In the end, USDA officials rejected the waiver request based on what they said was the logistical difficulty of implementing the plan and how difficult it would be to track whether it reduced obesity.

Logistical challenges are one big reason many experts oppose the restrictions across the board. Broader bans that try to define “unhealthy” or “junk” foods run into complicated questions about how to do so.

Case in point: Kennedy, now the top government health official in the country, is currently leading an online campaign to get seed oils like canola and sunflower out of fast food, while the vast majority of nutrition professionals say there is no evidence that they are responsible for health harms.

Even soda restrictions could pose a challenge, said Katie Bergh, a senior policy analyst at the Center on Budget and Policy Priorities (CBPP), a nonprofit research and policy institute. One state bill she had reviewed, for example, exempted soft drinks that were 50 percent or more fruit juice, so a fruit-forward beverage with plenty of added sugar might sneak through, but a diet soda would be banned. Either way, she said, working out those kinks, how to make distinctions, and how to roll out new rules to retailers across the country would cost money.

“Right now, SNAP is a very efficient program,” she said. “If you look at what the federal government spends, roughly 94 percent of that amount is on benefits that go directly to families to buy food. Only about 6 percent is spent on both federal administrative costs and the federal share of state administrative costs.”

Jerold Mande, an adjunct professor of nutrition at Harvard T.H. Chan School of Public Health and the CEO of Nourish Science, who has been involved in SNAP policy for decades, brushed off that concern. He’s excited about Rollins and Kennedy bringing this issue to the forefront, and good policies, he said, could be written to eliminate confusion and minimize costs. For example, he said, since most SNAP purchases are made at major retailers like Walmart, lawmakers could even exempt smaller, neighborhood retailers if the new rules were a burden to them.

“I think if the purpose of the program is to improve diet quality, and the evidence shows that it’s not doing that, this needs to be addressed,” he said.

What We Know About SNAP, Diet Quality, and Health

It’s a statistic that gets repeated over and over, including by a conservative think tank pushing for the laws in states across the country—but it’s misleading. Very little data on SNAP purchasing exists, but in 2016, the USDA published a study on purchases using 2011 data from one leading grocer’s stores.

The data set shows a ranked list of “commodities” bought by households using SNAP dollars over the course of a year, with soft drinks totaling $357 million in spending. But soft drinks are only sitting at the top of the list because of how the items are categorized: Individual fruits and vegetables are each listed one at a time, rather than counted as a category. Pork and bacon are separate line items. Fresh chicken and frozen chicken are different.

When the purchases are combined into the categories we normally use to assess dietary patterns, things start to look different: SNAP participants spent about 40 cents of every dollar on basic foods including meat, poultry, fruits, vegetables, milk, eggs, and bread. Soft drinks accounted for 5 cents, and candy for 2. Their purchases were very similar to purchases made by families not using SNAP benefits.

In another study using national survey data, the USDA also looked at diet quality using a measure called the Healthy Eating Index (HEI). SNAP participants scored a few points lower than people with similar income levels not using SNAP benefits and higher-income individuals, but researchers cautioned the differences were so small, it was unclear if they were significant. Their main conclusion was that overall, American diets don’t align with the dietary guidelines, regardless of how they’re paying at checkout.

“The data that we do have on what SNAP participants buy shows that there’s really no meaningful difference in the types of food that people are purchasing with SNAP benefits versus other payment methods,” said Bergh at CBPP. “The diets of most Americans across income levels fall short of what nutrition experts recommend.”

Harvard professor Mande disagreed on the first point. Any data point, no matter how small, that suggests the program is failing to improve nutrition is problematic, he felt. “If you’re spending over $100 billion a year, and the purpose of your program is to improve the diet quality of Americans, and when you check how that’s going, you find out it’s actually doing the reverse, that should be a big problem,” he said.

And he pointed to other concerns in the USDA report. For example, analysts found a higher prevalence of obesity and diabetes among SNAP participants and more risk factors for heart disease. A higher percentage of SNAP participants reported drinking soda regularly.

Still, other data points led in a different direction. The USDA found children in households using SNAP had lower rates of obesity and lower prevalence of diabetes compared to families with similar incomes not using benefits. In addition, other research has found that SNAP use is linked to better long-term health outcomes, including fewer sick days and doctor visits and  $1,400 less in annual healthcare costs compared to non-participants with similar incomes.

Part of the issue, Schanzenbach said, is that it’s “devilishly hard” to study SNAP. It’s been around forever and there’s little variation. Comparing those who access benefits to people in similar income brackets who don’t is often the best you can do, but it’s still not a great research practice, she said, because those groups vary in many other ways.

Where the research is even more sparse, she said, is in illuminating whether restrictions will actually improve diets. After all, SNAP has the word “supplemental” in its name for a reason. The vast majority of SNAP recipients have other forms of income and could shift their dollars around to buy soda and candy when they want to.

And that’s their right, say many advocates and experts who feel that the entire discussion around SNAP recipients’ diets is demeaning, especially once it enters the political realm. Most Americans, after all—even those who eat as healthfully as possible—grab something sweet every now and then.

“People can have this conversation, and at the end of the day, those lawmakers will go and have a soda while they’re restricting what other people will have. There’s this idea of American values and about being to choose, but somehow that gets lost in translation if you are low-income. Then, your right to choose, it’s not in play,” said Gina Plata-Nino, the SNAP deputy director at the Food Research & Action Center (FRAC). “Every farm bill we see it, and at its core is, ‘Who is deserving?’”

Similarly, Bergh said restrictions could be burdensome and stigmatizing. “Diet-related chronic disease is a society-wide problem. So, if you’re looking for a solution, it needs to be one that cuts across income levels and that addresses all of society,” she said.

Less Controversial, Better Evidence: Nutrition Incentives

A bill that introduced restrictions on using SNAP benefits to pay for soda, he suggested, could include a provision that requires all retailers accepting SNAP to stop promoting soda through the use of aisle “endcaps,” which have been shown to be effective marketing. And that could encourage all shoppers to make healthier choices.

Nearly across the board, experts who have spent time on the issue—whether they are for or against soda bans in SNAP—support policies that incentivize those using the benefits to buy more fruits, vegetables, and other healthy foods.

At about 40 farmers’ markets across Maryland, Washington, D.C., and Northern Virginia, for example, Freshfarm offers SNAP participants a dollar-for-dollar match to purchase foods from local farmers. Their accounting shows SNAP participants spend 55 percent of their benefits on fruits and vegetables and the other 45 percent on market foods like bread, meat, eggs, and dairy.

In fact, said, Nick Stavely, the director of incentive programs, the organization has been forced to implement a daily cap because the demand outstrips the funding.

The program is funded through a USDA effort called the Gus Schumaker Nutrition Incentive Program, which has historically been popular on both sides of the aisle and powers similar efforts at farmers’ markets across the country. Last week, a bipartisan group of Senators including Tommy Tuberville (R-Alabama) and Ben Ray Luján (D-New Mexico) introduced a bill to expand the program.

“I think, through our nearly 20 years of experience and the hundreds of conversations we’ve had with folks at the markets, incentives work better than restrictions,” Stavely said. Research supports the fact that incentive programs do increase consumption of healthy foods.

But experts like Mande want incentives as well as restrictions—and in this political moment, restrictions are easier to sell. “When you take a lot of smart, thoughtful people who have deep expertise and knowledge of public policy, and you lay out the facts, they say, ‘Of course, SNAP shouldn’t be by spending money on soda,” he said. In fact, many smart, thoughtful people say it’s not that simple.

Still, when asked if her thinking has changed since the issue surfaced in New York City so long ago, Marion Nestle told Civil Eats she thought that, if anything, the arguments now are stronger.

“WIC works fine without sodas. SNAP could too. I hope the USDA finally agrees to a pilot project—health advocates have been arguing for that for ages,” she said. “With that said, the entire system is terrible, and I wish we had universal basic income instead.”