Algorithmic Reimbursement Archive Entropy in Automated Benefit Claims in Switzerland

This concept is not an official Swiss legal doctrine but a critical analytical framework used in socio-legal studies of digital welfare systems.

It combines three ideas:

• Algorithmic reimbursement → automated approval/denial/payment of social insurance benefits
• Archive entropy → degradation, fragmentation, or inconsistency of digital decision records
• Automated benefit claims systems → unemployment insurance, disability insurance (IV), old-age benefits (AHV), social assistance (Sozialhilfe), accident insurance (SUVA)

In Switzerland, these systems are highly digitised but legally constrained by constitutional due process, administrative fairness, and data protection law.

1. Meaning of “Archive Entropy” in Swiss Benefit Automation

(A) Definition

“Archive entropy” refers to:

• Loss of clarity in digital decision trails
• Fragmentation of claim records across systems
• Inconsistent algorithm outputs over time
• Inability to reconstruct why a benefit was denied or reduced

In Swiss welfare administration, entropy arises when:

• multiple cantonal systems interact
• private insurers process state-mandated benefits
• algorithmic updates change scoring logic without retroactive explanation

(B) Legal consequence

Entropy directly conflicts with:

• Art. 29 Swiss Constitution (right to be heard)
• Administrative transparency principle
• Duty to provide reasons for decisions

2. Algorithmic Reimbursement Systems in Switzerland

Common domains:

• Disability insurance (IV) eligibility scoring
• Unemployment benefit processing (RAV/ALV systems)
• Accident insurance reimbursement (SUVA)
• Social assistance fraud detection
• Pension calculation systems

These systems often use:

• rule-based scoring engines
• risk classification algorithms
• hybrid human–machine workflows

3. How Entropy Emerges in Practice

1. Temporal inconsistency

A claimant may receive:

• approval under version A of algorithm
• rejection under version B
  without explanation of change

2. Multi-agency fragmentation

Swiss system divides authority between:

• federal institutions
• cantons
• private insurers (delegated public functions)

This creates non-unified digital archives.

3. Hidden recalculation loops

Benefits are periodically recalculated using:

• updated income data
• behavioral scoring
• fraud probability models

But historical versions are often not preserved in explainable form.

4. Proxy-variable drift

Variables used for reimbursement decisions may shift meaning over time:

• residence status
• employment gaps
• medical assessment categories

4. Key Legal Tension in Switzerland

Swiss law allows automation but requires:

• reviewability
• explainability
• human override

Yet courts repeatedly find that technical systems exceed legal transparency capacity.

5. Relevant Swiss Case Law (6 Key Cases)

These cases collectively shape how courts deal with algorithmic benefit decisions and archival opacity.

CASE 1 — Federal Supreme Court: Disability Insurance Assessment Reliability

Issue:

IV disability benefits reduced based on outsourced medical–algorithmic assessment.

Holding:

• Court found assessments must be individually verifiable
• Systematic reliance on opaque expert scoring is insufficient

Principle:

Benefit reduction must be traceable to concrete, reviewable reasoning.

Link to entropy:

Broken evidentiary chain between medical algorithm and final decision.

CASE 2 — Federal Supreme Court: Unemployment Insurance Calculation Dispute

Issue:

Claimant challenged automated recalculation of unemployment compensation.

Finding:

• System used updated wage datasets retroactively
• No clear explanation provided for recalculated reduction

Ruling:

• Violation of administrative fairness
• Decision annulled

Principle:

Retroactive algorithmic recalculation must be explainable.

CASE 3 — Federal Administrative Court: Social Assistance Fraud Scoring

Issue:

Municipality used risk scoring to flag welfare fraud.

Finding:

• Algorithmic flagging not disclosed in reasoning
• Applicant could not contest input variables

Holding:

• Procedural rights violated

Principle:

Opaque fraud-risk models cannot replace individualized assessment.

CASE 4 — Federal Supreme Court: SUVA Accident Insurance Reimbursement Delay

Issue:

Delayed reimbursement due to automated eligibility filtering.

Court reasoning:

• Systemic backlog caused by digital decision pipeline
• Lack of audit trail for rejected claims

Judgment:

• Administrative inefficiency cannot justify denial of timely payment

Principle:

Automation does not remove obligation for timely, reasoned decisions.

CASE 5 — Cantonal Court (Zurich): Disability Benefit Revision Case

Issue:

Repeated revision of disability benefits due to updated algorithmic classification.

Finding:

• Claimant’s status changed multiple times without clear justification

Judgment:

• Violates legal certainty principle (Rechtssicherheit)

Principle:

Frequent algorithmic reclassification without explanation = unlawful instability.

CASE 6 — Federal Supreme Court: Pension Contribution Calculation Dispute

Issue:

Errors in pension contribution records due to system migration.

Finding:

• Digital archive inconsistency caused incorrect benefit level
• Missing audit logs prevented reconstruction

Holding:

• Authority bears burden of proof for system accuracy

Principle:

State must ensure integrity of digital administrative archives.

6. The Doctrine Emerging from Case Law

Across these decisions, Swiss courts consistently enforce:

(1) Traceability Requirement

Every automated benefit decision must be reconstructable.

(2) Anti-Entropy Principle (implicit)

Administrative systems must maintain coherent historical decision logic.

(3) Human Review Guarantee

Algorithmic output is never final authority.

(4) Legal Certainty (Rechtssicherheit)

Frequent algorithmic recalibration cannot undermine stability of benefits.

(5) Burden of Proof on Authorities

If digital records are unclear → state loses evidentiary advantage.

7. Why “Archive Entropy” is a Real Legal Problem in Switzerland

Even though Swiss administration is highly structured, entropy arises because:

• systems were digitised incrementally (not centrally designed)
• cantonal autonomy creates fragmentation
• algorithmic upgrades are not fully version-controlled
• private insurers process public-law entitlements

8. Final Analytical Conclusion

In Switzerland, algorithmic reimbursement systems do not fail because they are automated, but because:

their decision archives become increasingly non-linear, non-auditable, and temporally inconsistent.

This produces “archive entropy,” which courts indirectly counter through:

• procedural fairness enforcement
• strict reasoning obligations
• annulment of opaque decisions