The current gold rush in civic tech is around HR1 work requirements implementation. Vendors are pitching states on systems to help them comply with the new federal mandates. The pitch sounds like operational efficiency, reducing burden on caseworkers, helping eligible people stay on benefits through the new verification gauntlet. What’s being built is administrative burden as a service.
A different era of civic tech would have stayed away from this. That this era hasn’t — the firms doing the work are the same firms attending Code for America Summit, getting Knight Foundation grants, listed on Henry Grunzweig’s State Capacity Ecosystem map — tells you where the field has landed.
Somebody has to do the implementation work. States complying with these requirements will procure systems regardless of what anyone thinks of the policy. If those systems are built well, fewer eligible people get kicked off benefits in the transition. There’s a real version of the capacity argument: better tools mean fewer wrongful denials at the point of contact. The pressure localities face in meeting these requirements deserves recognition.
Other questions deserve asking. About the games being played. About the tools we’re building earnestly while entrenching a system that will be nearly impossible to remove without people enlisted to ask questions about interpretation. In my public mechanics class earlier this semester, I had students contend with SNAP requirements and how they vary by state, leaving people on the ground as arbiters of whether someone can buy something. A cookie is candy in one state, groceries in another. Cashiers interpret policy at the register. That’s where the system meets the person and where the layer this reform conversation too often handwaves past.
A vendor who builds a clean, frictionless, well-designed verification system is helping the program do what the program was designed to do: reduce caseloads by introducing friction at the point of access. Building that well is its own kind of work.
The Schism
State capacity, as the term gets used in civic tech now, has come to mean a specific cluster of operational reforms: faster procurement, faster hiring, faster firing, faster deployment, modern technology stacks, test-and-learn culture, agile delivery.
The frame operates on a theory of change (pardon the jargon) that almost never gets stated: if the inside of government works better, downstream outcomes improve. The work is upstream. The downstream is assumed. This theory falls apart in a specific way. When the work the state is being asked to do produces harm at the point of contact — wrongful denials, administrative burden as deterrence, surveillance of the poor as eligibility verification — making the inside operationally cleaner scales the harm. A faster authority to operate (ATO) on a system that wrongly denies SNAP benefits delivers harm with better infrastructure.
A faster ATO is mostly a tomorrow problem. It’s almost never a question of law. It’s the person in charge telling people to do it a different way. That’s true for most of the things this reform agenda talks about. There are legal components, sure. In reality, most of it is about discretion. What the reformers want is the ability to fire people faster and remake the apparatus to move faster, like Silicon Valley. The state capacity discourse leaves the harm question alone because engaging would require asking which work the state should be doing in the first place, and that question sits structurally outside the frame.
What State Capacity Looks Like When It Shows up
One of the challenges in the business of translation is scale. My colleagues on the international side of public delivery don’t seem to have these same struggles, though public delivery is a flavored thing — your delivery can be only as good as the state’s desire to carry it out.
We position a lot of problems as financial ones, applied unevenly. When someone comes with a gift-wrapped data center, offers to bring jobs, or threatens to take your sports team, we’re magically able to muster the will to get creative and problem-solve through special taxation authorities, funding models, and other instruments that demonstrate collaboration. The procurement constraints turn out to be more flexible than they looked. The hiring happens. The capacity is there. We see far less motivation when it comes to downstream impacts of starved budgets over years and decades. The state has the capacity to be responsive to capital and unresponsive to a mother in Winnemucca whose SNAP card got turned off. That pattern is consistent enough that calling it a capacity gap obscures what it actually is. State capacity is a question of will more than ability.
Remaking Government in the Image of a Destructive Lot
Underneath much of the state capacity reform agenda sits the unstated argument that government should be remade in the operational image of Silicon Valley. The reformers want the ability to fire faster, ship faster, iterate faster, run experiments without process constraints, and replace institutional caution with venture-backed urgency.
Remaking government in the image of such a destructive lot, given their impact, would say a lot about your view of the world. Facebook on democracy. Uber on labor. Amazon on warehouse workers. The AI labs on knowledge work and copyright. The model produces speed and consequence at the same scale. Importing that model into the apparatus that determines whether a child eats, whether a disabled person retains custody, whether a tenant can stay housed — that’s a politics.
The civic tech world is sympathetic to this politics implicitly because most of its leadership came out of Silicon Valley and treats Silicon Valley’s operational practices as the baseline for what good looks like. The reform agenda arrives from a specific industry with a specific track record, and the people pushing it are mostly the people who came out of that industry. The populations the reforms will land on stay outside the reform conversation because the reform conversation happens between people who don’t bear those consequences.
The “government is bad at delivery” story has beneficiaries. Many bad actors are invested in the notion that government is bad at delivery, that improving it would make it harder to extract money via third parties, and that some people are fundamentally undeserving of public help regardless of their situation. People working on programs don’t feel this way. The language of graft, isolated stories of fraud, and the fallacies of “freeloaders” leave technologists as sidecar participants in a broader project hostile to the populations the technology is supposed to serve.
The Non-Profit Comparison
A community-based food pantry feeding 200 families a week in rural Nevada delivers state capacity. The food pantry has lower throughput, worse data, no procurement story, and nobody at Code for America Summit talking about it. The food pantry delivers the food. The civic tech world sits structurally adjacent to the contract. The funders, career paths, conferences, and publication venues all sit on the contract side of the system. The food pantry produces fed children. The metric the system rewards is the conference talk.
Using a government contractor to deliver public services is a failed model. At least non-profits are on the ground doing the worst work for the least pay to impact the most harmed populations — underpaid, undertooled, overworked, translating policy into actual harm reduction at the point of contact. The govtech vendor whose pitch is “we will help you implement work requirements” operates at a different layer.
Does Baby Have Hat
Sydette Harry, in her 2018 Code for America keynote, named the question every technologist should be answering:
When we go to make code, when we go to make apps, when we go to do these things, we are filling a need. We are assessing, we are looking, and we are trying to make something happen that was not happening before for the most vulnerable, for the people in need. It is not often a pretty question. It’s not innovative. The innovation is not you’re doing something new. Crochet has been around for centuries. The innovation is that the person, a baby didn’t have a hat. Now a baby has a hat. That is the innovation. Changing the state of something.
The person who needed the thing now has the thing. Filling the actual need is the work. Everything else — the framework, the platform, the procurement, the agile sprint, the test-and-learn cycle, the AI deployment — is means.
What we get instead are studies, articles, and papers nobody will read, and an opposition that will brute force their ideas into being while we’re still convening the working group. We should be working faster to tackle solvable problems today — faster in the sense of getting the actual thing to the actual person who needs it. Did the SNAP recipient get her benefits, on time, without a wrongful denial? Did the Medicaid enrollee get connected to a provider who could see her? Did the unemployment claimant get paid before her rent came due? Did the immigrant family get clear information in their language about what was happening to their case? The procurement reform, the hiring modernization, the agile delivery — those count as work when they answer those questions, and only when the answer is yes.
Public mechanics takes this as the unit of measurement. The work is the actual operations on actual things that produce or fail to produce the thing the person needs. The schism with state capacity is here whether we name it or not.
Part of the bet of the Portland Digital Corps delivery pilot last year was aimed at this idea that we could start just do things. I ran a delivery sprint with only volunteers, on a shoestring, with no funding. Lots of people wanted to help, which told me more than I realized about people wanting to help in some way — they just need a way to mobilize. We’ve gotten a decade of understanding about improving capacity building at scale, and figuring out how to convey impact. Technologists aren’t going to solve these issues alone.
Our next version of this work needs to enlist a wider net of practitioners, build modular teams that succeed in deploying towards hard problems and being willing to solve them in a resilient fashion. It won’t always be glamorous, pretty or fun.
Few meaningful things are.