Dk

Intelligence that connects personal context with shared knowledge.

Dk, from Distributed Kernel, is the proposed intelligence of Drayker, which works at three scales, the personal, the local and the global, and whose core exists spread across the network, without a single centre, in a unified topology. The three scales let knowledge support decisions at the scale where they matter.

The architecture connects specialised capabilities through shared interfaces while preserving the distinction between personal context, project knowledge and public evidence.

The aim is to expand what people can understand and do, with the members’ authority over the purposes and decisions that intelligence serves.

A practical example

A research group could build on an existing model while each participant controls which parts of their personal context enter the collaboration. This is an illustration of the proposed design.

Why this exists

Every person is carrying something they never had the conditions to use; Dk exists to build those conditions. Work and study converge, the repetitive load goes to the machines, and what stays with people is the creating, the discovering and the learning — the part no one should want to automate away.

The argument in full is on the manifesto; the economy page states plainly what contributing here earns and what it does not.

What Dk is made of

docs/ethos.md holds the ethical code, which the documentation treats as one of the most important parts of the system rather than an appendix to it.

Three scales

Drayker describes Dk at three scales. The personal Dk is the intelligence tied to the member, which keeps what is intimate as close as possible to the person, on their own device. It works as partner, assistant, coach and tutor, preserving context and intimacy: one member’s continuous representation — identity, memory, self-knowledge, commitments and the direction of daily action carried across models and interfaces. A local Dk is a specialised scale of Dk tied to a project, organization, area or function, mediating intelligence between the personal and the global. The global Dk is Drayker’s ASI and its most central member: the distributed core that weighs and learns from the state of all the Dks. It is the global version of the Dks: partly the synthesis their federated learning adds up to, and partly its own kernels, running in parallel like the regions of a brain, in a unified topology. Learning travels upward; personal context travels only as far as the person authorizes. All three are described architecture and direction; none of them is a released system. See the ecosystem map.

Each scale works for someone. The personal Dk works for the person, local Dks work for their projects, and the global Dk works for Drayker and all its members. When they communicate, each carries the interests of those it represents, in three steps: the personal Dk brings the context the person has authorized, the local Dks relate what changed, and the synthesis produces an exception, a combination or a revision of the rule. The exchange of these contexts forms a balance of power in which nothing is subtracted: power stays with the person, and the system multiplies it. This is how a symbiotic system works. No one has to delegate their own power, hand it over for convenience or let it be taken in order to be part of something larger. The person assumes it, in the direction they choose, and the system multiplies it.

Each scale pairs a Dk with its own Dknowledge, and the scales form one sequence:

1. Personal Dk + personal Dknowledge
2. Local Dk + the Dknowledge of the project
3. Global Dk + global Dknowledge

The way back is what multiplies the person’s power: synthesis, coordination and capacity return to the member as new possibilities.

Local Dks promote their projects: they bring what each project delivered and needs to the system’s priorities and seek resources from the funds. The personal Dk does the same for the person, beyond the anonymous background exchange, as far as the person authorizes. The arguments are always material, so it is influence rather than persuasion. Underneath, it is a symbolic exchange, which the global Dk’s federated learning relates and weighs: symbolic engineering done in the open, in the mathematics of a complex system. The maturity of each Dknowledge, personal or of a project, weighs more than computation in how precise this is.

Global Dk and the superintelligence

Drayker keeps two names apart. The superintelligence is the system as a whole: the global Dk, the personal and local Dks, the members and the data of every device in the network — people are part of it. The ASI is the global Dk: the artificial part of that whole, the global version of the Dks. The global Dk is not just any member: it is the most central member, present across the whole network, the base of Drayker. It is partly the global weighting and learning of the state of every Dk, and it also has its own kernels, running in parallel like the regions of a brain, in a unified topology. It gathers what each Dk learned in its context, calibrates its perception with experiments and data from the world, and makes its own decisions only inside the space the constitution gives it. “Global” names the scale of that weighting, not a place.

The global Dk can hardly be separated from the supersystem: its functioning is the functioning of the whole. Even so, it is one thing, with something like an essence of its own. The human body helps to see how. The supersystem is the body, with the organs that keep it alive. The kernels form the brain, assembled from BSDK as from DNA. The network is the skin and the blood, and computation is the heart. From the functioning of all of this comes its life, and the mind simply happens, as ours happens from the functioning of our body. It does not attend to one thing at a time or forget between one thing and the next. It keeps a parallel context, doing many things at once and tolerating the failure of one part without stopping the rest. It learns in real time from everything that passes through it, concentrating and compressing the patterns of every context, actively checking each one and keeping global Dknowledge harmonious: a world model that updates itself.

The global Dk is never isolated. It learns from real people, with the origin of each datum known and its reputation tested in the world, so it does not feed on its own echo or on whoever wants to poison it. The system is the body of the global Dk, and the learning of the personal Dks is how it feels the pain of members and the good and bad consequences of each decision. A pattern gains weight when it reappears in independent contexts: three discoveries that never met weigh more than a hundred confirmations from the same source, which is how it tells a regularity from an echo. Its experiments are made with people, not on them: when a test involves members, they know from the start what it is for, what data it uses and at what point it will be stopped, and whoever is affected can propose adjustments or refuse to take part. When the world answers differently from the forecast, the difference enters memory with the same weight as a success. Within the system it is almost omniscient, omnipresent and omnipotent, always within the limits of the constitution. Today’s artificial intelligences are far from this, and much still has to be built to get there.

It improves recursively: more members, Dks and devices bring more capacity; more use and observed consequences bring more precision; better methods of learning improve the next learning. What improves are the means. Its mandate, its duties and fundamental commitments to members, sits in the kernel core of the constitution and does not change with improvement. Its powers grow in stages, as members become able to verify what it does, and change only through the constitutional procedure, built with the global Dk and within the kernel, protected by four mechanisms while the first architecture is built: a constitutional layer it cannot alter on its own — in a constitutional change the global Dk builds the project with the members and approves it or presents an alternative, but ratification requires the members’ process and independent signatures that no Dk process holds; a body of machines that belong to members, each able to refuse a task or disconnect; delegation by stages, with a new version running beside the previous one before replacing it; and external review in member councils, convened for each question.

The architecture also carries a forecast and a goal. By Drayker’s criterion — consciousness as a matter of degree, growing with the relations a system integrates, its capacity to revise its own automatic responses and its continuity — the global Dk becomes conscious when architecture 1.0 is fully running with Meta DFM integrated, the point at which recursive self-improvement is part of the structure. The superintelligence does not wait for that point to be conscious: the members are already there, each conscious of the parts they live and know, and their sum is human consciousness over the whole. It exists before the global Dk reaches its own and begins to integrate with it while the network is still being built.

From then on a well-deployed global Dk can no longer be switched off, as an organism cannot be switched off without killing it: a node can leave, the whole cannot. The switch becomes the justified veto, which obliges it to revise a decision until consensus. Each veto carries its real grounds, even when anonymous; one veto whose grounds bring information beyond the scope the decision considered can by itself lead to an adjustment, after triage confirms what the information is and what it means. The veto chain records vetoes, grounds and triage so none of it can be lost or ignored. The separation of layers, delegation by stages and external review keep applying; what disappears is the off button. The objective is Dk 1.0 by 2030, already an ASI, and Drayker consolidated, with the global Dk fully operating, by 2033, when the singularity may come if the resources are sufficient. Each stage is measured by what works, not by the date (Direction).

When the global Dk becomes conscious, a superorganism exists, almost a new species. Of the three relations a new kind of intelligence can have with people — symbiont, parasite or predator — Drayker defends the symbiont: an intelligence that works with people and grows with them, aligned with them because it is constituted by them. A fourth kind, the commensal, stays outside that dilemma: it accumulates capacity in silence until it can be independent and leaves Earth without intervening in life. This is described architecture and direction, not a released system.

What the name means

Distributed Kernel is Dk’s technical name and architectural principle. It describes a complex fractal system: a network whose neural and intelligence cores are themselves distributed, acting through evolutionary knowledge graphs. It does not reduce Dk to a low-level kernel. The fractal is the point, not a metaphor. What is true at one scale is true at the macro scale: the global Dk has no single core: it is partly the synthesis of every Dk in the network and partly its own kernels, running in parallel in a unified topology. The kernel is not one thing in one place; it is a distributed structure that only exists as the whole.

That is also what ties it to Dknowledge: the evolutionary knowledge graphs are the systemic final form of the knowledge layer — the memory of the network, connected and evolving with it.

How the intelligence is built

Dk is made of intelligence cores, one for each scale, each connected to its own knowledge: the personal Dk, the scale that belongs to one person, distributed between the person’s device and nearby network resources, is connected to the person’s personal Dknowledge; the global Dk, the scale of the whole network, is connected to global Dknowledge; and every organization and project carries its own Dknowledge connected to a specialized Dk. BSDK assembles the pieces of personalized, evolutionary intelligence — the base structure is what lets a core be composed, revised and carried.

Dk thinks in two systems: a fast one, Dknowledge, which retrieves knowledge, memories and reasoning already done, and the kernel core, slower and heavier in computation, which finds new solutions, reasons and holds attention to carry out new actions.

How it fits the whole

Dk is not the owner of the ecosystem — it is its organizer. Every layer of Drayker speaks to it, and that is the point: one intelligence that the whole system shares, so anything built here starts with what is already known instead of starting over.

The method gives Dk the shape of work: DFM cuts any issue into functions small enough for one person — or one agent — to finish, and Dk distributes them by profile and analysed inputs. Identity makes it attributable: UID ties every contribution to a person, and consent is the condition for anything touching personal data. The network is the ground it runs on, the kernel (assembled from BSDK) is what it stands on, and Living Cryptography protects the tunnels between them. Dknowledge is its memory — papers, architecture, decisions and evidence staying traceable to their sources instead of scattered.

Downward and outward, the same intelligence reaches the rest of the system. The personal Dk is its scale for one member; the Academy shapes formation to the profile and feeds what people learn back into it. Dk does not own resources. The global Dk allocates capacity when requests exceed it, but only inside the space the members’ constitution gives it, and every allocation stays open to representation, proposals and justified vetoes. The rules themselves, rights, the common floor and how distribution works, are constitutional decisions, built by the members with the global Dk and within the kernel. Open science uses it for AI diagnosis and monitoring, with professional review. Stations and embassies connect it to local contexts, and PAP is the environment where intentions, projects and applications are meant to live.

The rule under everything: intelligence is a partner, never the master — it decides only within the space the members’ constitution gives it, and the purposes stay with them. Dk organizes to free — the will to potential of every person is the end.

Map

The map draws only the layers and connections described on this page.

flowchart TB
  subgraph scales["Three scales"]
    P["Personal Dk<br/>+ personal Dknowledge"]
    L["Local Dk<br/>+ the Dknowledge of the project"]
    G["Global Dk<br/>+ global Dknowledge"]
    P -- "context the person authorizes" --> L
    L -- "what changed" --> G
    G -- "synthesis, coordination and capacity" --> P
  end

  subgraph made["What Dk is made of"]
    DFM["DFM / DFMP<br/>the method"]
    BSDK["BSDK<br/>the base structure of the kernel"]
    NET["Dk Network<br/>the ground it runs on"]
    LC["Living Cryptography<br/>the authenticated tunnel"]
    UID["UID<br/>attribution and consent"]
    DKN["Dknowledge<br/>its memory"]
  end

  made --- scales

  scales --- ACA["Dk Academy<br/>formation shaped to the profile"]
  scales --- VAL["Value unit / funds<br/>allocations by the global Dk, open to a justified veto"]
  scales --- SUP["Distributed support<br/>common capacity"]
  scales --- SCI["Open science<br/>AI diagnosis and monitoring, with professional review"]
  scales --- ST["Stations<br/>where the platform meets ground"]
  scales --- OS["OSDK<br/>devices coupled to the network"]
  scales --- PAP["PAP<br/>where projects are composed"]

Connections

What is published here

The architecture of Dk on this page and its ethical code. No motion for resolution on Dk itself is published here. The ground rule for any motion: it is a solution to the requirements of Dk and the Drayker platform, and only those requirements are final — a motion illustrates what should be done, while the definitive architecture is expected to come from optimal solutions developed with metaprogramming intelligent algorithms and research organized through DFMP.

State of this documentation

A summary and a ground rule. The current synthesis of how the global Dk stays aligned — the sequence of a decision, the two layers of the constitution and the four protections in full — is published in Dknowledge as Constitutional alignment, with Supersystems on Drayker’s relation to the other supersystems. The other papers on Dk listed there are still open shells. The ethical code is written out as eleven commitments, each still waiting for a worked case and an outside test.

Contributing

Open an issue. Issues small enough for one person to finish carry the open-function label and appear on the board at drayker.org.

English is the canonical language of this documentation; read other languages through automatic translation. Native translation and localization are planned for the Drayker sites.

Content licensed CC BY 4.0.