Inference Investigation

Claim investigated: Federal threat assessment procurement patterns may systematically avoid proprietary system naming in public contracts, creating transparency gaps that extend beyond vendor-specific attribution questions. Entity: MOSAIC Original confidence: inferential Result: STRENGTHENED → SECONDARY

Assessment

The strongest case for the claim is that the established facts document a pattern of federal agencies deploying threat assessment systems through service contracts rather than product licensing, and that entity conflation (e.g., MOSAIC referring to entirely different systems) obscures both procurement and oversight. The strongest case against is the lack of a specific primary source contract record showing a named system omitted in a service contract, and the risk that absence from databases stems from incorrect search targets (e.g., searching for 'MOSAIC' when the relevant contracts use vendor name 'Gavin de Becker & Associates' or generic threat assessment descriptions).

Reasoning: The inference is strengthened by the convergence of (a) documented systematic conflation of MOSAIC entities, (b) established fact #14 showing agencies deploy proprietary tools through service contracts rather than product names, and (c) established fact #17 showing threat assessment systems deployed without corresponding named procurement records. However, it remains secondary because no single primary source demonstrates intentional avoidance across all federal threat assessment systems, and the claim's generality requires case-by-case verification.

Underreported Angles

• The role of FOIA Exemption 7(E) in enabling this pattern: if agencies protect algorithmic criteria from FOIA disclosure, service contracts become a structural double shield — obscuring both the existence of the system and its methodology.
• The possibility that the UK Home Office MOSAIC system, which is a case management system for asylum applications, may itself mask algorithmic risk scoring or threat assessment logic embedded within it, creating an international parallel to the U.S. federal pattern.
• The absence of a single GAO report specifically examining the use of service contracts versus product licensing for algorithmic threat assessment tools across civilian federal agencies (as opposed to defense/intelligence contexts).

Public Records to Check

• USASpending: search for contracts containing 'threat assessment' AND 'Gavin de Becker' NOT 'Palantir' for agencies: GSA, DHS, DOJ, FBI, USSS; filter by contract vehicle type 'Firm Fixed Price' vs 'Cost Plus Fixed Fee' Would confirm whether the system is procured as a service (e.g., consulting) rather than a named software license, which would support the inference of systematic avoidance.

• court records: 'Gd&Associates' AND 'threat assessment' AND 'protective order' in federal district court PACER records (2000–2025) Would reveal whether the Gavin de Becker MOSAIC system has ever been challenged under Daubert or FRE 702 admissibility standards in protective order proceedings, which would contradict or support the claim of systematic legal immunity.

• SEC EDGAR: filings by 'GAVIN DE BECKER & ASSOCIATES' (filer CIK) from 2000 to 2025 — specifically look for 'Acquisition', 'Asset Purchase', or '8-K' referencing asset sale or platform licensing Would confirm whether the threat assessment system itself (or its intellectual property) is formally a product with an asset value, rather than purely a consulting service, which would bear on procurement classification.

• GAO reports: search GAO.gov for 'algorithmic transparency', 'threat assessment', 'service contract' AND NOT 'defense' from 2018–2025 A GAO report explicitly examining the use of service contracts to avoid product naming for civil law enforcement algorithmic tools would provide direct primary evidence. An absence of such a report would itself be a finding.

Significance

SIGNIFICANT — The claim goes to the core of algorithmic accountability: if procurement patterns systematically strip identifying information from public records, then all downstream transparency efforts (GAO oversight, FOIA requests, congressional investigations, court challenges) are structurally hobbled before they begin. This is not merely a procedural curiosity but a systemic obstacle to democratic accountability for algorithmic power exercised by the state.