Dave Guarino

Many public programs—but particularly social safety net ones—are deeply, deeply complex. Handbooks for administrators run in the thousands of pages. The laws governing themselves are hundreds of pages. This complexity has a number of sources.

First, it's important to remember public programs are changed over time: new laws are passed, new rules promulgated, new edge cases and problems found that require changes. But almost all these changes are ADDITIVE: making the program *more* complex to accommodate an outlier.

Recently a human services administrator said to me, "when you look at the rules, you can almost see the history: 'Oh, here's the 70s. Now here's the 80s and the War on Drugs. Ah this bit is the 90s and Welfare Reform...'" Public policy ends up like the rock layers in the ground.

And this additive process of amending public policy means it grows sprawlingly complex: X person isn't eligible, though they SHOULD be. So we add an exception! But not everyone in that bucket should be eligible, so there are exceptions to the exceptions. Now: verify it all.

So you end up with hard ethical and distributional questions in public policy/ program design: We are adding complexity to increase eligibility for a program, *but* that complexity will be borne by *every* client who now, say, has to answer that additional question.

And the thing is, the more complex a program is, the larger the *hidden* cost of people simply getting overwhelmed and giving up even through they're otherwise eligible. So you can make more people eligible but also drop previously-eligible people by creating a friction barrier.

And the thing about approaching public programs from the vantage point of technology and service design is we can actually *measure* the costs of this complexity: - Conversion dropoff on a complex application form - The funnel of eligibility steps (say, 15% loss at interview)

I want to emphasize that the additive complexity of programs that comes from well-meaning legislators and advocates comes from a place of good intentions. And any given addition isn't hugely costly. But—iterated over decades—we end up with deeply costly levels of complexity.

And the thing is even when we try to simplify programs, sometimes we don't even really do that: we add a simplified version for some subcategory, but that's now just basically *another* line in a spaghetti diagram, even if it leads to a less complex ball of yarn.

And so here are just some of the costs of public policy complexity: - Harder for people it serves to understand - High-cost to administer - Harder to garner and maintain political support (take ACA's complex market subsidies vs. the simplicity of Medicare)

My other favorite example is around integration of different safety net benefits: everyone wants to talk up Medicaid and SNAP dual enrollment but A HOUSEHOLD IS LITERALLY DEFINED DIFFERENTLY ACROSS THE TWO PROGRAMS!! (Aside: our IBI team is doing heroic work on this problem.)

The danger of talking about streamlining complexity in public policy is political: it can be easy to propose "simplification" in such a way that it is a trojan horse for significant cutting. What we need is to truly simplify based on real costs and benefits.

But here we also have a political economy problem: there is no natural constituency for public policy simplification. In fact, many groups benefit from existing complexity, and in general it's a situation of concentrated benefits and diffused costs.

But there's also an epistemic barrier: often the costs of complexity—to the core program goals itself—are much harder to see than the ostensible benefits or justification of that complexity. Way easier to see the new eligible person than the one who dropped off on that question.

I will say that, in part because technology itself *must* model this complexity, that technology—both through implementation and higher-resolution measurement of peoples' experiences—can help us be more empirical about the costs of this complexity to people and outcomes.