Methodology

How this investigation was conducted: OSINT techniques, source categories, and verification approach.

Research Approach

This is an open-source intelligence (OSINT) investigation. Every finding is sourced from publicly available records that can be independently verified.

Data Sources

Federal Lobbying & Legislative Records

  • OpenSecrets - Federal lobbying client profiles, lobbyist lists, and spending reports
  • Senate LDA - LD-2 lobbying disclosure filings with specific bill references
  • Congress.gov - Federal bill text and status (H.R. 3149, S. 1586)

IRS Nonprofit Filings

  • ProPublica Nonprofit Explorer - Form 990 filings for NVF, STF, Windward, Hopewell, DCI, ConnectSafely
  • IRS Business Master File - Regional extracts (eo1-eo4.csv) covering all US tax-exempt organizations
  • Schedule I Analysis - 4,433 grants totaling approximately $2.0 billion across all five Arabella entities

State Lobbying & Campaign Finance

  • Colorado SODA API - Lobbyist income data (Headwaters Strategies expenditures)
  • Colorado TRACER - 132MB bulk campaign finance data, contribution analysis
  • F Minus - Louisiana lobbying client-firm relationships
  • LA Board of Ethics - State lobbying registrations
  • FollowTheMoney.org - Multi-state campaign contribution search

State Legislative Records

  • Louisiana, Utah, Colorado, Texas legislative databases
  • LegiScan and FastDemocracy - Bill tracking and status

Organizational Research

  • WHOIS/DNS - Domain registration dates, hosting infrastructure, email providers
  • Wayback Machine CDX API - 100+ historical snapshots of DCA website deployment
  • Corporate registries - CO, DC, DE, VA Secretary of State databases, OpenCorporates
  • For Good (Network for Good) - DAF grant recipient data (59,736 recipients, ~$1.73 billion)

Published Reporting

  • Bloomberg Government, Insurance Journal, Deseret News, ACT The App Association, Tech Transparency Project, Kansas Reflector, and others

Verification Approach

  1. Primary sources only - Every factual claim cites a primary source (IRS filing, Senate disclosure, state database, legislative record, or published reporting)
  2. Cross-referencing - Findings were cross-referenced across 24+ analysis files to identify patterns and contradictions
  3. Negative results documented - When searches found no evidence (e.g., DCA EIN, Arabella child safety grants), these negative results are explicitly documented as findings
  4. Proven vs. unproven clearly separated - 283 documented findings are kept strictly separate from 9 structurally possible but unproven hypotheses

Tools

Claude Code (Anthropic’s CLI tool, running Claude Opus) was used as a research assistant for:

  • Bulk data processing - Parsing 4,433 IRS Schedule I grant records, 59,736 DAF recipients, 132MB of campaign finance data
  • Cross-referencing - Identifying patterns across 24 analysis files
  • Drafting - Intermediate working documents and structured data summaries
  • Web searches - Against public databases (OpenSecrets, ProPublica, state portals, WHOIS/DNS, Wayback Machine)

Claude Code did not independently choose what to investigate, decide what constitutes a finding, or determine what to publish. The tool does not change whether Meta’s LD-2 filing lists H.R. 3149, whether DCA has an EIN, or whether Stefanski admitted tech funding under oath. The records exist or they don’t.

Verify the Findings

If you want to verify any finding, the source URLs and database identifiers are provided throughout the Repository. Start with the primary records, not with this repository.