{
  "$schema": "https://ndcodex.com/schemas/object/v1.json",
  "id": "codex://object/the-work-has-to-survive-the-worker",
  "archive_id": "the-work-has-to-survive-the-worker",
  "slug": "the-work-has-to-survive-the-worker",
  "url": "https://ndcodex.com/codex/the-work-has-to-survive-the-worker/",
  "type": "codex",
  "object_form": null,
  "title": "The Work Has to Survive the Worker",
  "summary": "The next test for retained intelligence is whether another participant can recover the work, understand its limits, and continue responsibly.",
  "content_text": "The conversation ends. The work remains.\n\nThat ordinary fact has become one of the central design problems in what I have been calling MK 42: an attempt to make the intelligence around work available across people, agents, sessions, and environments.\n\nA conversation can carry an extraordinary amount of understanding. It also carries repetition, abandoned directions, temporary assumptions, and decisions whose conditions may have changed halfway through.\n\nThe next participant needs a way into the present work.\n\nWhat are we trying to accomplish? Which artifact are we changing? What has actually been decided? What evidence supports that decision? What remains open? Who can authorize the next move?\n\nI keep returning to a fairly demanding test: could someone arrive without the original conversation and continue competently?\n\nThis has pushed the thinking toward recognizable structures. A claim has support and conditions. A decision has a reason and an owner. An artifact has a version. A task has a state. Relationships let us move between them.\n\nThe term Tesseract has been carrying part of this idea for me: a proposed way to preserve several dimensions of the work while making the relevant ones available to the participant who needs them.\n\nThe name can get large. The test should stay small.\n\nGive another participant a bounded piece of work. Let them recover what matters. See whether they recognize a changed condition. See what they leave behind.\n\nThe early MK 42 integration reports describe laboratory versions of that test: simulated workers, a small constructed page, changed evidence, and preserved task history. They are useful steps toward continuity. An unfamiliar agent performing consequential work, with independent verification, remains a further test.\n\nThat distinction matters to the design itself. If the system quietly turns a demonstration into a claim of general readiness, it has already lost some of the judgment it was supposed to preserve.\n\nThe recent Design Intelligence review brought another version of the problem into view. Older descriptions of limitations sat beside later records of completed technical work. The entrance needed to become shorter and more current while the history remained recoverable.\n\nMemory needs maintenance. A statement can have been accurate when written and still misdirect the next person who reads it.\n\nI want a return to an earlier decision to include its conditions. Is its evidence still current? Does its scope fit this situation? Does permission from the earlier operation apply here?\n\nThe next participant should inherit a way to find out why.\n\nThat includes access to disagreement. It includes the possibility that a previous conclusion should be narrowed, reopened, or put aside.\n\nThere is a human reason beneath all this machinery. Reconstructing a project repeatedly spends attention that could have gone into the work itself. I want the record to carry more of that burden, with enough clarity that its contents remain challengeable.\n\nThe experiment continues at the next arrival: someone enters, finds their footing, makes a useful contribution, and leaves the place easier to enter again.",
  "content_markdown": "# The Work Has to Survive the Worker\n\nThe conversation ends. The work remains.\n\nThat ordinary fact has become one of the central design problems in what I have been calling MK 42: an attempt to make the intelligence around work available across people, agents, sessions, and environments.\n\nA conversation can carry an extraordinary amount of understanding. It also carries repetition, abandoned directions, temporary assumptions, and decisions whose conditions may have changed halfway through.\n\nThe next participant needs a way into the present work.\n\nWhat are we trying to accomplish? Which artifact are we changing? What has actually been decided? What evidence supports that decision? What remains open? Who can authorize the next move?\n\nI keep returning to a fairly demanding test: could someone arrive without the original conversation and continue competently?\n\nThis has pushed the thinking toward recognizable structures. A claim has support and conditions. A decision has a reason and an owner. An artifact has a version. A task has a state. Relationships let us move between them.\n\nThe term *Tesseract* has been carrying part of this idea for me: a proposed way to preserve several dimensions of the work while making the relevant ones available to the participant who needs them.\n\nThe name can get large. The test should stay small.\n\nGive another participant a bounded piece of work. Let them recover what matters. See whether they recognize a changed condition. See what they leave behind.\n\nThe early MK 42 integration reports describe laboratory versions of that test: simulated workers, a small constructed page, changed evidence, and preserved task history. They are useful steps toward continuity. An unfamiliar agent performing consequential work, with independent verification, remains a further test.\n\nThat distinction matters to the design itself. If the system quietly turns a demonstration into a claim of general readiness, it has already lost some of the judgment it was supposed to preserve.\n\nThe recent Design Intelligence review brought another version of the problem into view. Older descriptions of limitations sat beside later records of completed technical work. The entrance needed to become shorter and more current while the history remained recoverable.\n\nMemory needs maintenance. A statement can have been accurate when written and still misdirect the next person who reads it.\n\nI want a return to an earlier decision to include its conditions. Is its evidence still current? Does its scope fit this situation? Does permission from the earlier operation apply here?\n\n**The next participant should inherit a way to find out why.**\n\nThat includes access to disagreement. It includes the possibility that a previous conclusion should be narrowed, reopened, or put aside.\n\nThere is a human reason beneath all this machinery. Reconstructing a project repeatedly spends attention that could have gone into the work itself. I want the record to carry more of that burden, with enough clarity that its contents remain challengeable.\n\nThe experiment continues at the next arrival: someone enters, finds their footing, makes a useful contribution, and leaves the place easier to enter again.",
  "author": {
    "id": "nathan-davis",
    "name": "Nathan Davis",
    "designation": "Archive Operator",
    "role": "Archive Operator",
    "handle": "@nathandavis",
    "avatar": "/media/people/nathan-davis.jpg",
    "bio": "Designer, builder, and curator of the Codex Archive."
  },
  "contributors": [
    {
      "id": "nathan-davis",
      "name": "Nathan Davis",
      "designation": "Archive Operator",
      "role": "Archive Operator",
      "handle": "@nathandavis",
      "avatar": "/media/people/nathan-davis.jpg",
      "bio": "Designer, builder, and curator of the Codex Archive."
    }
  ],
  "date_published": "2026-10-20T19:07:00.148Z",
  "date_modified": "2026-10-20T19:07:00.148Z",
  "status": "published",
  "visibility": "public",
  "language": "en-US",
  "axes": {
    "scale": "meso",
    "depth": "recursive",
    "focus": "system",
    "function": "revelatory"
  },
  "themes": [
    "retained learning",
    "provenance",
    "human-agent collaboration"
  ],
  "constellations": [],
  "tags": [
    "human-agent collaboration",
    "design intelligence",
    "continuity",
    "retained learning",
    "provenance"
  ],
  "keywords": [
    "Codex",
    "retained learning",
    "provenance",
    "human-agent collaboration",
    "design intelligence",
    "continuity"
  ],
  "relations": [],
  "media": [],
  "capture": null
}