Measure the present, and show what is missing.

NovusLens measures how an organisation is built now, and it keeps a visible record of what it cannot yet see. It does not offer a model of how the business will turn out. Every page in this section rests on that choice, and this page sets out the rules behind it.

Measure the current structure

Complexity, Agility and RecoverabilityRecoverabilityThe R in C/A/R: how well a team or service restores acceptable service after a failure. Recovery times and tested alternatives can supply evidence. (C/A/RC/A/RThree separate scores for a team or service: Complexity, Agility and Recoverability. Each uses current evidence and the customer’s stated scoring rules.) and critical path lengthCPLCritical path length. The total coordination weight of the longest chain of dependencies in the model. NovusLens reports it as index points, not elapsed time. (CPL) are measures of current structure, recalculated as evidence changes. They are not predictive models, and there is no outcome model in the product to test against. The foundation paper states the principle as assumptions 4 and 7: the measures come from recorded observations that stand for aspects of the structure, and what-ifWhat-ifA scenario that changes one or more dependencies in the model and reports the result. A what-if changes nothing in the records, and its result is indicative. scenarios are indicativeIndicativeThe label on a scenario or what-if. An indicative result shows one possibility under stated assumptions. It is never a measurement of the future., not predictive.

The practical consequence is that a figure is a statement about today’s evidence. It is not a promise about next quarter. The C/A/R page and the CPL page show what each one measures and what it leaves out.

Scenarios are indicative

A what-if is a recalculation of the same measure for a hypothetical structure. Take a dependencyDependencyA link where one unit needs something from another. A dependency is not a reporting line: it shows where work relies on something else. that is currently RigidRigidA handoff that takes more than four days. It adds 3 points to the dependency-chain index. and suppose it is moved to BufferedBufferedA handoff that takes more than one day and up to four days. It adds 1 point to the dependency-chain index.. The scenario reports the index before, the index after and the difference between them. Those numbers describe the structure under the stated change. They do not estimate what the change would do to delivery, cost or customers.

Each scenario states that it is indicative of structural direction. The answer checks add that statement if the model leaves it out. A change is shown with its trade-offs alongside it, so the gain is never presented without the cost of making it.

Gaps are shown, not filled

An unknown input stays unknown. The product does not invent a value to fill it.

• A C/A/R measure with no current calibratedCalibrationThe customer's own rule that turns raw figures, such as a gauge, a workflow history or an incident log, into a C/A/R score. A score moves when the calibration moves, and the change is recorded. evidence reports unknown and gives the reason. Unknown is not zero.
• A CPL figure excludes links whose handoffHandoffWork or information passed from one team or service to another. Its measured or stated completion time determines the dependency class. is unknown and counts them. The screen shows how many links on the path were measured.
• An assessment’s coverageCoverageWhat the available records cover, and what is missing. A count such as “3 of 5 dependencies measured” names the measured portion and the gap. note counts cellsCellA box on the map that stands for a team, service, supplier or AI agent. It is a unit you can inspect, not necessarily a department or a person. with records. It does not establish that all required records are present.

The organisation mapMapThe view of teams, services and their dependencies for the selected part of the organisation. Details appear beside the map. makes the same point for dependencies. A missing record is a question to investigate. It is not evidence that no dependency exists.

Disputed claims are held as findings

When two accounts of the same fact disagree, NovusLens keeps both and raises a findingFindingA difference in the records that needs a decision. It includes its sources and can describe a slow handoff, disputed fact or missing owner.. It does not pick the winner by source order or by which record is newer.

• In the C/A/R calculation, conflicting versions of a record are kept, a conflicting-evidence finding is recorded in the calculation’s results, and neither version is used. A score is withheld when the conflict falls on one of its required inputs.
• The map holds different accounts side by side. When a person records a resolution to a challengeChallengeA recorded dispute of a value. Both the original and the disputed version are kept, and the challenge is listed under Findings., both accounts stay in its history.
• The answer layer records a conflict as its own status, so a model cannot read a disputed value as an absent one.
• Ownership disputes raise disputed-value or contested-structure findings.

Support is re-evaluated, and history is not rewritten

A conclusion rests on current support. When that support expires, is withdrawn or is superseded, the conclusion is re-evaluated. The earlier record stays in the history.

• A measurement is current for its declared window, which is 30 days by default. Once it lapses it is excluded from the current picture and counted.
• A boundary feedBoundary feedRecords supplied by the customer’s own systems about activity at the edge of an agent or service. NovusLens reads those records; it does not watch the agent. is refused if it is older than three refresh intervals. It can no longer support a finding.
• A purgedPurgeRemoving a stored document’s contents while keeping a record that it was removed and why. Conclusions that used it must be checked again. document leaves a marker recording its removal. If the same bytes come back, they receive their own identifier, and nothing they supported is restored by accident.

The rule is that support is re-checked, and the record of what was once believed is kept.

The gap between the org chart and reality is the finding

Organisation charts, service-level agreementsSLAService-level agreement: a recorded service commitment, such as a target percentage of availability. A declared target is kept separate from measured attainment. and regulatory maps are written accounts. They describe how work is supposed to be arranged. Measured reality can differ from them, and the difference is what the product is for. The foundation starts from documents. The useful result is the difference between those accounts and recorded activity.

For external suppliers the same distinction applies. The product compares a declared service level with measurements from the organisation’s side of the boundary. A shortfall becomes a finding. The two views are kept apart. They are not averaged into one number.

Agents propose, humans resolve and commit

NovusLens’s evidence workflow follows one rule. The tool proposes and surfaces. People resolve and commit. This does not control how the customer’s agents act. The foundation sets this as assumption A8.

• A finding moves through open, acknowledged, accepted and resolved. The lifecycle is owned by people. Agents surface findings and humans resolve them.
• Where ownership or structure is disputed, the product raises a contested-structure finding for a human to decide. It does not decide which account is right by itself.
• For answers, the model proposes a structure and the application checks it against the records it retrieved. A recorded fact with no supporting record is withheld on its own, and the rest of the answer stands.

Each sentence in an answer is linked to the records that support it, so a reader can see what the answer rests on.

What a measure cannot show

A reproducible calculation proves that the arithmetic is repeatable. It does not prove that the quantity is the right one. A calibration written by the organisation makes its assumptions attributable, so that they can be challenged. It does not make them valid. The organisation must test whether a measure describes what its name says, using its own history. This page does not answer that question for you.

What to do next

Read a figure’s rungRungOne of three levels of support for a recorded value: told, shown or measured. The level appears beside the value., window and reason before you act on it. For a what-if, read the statement that it is indicative and the trade-offs together. When a finding is raised, the accountable ownerAccountable ownerThe person recorded as answerable for a team, service or agent. NovusLens does not assume that this person is also the one who runs it day to day. decides what happens to it. For the boundary between what the product does and what it leaves to your teams, read the Layer-A/B boundary. For how claims move between told, shown and measured, read the evidence ladder.

Where it is implemented

• apps/reasoning/src/conway_reasoning/tools.py produces scenario results with index_points, delta_index_points and the indicative sentence.
• apps/reasoning/src/conway_reasoning/guards.py checks that what-if answers carry the indicative framing.
• apps/reasoning/src/conway_reasoning/answer_contract.py assigns each record a status (recorded, missing, explicitly absent or conflict), and withholds a recorded fact that has no supporting record.
• apps/reasoning/src/conway_reasoning/claims.py links each answer sentence to the retrieved records that name what it states.
• apps/graph/src/conway_graph/car_assessment.py withholds figures under conflict and reports unknown with a reason.
• apps/graph/src/conway_graph/engine.py excludes unknown links and expired observations from the current critical pathCritical pathThe dependency chain with the greatest total coordination weight in the map. It identifies a structural constraint. It does not state elapsed time., and counts them.
• apps/graph/src/conway_graph/governance.py compares declared service levels with boundary-measured figures for external nodes.
• apps/mesh/src/conway_mesh/model.py defines how disputes are stored and how findings progress through review by people.
• apps/graph/src/conway_graph/evidence.py and apps/connectors/src/conway_connectors/boundary_contract.py keep removal markers for purged documents and refuse stale feeds.