure that Sovrin scales by supporting a decentralized system of interacting peers that can scale to any size.
The decentralized nature of claims and claim schemas solves the flexibility problem because people can use Sovrin for the whatever identity problem they face. Everyone can design and use whatever claims will solve their problem.
DIDs and zero knowledge proofs provide tools for increased privacy by limiting correlation and supporting minimal disclosure.
Sovrin supports consent becausethe identity owner is structurally part of all identity transactions. Sovrin automatically records what was shared and under what terms.
Most physical world identity transactions are self-sovereign. They put people at the center and use decentralized credentials to transfer trustworthy attributes about the identity owner. The naturally support scalable, flexible, private interactions that take place with the identity owner's consent. The Internet introduced the proximity problem and the available solutions and their inherent limitations led us the situation we're in now.
Sovrin capitalizes on decades of cryptographic research and the now widespread availability of decentralized ledger technology to rethink identity solutions so that we can have scalable, flexible, private interactions with consent despite the issues that distance introduces. Sovrin introduces protocols for identity that govern interactions so as to solve the five problems of identity.
Despite the demise of Kynetx, the platform continued to be open and available. Fuse was still running on it and my students were using it for class and research. But Fuse stopped working for good last spring when the MVNO we were using to process cellular data from the car devices shut down. And the new pico engine is working so well that we use it for everything now.
KRE was started in 2007 and envisioned as a cloud-based programming platform for events. While we had several different uses for it over the years, the focus on cloud-based program evaluation never changed. KRE was a PaaS play and so we built it with the idea that it would be a big chunk of infrastructure that we maintained for use by our customers.
Back at iMall in the 90's, Steve Fulling, Wade Billings, Mark Horstmeier, Cid Dennis, and I developed a core competancy around running big server farms. And AWS was still a fairly new thing. So, we built KRE using Dell servers, Linux, virtual servers, Puppet, and other technology we were familiar with. When we built iMall, not many people could do this well and we got good at running large infrastructure. So when Kynetx started up, that was our natural path. If I were doing it today, or even just 5 years ago, I'd do it on AWS.
Over the past 10 years, KRE has operated day in and day out without fail. The only time it's been offline is when we had to physically move the servers from one data center to another. Turning off KRE and retrieving the servers is the final step in the long, exciting dance that was Kynetx. A few weeks ago I realized that the platform I'd poured my soul into for the last 10 years was no longer needed. But the ideas that it spawned live on in the pico engine. Shutting it down is bittersweet, but I'm excited for the future.
People sometimes ask "Is Sovrin decentralized?" given that it relies on a permissioned ledger. Of course, the question is raised in an attempt to determine whether or not an identity system based on a permissioned ledger can make a legitimate claim that it's self-sovereign. But whether or not a specific system is decentralized is just shorthand for the real questions. To answer the legitimacy question, we have to examine the reasons for decentralization and whether or not the system in question adequately addresses those reasons.
This excellent article from Vitalik Buterin discusses the meaning of decentralization. Vitalik gives a great breakdown of different types of decentralization, listing architectural decentralization, political decentralization, and logical decentralization.
Of these, logically decentralized systems are the most rare. Bitcoin and other decentralized ledgers are, in fact, logically centralized. There's one ledger. But the architecture (no infrastructural central point of failure) and politics (no one controls them) of Bitcoin are decentralized. But of course, these aren't binary options. There's a spectrum.
For example, while it's true that Bitcoin miners aren't centrally controlled in some aspects (e.g. who can use them, who can verify transactions), there are points of control such as the code itself. No one person has control but many have a lot more than others. I'm not saying this to throw shade on Bitcoin. Rather, I'm making the point that not even Bitcoin is completely governed by technology. There's some balance between the business, technical, and legal agreements that make up any functioning decentralized system.
The more important question about decentralization is: does the level of decentralization serve the goals of the system. There are several good reasons for going to the effort of creating a decentralized system:
Avoiding common mode failure—This is one of the most frequently stated reason for decentralization. An architecture built from many parts is less likely to fail if one of them fails. Of course, this assumes that the many components are put together combined in a way that makes this possible. And it also assumes that the failure isn't due to something they're all susceptible to (otherwise known as the monoculture problem).
Resisting attacks—Attacking a decentralized system requires taking on many components in a coordinated manner. Lack of central control points makes attacking decentralized systems much more expensive.
Avoiding collusion—Collusion is, as Vitalik puts it "coordination that we don’t like." When participants in the system conspire to cheat or gain an unfair advantage, we call it collusion.
So, rather than asking "Is Sovrin decentralized?", we might ask how the Sovrin network avoids common mode failure, resists attacks, and makes collusion difficult.
Is Sovrin Resilient to Common Mode Failures?
There are many kinds of common mode failures, but there are techniques we can use to avoid them. Sovrin is built on a distributed ledger that is readable and writable by nodes on the network. These nodes are operated by organizations of different types, industry affiliations, and size. They are spread out around the globe. Nodes are operated on different hardware, in different kinds of data centers, with different operating systems. There are no secret nodes. Everyone running a node is known. The code is open source.
There are several shortcomings currently: First, there is only one implementation. Second, most of the development is being done by one company. Both of these will be remedied over time as Sovrin becomes more self-sufficient.
Is Sovrin Resistant to Attacks?
Some attacks are mundane and can be handled using the same techniques as are used for common mode failures. One of the more sophisticated kinds of attacks that decentralized systems face are known byzantine faults. Sovrin is resilient to Byzantine failure because of the underlying consensus protocol of the ledger. Byzantine fault tolerance protects the network from coercion attacks and asymmetric information attacks.
Does Sovrin Resist Collusion?
One of the core principles of Sovrin is diffuse trust—the idea that no single node, person, or organization has to be trusted in order to trust Sovrin. Diffuse trust also ensures that many parties have to agree to take action. This shows up in the consensus protocols for the network as well as the decisions about what code to release. There are no purely technical solutions to collusion. Ultimately every decentralized system has some level of governance that backstops the technology to resist collusion. Transparency and diversity are tools that help systems resist collusion. As Vitalik says:
This third kind of decentralization, decentralization as undesired-coordination-avoidance, is thus perhaps the most difficult to achieve, and tradeoffs are unavoidable. Perhaps the best solution may be to rely heavily on the one group that is guaranteed to be fairly decentralized: the protocol’s users.
Sovrin Foundation's role is to support users. Sovrin Foundation is not an industry association for just this reason. We must be vigilant in making decisions that discourage collusion of all kinds.
Power and Self-Sovereignty
You might look at the preceding questions and think "Ok, I get that Sovrin uses decentralization to protect itself from failure, attack, and collusion. But I'm interested in decentralized systems that protect human freedom and autonomy." That's an important point that doesn't have to do with resilience so much as it does power.
One of the great advantages of blockchain-based systems is not just that they're architecturally decentralized for resilience, but politically decentralized for diffuse control. Bitcoin achieves this, for example, by allowing anyone to validate transactions on the network using a sophisticated proof of work algorithm to protect itself against Sybil attacks. We call these kinds of ledgers permissionless because anyone can read and write transactions on the ledger.
Sovrin supports broad participation through a combination of business process, technology, and legal agreement. Sovrin is "public" meaning anyone can initiate identity transactions that are validated using Sovrin's consensus algorithm. While validators on the Sovrin network are known (the definition of a permissioned ledger), they don't see inside the identity transactions and can't make judgments about which transactions to allow or disallow based on content. Validators who don't abide by the rules of the Sovrin network can be taken offline or replaced.
Sovrin does not have "identity providers" because anyone can be the source of their own identity. Sovrin's primary support for identity transactions uses something called a verifiable claim that works like a physical credential such as a driver's license of password. Sovrin's trust model for verifiable claims has four important and desirable properties that all underscore its support for autonomy:
Claims are decentralized and contextual—There is no central authority for all claims. Every party can be an issuer, an owner, and a verifier. The system can be adapted to any country, any industry, any community, any set of claims, or any set of trust relationships.
Verifiers do not need to contact issuers to perform verification—Claim verifiers (the people or organizations relying on a credential) don't have any technical, contractual, or commercial relationship with claim issuers (the people or organizations making the claim).
Verifiers make their own trust decisions about which credentials to accept—there's no central authority who determines what credentials are important or are used for what purpose.
Owners are free to choose which credentials to carry—People and organizations are in control of the claims made about them (just as they are with physical credentials) and determine what to share with whom.
All of this points to an incredible amount of autonomy and control by the people and organizations using Sovrin.
Perhaps the most important questions about Sovrin and decentralization is does it provide people and organizations with self-sovereignty. That's ultimately why the power question matters. Last October, I wrote in On Sovereignty:
Sovereignty is about relationships and boundaries. When we say a nation is sovereign, we mean that it can act as a peer to other sovereign states, not that it can do whatever it wants. Sovereignty defines a boundary, within which the sovereign has complete control and outside of which the sovereign relates to others within established rules and norms.
Self-sovereign identity describes the same situation. A self-sovereign identity system defines the things over which an entity (person or organization) has complete control along with the rules of engagement for its relations with other entities in the SIS system.
As the previous discussion makes clear, Sovrin not only defines those boundaries, but puts powerful choices in the hands of anyone using it about what personal information they share and how they share it. Sovrin is built from the ground up to use pairwise pseudonymous identifiers and support minimal disclosure as a means to protect sovereignty. This is an important point. While we often speak of privacy, we don't often link it to control. Privacy protects control and control protects privacy.
The Internet has shown tremendous resilience to attack while offering unprecedented access for everyone to publish and use information. Sovrin Foundation is working to make this same vision a reality for identity.
Some FAQs About Sovrin's Governance
The following questions and answers fill in some details that may be helpful in evaluating Sovrin as a decentralized identity system:
Who decides who can use the Sovrin?—The Sovrin network is public. Anyone can use it.
Who decides who can write to the ledger?—Sovrin is, currently, a permissioned ledger so that people can afford to create pairwise pseudonymous identifiers for every relationship they have. Consequently, Sovrin's ledger has a known set validator nodes who write transactions to the ledger and achieve consensus. Sovrin Foundation has legal agreements with the organizations who run validator nodes that control how they operate. The goal of Sovrin Foundation is to have stewards of different legal jurisdictions, industries, sizes, and structure to ensure broad representation and avoid monoculture.
Who decides who can read from the ledger?—The Sovrin network architecture allows for observer nodes who can read, but not write, the ledger. The provisional Sovrin network does not have any observer nodes. Observer nodes will also be subject to the Sovrin Trust Framework.
Who decides how code is updated?—The Sovrin Foundation has a Technical Governance Board (TGB) that makes decisions about what code validators and observers will run. The code is open sourced at the Hyperledger Indy project is the code that forms the basis of the Sovrin network, but validators run known builds that Sovrin Foundation's TGB authorizes.
How is contention resolved?—Contention about transaction is resolved automatically by the code that validators run. Contention about code is first handled by the TGB with the Board of Trustees as the court of last resort. Ultimately validators could refuse to run code that makes certain changes that they disagree with. This could result in a fork of the ledger, as we've seen with other ledgers, but I think the formal structures embodied in the TGB and Board of Trustees make this less likely. The purpose of the Sovrin Trust Framework (PDF) is define how many of the most common points of contention are resolved or avoid them in the first place.