Give Everyone the Benefit and There's No One to Investigate

In health insurance, credit, and employment, private actors mine ZIP codes, prescription histories, résumé gaps, and spending habits to assign risk and deny access, often through algorithms that are neither visible nor accountable.

As long as profits depend on refusing “expensive” or “unusual” cases, institutions will always find new proxies and new data points for exclusion.

The more granular the surveillance gets as a consequence, the more people are shut out before they even know the rules.

Data-driven gatekeeping defines the injustice of the present era.

Risk is atomized and cost is offloaded: people skip routine care, avoid seeking loans, or self-censor their ambitions, all out of fear that one mistake, or one misread data point, will brand them for years. The insurers, lenders, and employers doing the sorting bear none of that cost, so they have no reason to change it.

Universal pooling is the proven way to remove the incentive to select profitable enrollees — the strongest lever for breaking the gatekeeping cycle.1 It is the surest remedy for one mechanism, not the only way to fight every surveillance practice, and it still leaves allocation decisions, administrative exclusions, and surveillance for other purposes to be governed separately.

When every resident contributes to a single public fund and draws from it as needed, there is no profit in cherry-picking the healthy, the wealthy, or the “ideal” applicant.

Taiwan’s National Health Insurance runs on this model. After 1995, Taiwan rapidly achieved near-total coverage, erasing previous disparities and making residency and age the only criteria for care. Studies document significant increases in access and high satisfaction; administrative overhead is dramatically lower than in fragmented, private insurance markets. Similarly, in Canada’s single-payer provinces, nearly all public health dollars go to direct care rather than underwriting or administrative bureaucracy. Administrative spending in Canada is consistently lower than in the U.S., by as much as half. Risk selection is functionally eliminated, and preexisting conditions no longer bar anyone from the system.

Diagram

Proving you qualify versus being enrolled by default

View source

%% title: Proving you qualify versus being enrolled by default
%% caption: A means-tested benefit pays vendors to score applicants and mails families recertification packets, so a missed envelope ends coverage and the shrinking rolls invite the next cut. Enrolling everyone by residency leaves nothing to score and no packet to miss.
flowchart TD
  classDef system stroke-width:1.2px;
  classDef gate stroke-width:1.6px;
  classDef failure stroke-width:2.2px,stroke-dasharray:6 3;
  classDef repair stroke-width:2.4px,font-weight:bold;

  subgraph Targeted ["Means-Tested: Prove You Qualify"]
    direction TB
    T1["Eligibility Rule<br/>Benefit only if you prove need"]:::system --> T2["Vendor Fraud Scores<br/>ZIP, spending, credit history"]:::failure
    T2 --> T3["Recertification Packet<br/>Forms, documents, a deadline"]:::failure
    T3 --> T4["Missed Envelope<br/>Benefit lapses, medication unfilled"]:::failure
    T4 --> T5["Fewer Enrolled, Fewer Defenders<br/>Budget cut, rule narrowed"]:::failure
    T5 --> T1
  end

  subgraph Universal ["Universal: Enrolled by Residency"]
    direction TB
    U1["One Pool for Every Resident<br/>No means test at the door"]:::repair --> U2["Nothing to Score<br/>Name, birth date, residency"]:::repair
    U2 --> U3["No Packet to Miss<br/>Low overhead, everyone defends it"]:::repair
    U3 --> U4["Medication Gets Filled<br/>A setback does not end coverage"]:::repair
  end
  Targeted ~~~ Universal

Hybrid systems, like Switzerland’s, show what happens when universality is compromised. Despite mandatory basic coverage, risk selection creeps back in through deductibles, co-pays, and optional plan upgrades. As long as part of the system operates for profit and retains risk, the logic of segmentation (searching for data proxies) returns.

Incremental reforms do not fix the underlying incentive. Banning ZIP-code redlining, for example, only prompts insurers to switch to other variables like utility-bill payment or credit-card usage. “Fairness” mandates and transparency rules are routinely gamed or circumvented. Most so-called “score reviews” end with the same automated denials; seeing your score means little if there’s no real path to challenge the model. And when payday loan rates are capped, exclusionary practices shift to new, less-regulated corners of the market.

Universal provision changes the incentives at the root. In Taiwan, after the shift to a universal pool, entry was no longer dependent on hidden scores or surcharges. Coverage shifted from privilege to default. In Canada, early access to primary care and a decoupling of employment from coverage have steadily reduced avoidable hospitalizations and improved public trust, even as wait times for some elective procedures remain a challenge.

Universal models redirect resources away from exclusion and into service. Lower administrative costs and less time spent fighting denials mean more resources for providers, care navigators, and community outreach. The contrast is visible in U.S. states that expanded Medicaid: hospitals faced lower uncompensated care costs, freeing funds for prevention and outreach, even if the system remains imperfect.

The same holds where the gatekeeper is the state, and it answers the usual objection that universality is unaffordable. Social Security, which nearly everyone receives, spent about half a percent of its total cost on administration in 2024.2 SNAP’s state agencies spent $11.1 billion administering about $90 billion in regular benefits in fiscal 2023, and most of the federal share of that went to certifying who qualifies.3 Much of what a means-tested program spends goes on deciding who may have it.

Means-testing is a market of its own. Software vendors land multi-year contracts to root out “fraud,” data brokers sell ever more granular risk scores to justify denials, and consultants sell “equity reviews” as subscriptions that rarely shift power. An earlier essay follows investors who profited from exclusion before, from the 1493 papal decrees through redlining and convict leasing to risk-based insurance. In a public program, the buyer of the gatekeeping data is the agency doing the sorting.

The people being sorted pay the rest of the bill. Working families are tripped up by paperwork, missed deadlines and the randomness of recertification cycles, and benefits lapse for reasons that have nothing to do with need. Medication goes unfilled, and a preventable emergency can wipe out months of progress. If everyone is enrolled by default, no one can lose a lifeline over a missed envelope or a misplaced document.

Universality also breaks the chain where a single denial can haunt someone for years. Under patchwork systems, a loan denial, mortgage rejection, or insurance lapse can cascade into higher premiums and barriers across sectors. In practice, the stigma and shame of asking for help are reduced, and people trust the system enough to seek care or credit before a crisis hits.4

Skeptics argue that universal provision brings new problems: wait times, provider shortages, or bureaucratic opacity. But comparative evidence from Germany, France, and Canada suggests otherwise. Where funding follows patient outcomes and real-time data highlight bottlenecks, health systems can adapt quickly; Alberta’s team-based models and rural incentives have reduced physician shortages and burnout, though not to the exact percentages sometimes cited. And while no system is immune to opacity, countries like Germany have begun public algorithm audits and community oversight, making exclusion harder to hide.

For universal provision to succeed and stay accountable, three guardrails are essential.

1. Data collection should be strictly limited to what’s necessary: name, date of birth, proof of residency. Strong privacy laws in several states and countries have reduced insurers’ third-party data purchases dramatically.
2. Second, proxy hunting and public audit registries are needed to catch covert risk factors; where transparency and enforcement are strong, hidden surcharges and discriminatory practices decline.
3. Third, every denial should require human sign-off alongside any algorithmic veto. Recent regulation in Germany shows how oversight can make exclusion harder to sustain.

Steps short of universal pooling are already being taken where political will exists. Washington’s Cascade Select public option became available statewide in 2025, four years after launch, and pulled insurers back into counties they had abandoned. New York City’s Local Law 144 requires bias audits and public notices for automated hiring tools; compliance has been thin and direct evidence of improved callback rates remains limited, but the law marks a real step toward transparency and oversight. The place to start is automatic enrollment: use school lists for free meals, connect birth records to healthcare, and have agencies publish what share of their budgets reaches people.

None of these advances require new technology. They require a new social contract. Treat health care, credit, and work as rights rather than commodities. Make risk collective rather than a tool for exclusion. Bulldoze the maze, don’t merely rearrange the walls. Each layer of means-testing is a hidden tax that funds delay, denial and the profits of whoever keeps help out of reach. Once exclusion stops paying, the incentive to gatekeep weakens — and with it the market for the data-mining that supports exclusion. A weaker data-mining market means less surveillance aimed at selecting and excluding people. That removes one powerful driver of shadow profiles.