Every once and a while an idea comes around that is both quite popular among Very Serious People and pretty dubious. One of the latest examples of this age-old phenomena is means testing. Programs are said to be means tested when they phase out at the point where the government believes beneficiaries would be able to do without. The Affordable Care Act’s Medicaid Expansion, for example, isn’t available to households that make more than 138% of the federal poverty line. While means testing makes a degree of sense in situations where programs are designed to have a very limited scope, they ultimately restrict our ability to truly reimagine the way American society is constructed and make for worse policy.

A classic example of this failure of vision was Hillary Clinton’s 2016 college tuition plan. It promised to eliminate tuition at public colleges and universities for students who made under $125,000 a year. This cap on who would be eligible for the program proved controversial, especially among those who supported Bernie Sanders in the primary and his more universal version of the same plan. The campaign responded to these criticisms by appealing to the idea that wealthy families should not have their tuition subsidized by taxpayers. The wealthy are already capable of sending their children to college, they claimed, and so should not benefit from a program funded by tax dollars.

I find this justification to be deeply suspect. There are all kinds of public goods that the wealthy benefit from that are not means tested, from public parks to roads to K-12 education. The rich are not required to pay a fee to use a local park. Instead, these services are simply funded from federal and state budgets. If policymakers would like to place a greater portion of the burden for funding these programs on the rich they have a variety of well-worn methods available to do so, from raising rates in the upper tax brackets to changing the tax rates charged for capital gains. Why the campaign didn’t seek to fund their education plan in a similar way is beyond me.

This refusal to place college education under this well-known and popular funding scheme is a failure of imagination of the highest order. Instead of coming out and saying that they believed a college education should be a right for all those interested in pursuing it, the Clinton campaign yielded to existing orthodoxy and spoke of college as a privilege that the government would help those who met certain criteria afford. While the overall effect of the their program might have seemed similar to that proposed by the Sanders campaign, it failed to address lingering problems in college access and affordability, especially considering that college is still quite expensive for families beyond the $125,000 threshold.

As well, the Clinton campaign’s proposal lacked the universality that would give it sticking power. The power of universality in shaping public opinion is apparent in the difference between how Americans perceive Medicare and Medicaid. Medicare, the universal health insurance program for the elderly, is astoundingly popular. Medicaid, the insurance program for the poor, is markedly less so. This difference is likely due to the fact that relatively few Americans are involved with Medicaid compared to the deep involvement all elderly Americans and those close to them have in Medicare. In other words, Americans tend to believe in the importance of programs that help people like them, not abstract others somewhere else in the country.

At the root of my objections to both the Clinton campaign’s education plan and means testing more broadly, though, is a doubt that means-tested programs constitute a fundamental reimagining of the systems they work within. If we merely aim to help individuals on the margins of a system better integrate themselves into it, are we really addressing the broader structural problems that created those marginalized populations in the first place? If we intend fundamental change, our best option is not to accept the failings inherent in the current system. Our best bet is instead to attempt to root them out and create it anew.