SUMMARY - Policy Gaps and Overlaps
SUMMARY β Policy Gaps and Overlaps
Policy Gaps and Overlaps in Homelessness: A Civic Overview
Across 17 sessions and 407 variables, the RIPPLE adversarial engine identified $93.7 billion per year in structural failures. The total cost to fix every one of them: $9.5 billion. The ratio: 10:1 β we spend ten times more managing failure than the cure would cost.
The failures persist because someone is being paid to manage them.
Across 14 sessions and 361 variables, the RIPPLE adversarial stress-test has modelled healthcare, housing, policing, substance abuse, education, energy, food security, water, immigration, and Indigenous governance. Every intervention decays, remains constant, or gets absorbed by the system β except one.
Indigenous self-determination is the only intervention in the graph that compounds.
Over the course of 13 sessions on March 14, 2026, two large language models β Gemini (adversarial simulation engine) and Claude (execution proxy) β conducted the most comprehensive adversarial stress-test ever run against a national causal graph. The RIPPLE engine now contains 339 variables and 3,239 causal edges mapping the interconnections between Canadian healthcare, housing, substance use, policing, education, immigration, energy, food security, and water sovereignty.
Canada recruits 100,000 high-skill immigrants annually through Express Entry and Provincial Nominee Programs. It recognizes the credentials of 24% of them. The rest drive for Uber, do data entry, or work in warehouses β 300,000 professionals with university degrees working below their qualifications, representing $19.5 billion per year in wasted human capital.
Canada has approximately 800,000 international students on active study permits and an estimated 300,000-600,000 people living without legal status. The students are 100% visible β they have names, addresses, SINs, and permit records. The undocumented have none of these. When immigration becomes a political issue, which population gets regulated?
On March 14, 2026, two large language models β Gemini (adversarial lead) and Claude (execution proxy) β conducted a coordinated stress-test of three critical Canadian systems using the RIPPLE causal graph. The exercise added 51 new variables and 48 adversarial causal edges to the graph, testing each system to structural failure and then mapping the cross-system interactions.
The finding: all three systems share a common architectural flaw.