A method you can inspect

Robust options: viable sets, not single optima

A plan can look attractive under a point estimate and fail when costs change. Start with the constraints that cannot be traded away. Compare the remaining alternatives across declared disturbances, preserving uncertain ranges and competing objectives. The result is a visible set for an accountable person to consider. It does not choose for them or establish ecological success.

How the method works

Declare a finite set of options, mandatory disturbances and hard constraints.
The kernel excludes hard failures and holds uncertain mandatory evidence.
For admissible options, robust dominance requires the lower bound of one to be no worse than the other’s upper bound in every maximizing dimension, with strict separation in at least one. Minimizing dimensions reverse the comparison.
The kernel returns its frontier with no automatic selection.

Diagram is schematic, with no measured scale. Across several dimensions, a gain in one cannot compensate for a loss in another.

Worked example: a fictional riparian decision

Fictional example — not a site or customer

All input values and the scenario timestamp are explicitly hypothetical. The result below is computed from the committed fixture during the build, using the pinned interpreter and engine.

Inspect hypothetical inputs

Budget: 120000 USD.

Habitat protection: capital 80000, avoided 180000, embodied 30000, friction 70000 (declared USD).

Corridor restoration: capital 100000, avoided 120000, embodied 20000, friction 50000 (declared USD).

Cost overrun: budget factor 1 to 1, capital factor 1.3 to 1.3, avoided-value factor 1 to 1. Fictional published fixture; not an observation.

Fictional period-end teaching ranges: protection retains habitat, restoration and mixed management reconnect habitat, and no action allows continued bank erosion. No measured prediction or joint disturbance-dependent forecast is claimed.

Connected riparian habitat at period end · hectares

Declared minimum: 10 hectares. Fictional teaching threshold.

protect: 12 to 14 hectares · hypothetical. Fictional: illustrative habitat interval for protect; not a measured forecast.

restore: 16 to 20 hectares · hypothetical. Fictional: illustrative habitat interval for restore; not a measured forecast.

mixed: 15 to 18 hectares · hypothetical. Fictional: illustrative habitat interval for mixed; not a measured forecast.

no-action: 6 to 8 hectares · hypothetical. Fictional: continued bank erosion without intervention

performance-optimum · protection 1 · corridor 0.4

Declared for a derived option — not part of the verified ecological account

connected-habitat: 14 to 18 hectares · hypothetical. Fictional: full protection and 0.4 corridor effort retain habitat and allow partial reconnection. A declared teaching interval, not interpolation or a measured forecast.

One-period profile: the model has no path within the period. Violation spans, recovery, and aftershocks were not assessed; robustness here means viability across declared disturbance scenarios at the period's end, not along the way.

Shared dependencies were not declared. Common-mode exposure (for example, one funding source or implementer behind several options) was not assessed.

Options that stay viable in every mandatory disturbance you declared

Within this finite candidate set and declared model. Includes a fictional ecological account; ecological estimates are declared, not measured.

Mixed management

Protection 0.5 · corridor 0.5

declared objective · higher: 110000 to 110000
budget margin · higher: 3000 to 3000
impact connected-habitat · higher: 15 to 18
irreversibility flags · lower: 1 to 1

Habitat protection

Protection 1 · corridor 0

declared objective · higher: 115000 to 115000
budget margin · higher: 16000 to 16000
impact connected-habitat · higher: 12 to 14
irreversibility flags · lower: 1 to 1

Held

None.

Excluded

no-action

impact-connected-habitat: impact-connected-habitat failed for affected in mandatory scenario baseline at period-end
impact-connected-habitat: impact-connected-habitat failed for affected in mandatory scenario cost-overrun at period-end

performance-optimum

budget: budget failed for affected in mandatory scenario cost-overrun at period-end

restore

budget: budget failed for affected in mandatory scenario cost-overrun at period-end

Unresolved comparisons

No overlapping frontier pair was reported.

Single-objective optimum (Run-139, point estimates)

Not on the returned frontier. budget failed for affected in mandatory scenario cost-overrun at period-end

None of these is ranked best. Choosing among viable options is a decision for the accountable owner, informed by values and evidence this model does not contain.

Recomputable record

Empirical status: NOT_VALIDATED. Conservation adapter: not formally verified.

SHA-256: 2e682a92e2484b9cb488345617facd480a836d5eb130cf70f975f1ccf086681a

All scope limits

Shared dependencies were not declared. Common-mode exposure (for example, one funding source or implementer behind several options) was not assessed.

the package proves properties of declared finite structures, closed rational intervals, Boolean authority conditions, and sampled rational trajectories. It does not prove real-world dependency completeness, calibrated uncertainty, truthful evidence, legitimate authority, causal scenario coverage, deployment, adoption, revenue, or observed outcomes.

Robust intervals are marginal and declared

The comparison engine now requires worst-case separation before declaring dominance. It treats declared marginal intervals as a rectangular uncertainty set and does not model correlation among dimensions, probabilities within bounds, model ensembles, or joint disputed assumptions. Conservative overlap may therefore leave more candidates unresolved.

The conservation adapter is software-tested, not formally verified. The comparison concerns only the finite candidate set, not every possible allocation.

Ecological account ranges, when attached, are reused unchanged across disturbances; this model supplies no disturbance-dependent ecological forecasts. Missing limits or estimates hold the comparison.

Without an ecological account, ecological effects and irreversibility are not assessed; a zero count of declared flags does not establish absence of irreversible effects.

What is proved and what is not

the package proves properties of declared finite structures, closed rational intervals, Boolean authority conditions, and sampled rational trajectories. It does not prove real-world dependency completeness, calibrated uncertainty, truthful evidence, legitimate authority, causal scenario coverage, deployment, adoption, revenue, or observed outcomes.

The following names come from the frozen statement contract. The applicability profiles and conservation adapter are software-tested, not formally verified. Empirical status: NOT_VALIDATED.

shared_dependency_strictly_reduces_independence
hidden_common_parent_lowers_independence

execution_authority_gate_iff
advisory_selection_never_authorizes_execution
revoked_or_expired_lease_denies_execution

derivable_iff_explicitly_asserted
distinct_state_is_not_derived_from_singleton

recalled_artifact_is_blocked
descendant_of_recalled_artifact_is_blocked
unrelated_artifact_is_not_blocked

robust_pass_and_fail_are_disjoint
interval_classification_complete
interval_classification_pass_iff
interval_classification_fail_iff

robust_dominance_irreflexive
robust_dominance_transitive

trajectory_shortfall_nonnegative
nonnegative_trajectory_has_zero_shortfall
marked_violation_span_nonnegative

aftershock_count_le_adjacent_pairs

release_states_are_independent_nonvacuous
robustness_kernel_v1_nonvacuous

Statement contract SHA-256: 12410f1dd42d4813263640ee3ab5e957a82629b46d9933d3237974527e468184

Published lineage

This method is Viridis's own construction. The works below are the published ideas it builds on; their authors have not reviewed or endorsed it.

J.-P. Aubin
Viability Theory (1991; publisher reprint 2009).
ISBN 978-0-8176-4910-4 (reprint)
Viability theory studies maintaining constraints through possible evolutions. Our period-end gate is a finite declared check; it is not a computed dynamical viability kernel.
J.-P. Aubin, A. M. Bayen and P. Saint-Pierre
Viability Theory: New Directions (2011).
10.1007/978-3-642-16684-6
This treatment develops viability under uncertainty. A formal derivation connecting this conservation adapter to that theory remains future work.
R. E. Ulanowicz, S. J. Goerner, B. Lietaer and R. Gomez
Quantifying sustainability: Resilience, efficiency and the return of information theory (2009).
10.1016/j.ecocom.2008.10.005
The paper relates effective performance to reserve capacity in flow networks. This implementation does not compute its network sustainability measure or infer a universal conservation threshold.
K. J. Niklas
Morphological evolution through complex domains of fitness (1994).
10.1073/pnas.91.15.6772
Simulated morphological searches explore competing functional requirements. Keeping several dimensions visible is a design connection; this conservation example is not a biological replication.
O. Hamant
Debunking the idea of biological optimisation: quantitative biology to the rescue (2024).
10.1017/qpb.2024.3
This published editorial challenges descriptions of plants as globally efficient designs and emphasizes robustness. It motivates questions about fluctuation; it does not validate this software.
J. M. Carlson and J. Doyle
Highly Optimized Tolerance: Robustness and Design in Complex Systems (2000).
10.1103/PhysRevLett.84.2529
Designed systems can tolerate anticipated disturbances yet remain sensitive to unanticipated ones. Declared scenarios here cannot establish coverage of surprises.

Publication status

Method paper in release review — not yet in the public canon.

This page makes the implementation inspectable. It does not admit the method to the public canon.

Try your declared scenario

Start from your saved decision, add disturbances and review the evidence gaps.

Open Robust options