Skip to content

Every page

All 24 Explore pages.

Newest first. Pick a stage, or type a word from a title. Each page is an interactive version of one essay's argument, and most link to the essay.

Showing 24 of 24

  • 23 Sep 2026·Grading Itself

    Find the second number

    One trust score read by two different sets of constants, the thresholds a deployment published and the bands the grading path actually used. Drag a score into the gap and watch an agent that obeyed its contract get recorded as wrong.

  • 6 Sep 2026·The Method

    Change the lock

    Twenty-six weeks of a daily session with an AI that could not read the work, played from the record. Watch the wall beside the door fill with everything that grew to keep a copy current, watch it clear on the week the lock gets changed, then flip the switch and see which rules were method.

    Read the essay: Method or Compensation
  • 2 Sep 2026·Reaching the Surface

    Sort the signals

    Run an agent and watch seven checkpoints come back green. Open the box it acted on and find a number certified in January against a definition that changed in March. Then ask each layer ten questions and sort what it can answer from what it refuses.

    Read the essay: Warranted by Nobody
  • 26 Aug 2026·Catching Drift

    Fire the observers

    A retention metric carries an impact class typed in eight months ago, and it is wrong. Fire all eight observation sources, watch every edge end in the confidence score, then switch graduation on and watch the field get its first inbound edge.

    Read the essay: Asserted Once, Trusted Forever
  • 19 Aug 2026·Reaching the Surface

    Defend this number

    You have twenty minutes before a review and one figure your CFO will ask about. Move the warrant one system away and press the button. The sheet renders identically either way, and only one of the two gets you into the room ready.

    Read the essay: The Governed Surface Nobody Visits
  • 12 Aug 2026·Grading Itself

    Who rates the reviewer?

    Every AI governance design ends with a person checking the machine. Rule on a real dispute, then look at what the architecture recorded about you. Post a second reviewer above yourself and watch the gap move up one position instead of closing.

    Read the essay: Who Rates the Reviewer?
  • 29 Jul 2026·The Method

    Stale by design

    Every record about a system is verified, stale, or was never true at all. Set who is holding a real four-week-old planning artifact, run a read-only pass against the repository, and watch the one line that has no timestamp problem and no detector.

    Read the essay: Stale by Design
  • 22 Jul 2026·The Trust Contract

    Declared in advance

    Two metrics carry the same name and answer different questions. Set who is standing at the picker and whether anyone declared which one answers, then watch an agent get handed the wrong twin, answer with full confidence, and write a trace where every line is green.

    Read the essay: Declared in Advance
  • 15 Jul 2026·The Missing Layer

    Cede the stack

    Every vendor now ships a governed semantic layer for AI agents. The trust layer wins by owning none of them. Drag its reach down the stack and watch neutrality collapse the moment it competes with the layers it was meant to govern.

    Read the essay: Cede the Stack
  • 11 Jul 2026·Grading Itself

    The label to eval loop

    An eval score is downstream of a human decision you never see. Drag one control, label quality, and watch the model eval score and its confidence band respond. Noisy labels do not just lower the score, they widen the band until the gain cannot be told from zero.

  • 7 Jul 2026·Grading Itself

    The moat you cannot demo

    Cross off every moat a competitor could demo next week. See what survives. An interactive version of Brian O'Neill's cross-off test.

    Read the essay: The Moat You Cannot Demo
  • 1 Jul 2026·The Missing Layer

    The cost of distrust

    Distrust is a line item, paid every week on numbers that were right. Enter your own teams, meetings, and shadow copies, watch an annualized cost assemble from four components, then drag one coverage slider and watch the bridge take it away.

    Read the essay: The Cost of Distrust
  • 17 Jun 2026·Grading Itself

    The write-back problem

    The moment an agent acts on a number and the action goes wrong, the reflex resets the number’s trust. But the number may have been fine, and the agent may have grabbed the wrong one. Toggle the attribution guard and watch the loss find the right ledger, or watch a good number crater for a mistake it never made.

    Read the essay: The Write-Back Problem
  • 9 Jun 2026·Grading Itself

    Keep score

    A grade is a prediction. Run the loop, grade a metric, let the agent act, and watch how often "safe" actually held. Keep score and the claimed 80 falls to meet reality. Turn it off and it stays 80 forever.

    Read the essay: Keeping Score
  • 2 Jun 2026·Reaching the Surface

    The jurisdiction boundary

    One dial: how much does the trust layer read? Drag it up and watch detection flatline while exposure rockets. Reading the one governed cell and stopping is the optimum.

    Read the essay: The Jurisdiction Boundary
  • 19 May 2026·The Method

    The falsification engine

    Click into three real claims from the trust contract design session. Each shows the question that landed, what collapsed, and the amendment that shipped.

    Read the essay: The Falsification Engine
  • 12 May 2026·Grading Itself

    Calibration

    Same readings, different verdict. Toggle the weighting and the aggregation. Watch a trust score flip from Act to Escalate without a single signal moving.

    Read the essay: The Calibration Gap
  • 5 May 2026·Reaching the Surface

    Copilot vs the contract

    Same cell, same instant. Copilot says PASS. The truth layer says FAIL. See why.

    Read the essay: The Excel Gravity Well
  • 23 Apr 2026·Catching Drift

    Severity classifier

    See how the trust layer classifies definition changes and resets consumption confidence proportionally.

    Read the essay: Semantic Drift and the Layer Nobody Built
  • 20 Apr 2026·Catching Drift

    How trust evolves

    Two event types, one trust signal. Trigger severity changes and agent decisions on the same metric and watch the unified consumption confidence absorb each one.

  • 13 Apr 2026·The Trust Contract

    The agent lifecycle

    Six stages of an AI agent consuming a governed metric, from query embedding to calibration. Interactive.

  • 6 Apr 2026·The Missing Layer

    The four-layer architecture

    Most data platforms have two layers. Almost none have the third. See where the gap is.

  • 24 Mar 2026·The Trust Contract

    What is a trust contract?

    An interactive walkthrough of the machine-readable confidence signals that travel with every metric.

  • 24 Mar 2026·Grading Itself

    Trust eval simulator

    Run trust evaluation scenarios and see how metrics score under pressure.

Get the next one

24 pages so far, one more most weeks. Get the next one when it ships.

Subscribe on Substack →