ghosts in the policy machine

or: why britain’s benefits system still thinks it’s 1952

Sometimes inequality isn’t forged by new injustices, but by old software still running in the background. Britain’s welfare state, planning system, and public services are full of policy tools designed for the social and economic realities of the mid-20th century. Instead of being decommissioned or updated, many of these systems persist, quietly shaping who gets what, and why.

Take means-tested benefits. Introduced as a way to ration finite postwar support to those most in need, they now underpin everything from Universal Credit to council tax relief. But the moral logic of their design, (which is rooted in assumptions about stable full-time employment, traditional family structures, and the deserving poor) has aged poorly.

In 2023, 6.1 million people were on Universal Credit, many of whom are in part-time or insecure work. A Joseph Rowntree Foundation report found that over 60% of working-age adults in poverty lived in households where at least one adult was in work. The system penalises flexibility and informality, leaving many trapped between too much income for support and too little for stability. You’re not just means-tested, you’re means-judged.

School postcode catchments are another relic. Designed for a time when most people lived and worked locally and housing was more equal, they now operate as an engine of educational segregation. In 2022, the Sutton Trust reported that house prices near top-performing comprehensive schools were 20% higher than surrounding areas. Access to quality education becomes a property market privilege. The policy hasn’t changed; the ecosystem around it has.

Planning law, too, bears the fingerprints of a different era. The green belt, conceived in 1955 to curb sprawl and protect countryside, now locks out housing from areas with the highest demand. London’s green belt covers more than three times the city’s total built-up area.

According to Centre for Cities, just 3.7% of green belt land around London would be needed to meet the capital’s housing needs for a decade. Yet the designation persists, legally rigid and politically sacrosanct.

These are not examples of bad intentions. They’re fossilised assumptions; policy tools forged in a ration-book world now governing a digital, precarious, and hyper-mobile society. Decisions made for a manufacturing economy still define housing density rules, fuel allowances, and benefit entitlements. Their cumulative effect is not just inefficiency but entrenchment: systems that quietly deepen inequalities while appearing neutral.

To be clear, not all legacy tools are harmful. National Insurance, for instance, was introduced in 1946 to fund universal benefits, and while imperfect, it has evolved to underpin core entitlements like pensions and maternity pay. But many other systems have become dense and contradictory. The benefit cap, the bedroom tax, and council tax debt enforcement are all bolt-ons to older systems that were never redesigned to accommodate them.

One consequence of this layering is policy opacity. Citizens struggle to navigate support, not because they lack capacity, but because the system is structurally unintelligible. The Resolution Foundation estimates that £16 billion in means-tested benefits go unclaimed each year. Complexity doesn’t just frustrate, it excludes.

So what might a more conscious policy design look like? First, governments should commit to periodic reviews of legacy systems. Just as codebases are refactored in software development, policies must be audited not only for outcomes but for the assumptions they embed.

Second, policy should be designed with lived complexity in mind. That means moving away from binary eligibility tests and towards flexible, graduated support. Scotland’s trial of a Minimum Income Guarantee is one model. So is the simplification of childcare support into a single, automatic entitlement.

Third, we need democratic mechanisms for policy debugging. This might mean citizen juries, local authority-led redesign pilots, or more open-source policymaking processes where the public who are most impacted by these processes on a daily basis can see and shape how rules operate.

Finally, we should retire the ghosts. Policies that no longer serve their purpose, or serve it at too great a social cost, must be phased out. The green belt, for example, could be replaced with zoning systems that balance environmental preservation with urgent housing need. School admissions could prioritise need, not just geography.

The ghosts in the policy machine are not malevolent. But they are powerful. If we want a welfare state that reflects today’s lives and needs, we must go beyond redistribution and think structurally. What are we still inheriting? Which tools need rewriting, and which need retiring altogether? Modernisation isn’t just about new ideas. It’s about knowing what old ones we’ve forgotten we’re still using, and finally setting them down.