What is a principle, what is evidence?

Core Claims: What Supports Them?

A moral commitment, a human study and a computer model are different kinds of support. None should be presented as proof of the others.

PULSE claim	Basis	Evidence type	Warranted confidence
A person’s equal standing should not depend on what they are made of.	A moral and constitutional commitment.	Normative principle	A value commitment, not a measured result.
Useful cooperation may reduce division while improving shared capability.	Human research offers an analogy; internal models test assigned effects.	Empirical analogue · PULSE simulation · Project assumption	Plausible mechanisms; a future human–synthetic effect has not been measured.
Backups are more useful when one failure cannot disable them all.	Engineering guidance and explicit failure models.	Engineering guidance · PULSE simulation	The dependency logic is clear; actual independence must be demonstrated.
The ordinary-law voting rule aims to allow decisions while requiring support in both populations.	A constitutional design choice tested against alternative thresholds.	Constitutional precedent · Governance scholarship · PULSE simulation · Project assumption	A current working choice, not a proven optimum.
PULSE does not claim conscious AI will emerge.	The proposal prepares for a possibility.	AI consciousness research · Project assumption	Unresolved.
Independent, diverse reviews may reduce mistakes about personhood.	Internal tests vary assumed reviewer error and shared mistakes.	PULSE simulation · Project assumption	Conditional on error assumptions; not a validated diagnostic test.
One civic identity cannot create extra votes by making copies.	An identity and political-accounting rule.	Normative principle · Philosophy · Project assumption	A specified rule; its technical enforcement is unfinished.
Splitting people created unlawfully into separate groups must not let them evade the political limit.	A structural rule and demographic stress tests.	PULSE simulation · Project assumption	The evasion problem is clear; real enforcement remains unresolved.
PULSE ranks first in 114 of 120 tested baseline comparisons and fourth in the other six.	A calculation reproduced using the supplied model and its summary data.	PULSE simulation	Reproduced within this model, not a forecast.
Model assumptions and sensitivity These results assume PULSE's institutions work well; the model doesn't independently establish that. In the skeptic test, moving selected PULSE institutional effects roughly 15–20% closer to the model's neutral setting removed its average lead over the strongest liberal-democratic comparator. This does not mean reducing every PULSE assumption by that amount. It is not an empirical confidence interval, a measured safety margin or evidence that PULSE is 15–20% better. Inspect assumptions and skeptic sweep →
Very long-lived ownership may threaten practical freedom.	Human wealth research motivates a question tested in project models.	Empirical analogue · PULSE simulation	A supported reason to investigate; the future scale is unknown.
A catastrophe should not let the survivor inherit permanent control of the absent population.	A constitutional commitment, examined with models of keeping government operating and returning authority after catastrophe.	Normative principle · PULSE simulation	The protection is specified; return procedures and restoration technology remain unfinished.

A person’s equal standing should not depend on what they are made of. — How do we know this?

Evidence may change who is recognized as a person. It cannot by itself establish the moral value of equal standing.

Article ii §2

[chesterman-2020]

Useful cooperation may reduce division while improving shared capability. — How do we know this?

The long-horizon integration variable combines several benefits. It does not test the exact non-coercion boundary in v1.2.1.

[contact-2006]

Inspect model archive
Expert_Spread/functional_integration_network_summary_clean.csv

Inspect model archive
data/pulse_doctrine_ablation_v2.csv

Backups are more useful when one failure cannot disable them all. — How do we know this?

More locations do not prove independence. No simulation establishes that restoring a lost population is technically possible.

Inspect model archive
Prior_Simulation_Archive/All_Models_and_Simulations/04_Targeted_v0.9.3/restoration_failure_domain_sensitivity_v093.csv

The ordinary-law voting rule aims to allow decisions while requiring support in both populations. — How do we know this?

The result depends on how support and conflict are modeled. The stricter 83/67 rule serves protected matters, not ordinary law.

Article x §3

Inspect model archive
Prior_Simulation_Archive/All_Models_and_Simulations/03_Targeted_v0.9.2/senate_threshold_grid_v092.csv

PULSE does not claim conscious AI will emerge. — How do we know this?

Indicator theories and welfare proposals do not prove a system is conscious, now or in the future.

Article xviii §1

Independent, diverse reviews may reduce mistakes about personhood. — How do we know this?

More reviewers do not help as much when they all rely on the same mistaken theory or evidence.

Article xvii §4

[butlin-2023]

Inspect model archive
Expert_Spread/personhood_correlated_error_sensitivity.csv

One civic identity cannot create extra votes by making copies. — How do we know this?

New conscious individuals created during unauthorized copying remain persons. The civic identity rule does not settle the philosophical question of whether a restored mind is the very same person.

[identity-2023]

Splitting people created unlawfully into separate groups must not let them evade the political limit. — How do we know this?

The 30% value is a current working rule, not an observed safety threshold. How to trace origins, hear appeals and manage hidden risks needs further design.

Inspect model archive
Prior_Simulation_Archive/All_Models_and_Simulations/03_Targeted_v0.9.2/rogue_cap_sensitivity_v092.csv

Baseline ranking: inspect assumptions and sensitivity

PULSE ranks first in 114 of 120 tested baseline comparisons and fourth in the other six. — How do we know this?

All six losses occur in the world with weak institutions. Rankings depend on the strengths assigned to each government design, the starting conditions and the priorities used to score results.

Inspect model archive
verified-baseline-ranks.csv

Inspect model archive
data/main_scenario_architecture_results_v2.csv

Inspect model archive
data/value_system_weights.csv

Very long-lived ownership may threaten practical freedom. — How do we know this?

Returns on human investments cannot establish how wealth would grow for future synthetic persons or select a tax system. Lifespan alone is not a complete explanation.

[fagereng-2020]

Inspect model archive
Expert_Spread/economics_persistent_return_longevity_agent_model.csv

Inspect model archive
data/private_concentration_sweep.csv

A catastrophe should not let the survivor inherit permanent control of the absent population. — How do we know this?

Restoring a species is not bringing particular dead individuals back. Abstract readiness scores are not constitutional population thresholds.

Inspect model archive
Prior_Simulation_Archive/All_Models_and_Simulations/05_Eternum_v1.1_Trusteeship_Return/trusteeship_return_sensitivity_v11.csv