A New Approach
What this realization presents us with is an opportunity to reevaluate assertions that have been baked into the tech stack and make new ones. If a different set of constraints were used to make the assertions, the assumptions about the stack would change and it would function in a different way. If we use a human-centered approach to impose new assertions upon the same stack, the stack will naturally shift to better suit human wants and needs. But just like every engineering decision, tradeoffs must be considered in order to settle on something that functions well while still meeting the design goals; certain limitations are accepted in return for some desired functionality. Tradeoffs are defined by natural laws, and so it is paramount to be aware of the natural laws which apply to the technology as well as to the design goals. With human-centered design goals in mind, natural laws which apply to humans must be understood and leveraged to define the constraints of the solution. The first protocol that the Human-Centered Computing Foundation is going to reevaluate in this manner is DNS.
For computers as well as humans, there are individuals as well as networks, each with their own defining characteristics and natural laws which apply to them. Since DNS is a networking protocol, in order to make it more human-centered, we need to map a law which defines human networks onto this protocol which helps to define computer networks. All networks are defined first by the nodes within them and then by the number of connections between the nodes, and the principle which defines this network structure for humans is known as Dunbar’s Number. Though there is some debate around the value of the number, the most commonly accepted one is 150. And what Dunbar’s Number defines is the maximum number of meaningful social connections a single person can maintain.
Mapping this to DNS would mean enforcing an assumption that each endpoint within the domain – whether an A, AAAA, or CNAME record – is a machine being used by at most 151 people: the one person who represents the node of the network plus a maximum of 150 others in their meaningful social circle. We will be subjecting the computer, which is the node in the digital network, to the same law that humans are subjected to in their social networks. Though computer systems can scale to serve millions of people, a computer system operating within the 151 user limit can be said to be operating at human-scale.
If we set that as an assertion, that every machine within a network must operate at human-scale, then we can use that assumption to optimize network functionality around that trait. Shared resources or services provided to the network can enforce that as a limit to prevent abuse and applications can be designed and deployed on those machines without needing to factor in massive scalability per instance. In fact, they could scale in a different way: horizontally across peers in the network with a centralized discovery service.
Shared Services
One service inextricably linked to DNS and specific domains is email: everything after the “@” in an email address is a domain that someone has configured to work with the email protocol. As a protocol, it does not have any built-in guardrails to prevent abuse, and so all the big email providers go to great lengths to police email traffic and other protocols have been created specifically to limit and prevent as much abuse as possible. These preventative measures make it very difficult to self-host your own email inbox: the big email providers simply assume small email servers are bad actors by default and block delivery of all emails sent from it to all emails within their domain.
Email is a great example of a protocol that could be altered by imposing human-centered assertions on how it is used. One such assertion could be a limit of 150 emails per day, enforced by a client, a shared relay used by all clients, and dropping all emails that try to bypass said relay. These limits would prevent the domain and its TLD from being used for spam, provide the ability to monitor for phishing attempts, and ultimately enable individuals to self-host their own email by preserving the reputation of domains within the TLD. It would also prohibit the ability to run mass-mailing or newsletter campaigns from any domain within that TLD, but that is an acceptable trade-off; those things are more business-oriented than human-oriented and there are plenty of providers today that offer such services.
Applications
Applications consist of two distinct elements: the frontend, which is the interface that the user interacts with, and the backend, which is where the data lives and the application’s meaningful functions are executed against that data. Traditionally, an application’s backend is a tightly-controlled, centralized service run by the company who owns the application. All the application data from all its users is colocated within the backend service and made available to the frontend service through programmable interfaces.
Since the user data must exist regardless, a more human-centered approach would be for each user to host their data, advertise its availability through a simple discovery service, and then for sharing and transacting to be handled through peer-to-peer protocols. Though this distributed backend would inherently be less technically efficient than the centralized one, it would also enable an entirely different operational and business model for the application as a whole. With the value-creating data under full ownership and control of the users, they would have the power to decide when and how their data is used. Any business model which aims to monetize that data will have to directly incentivize participation of those users in order to get that value for itself. Users could also transact with each other directly by leveraging the discovery service to find the other user that they want to transact with.
Though this novel approach may not be able to fully replace certain applications which inherently function better based on centralized backends, it could at least provide an alternative model that works well for certain use cases and enable people to take more ownership over their digital lives.
Conclusion
With so many protocols that make up the tech stack, our first challenge was to decide where to start. We chose to start with DNS because DNS is a human interface protocol which has a far-reaching impact on other technical protocols as well as on how humans interact with technology. We see DNS as more than just a networking protocol; we see it as a way to create digital domains for everyone. We want to create a TLD which is more than just another string at the end of a URL, but one where people understand “all my data lives at my .self domain”, which physically exists on hardware in their own possession. Digital technology is largely mysterious to most people and we want to make it feel more concrete and real for them.
The technology industry so far has failed to bridge this gap, not only because it is easier to build and monetize things when they are completely under their control but also because this gap does present some especially challenging technical and commercial problems that must be overcome. Our ultimate goal for .self is that it becomes one of the pillars upon which this bridge can be built.
