Abundant Government: How Automation Can Save Proceduralism
How to keep the ‘good’ rules now strangling the goals they were meant to serve.
Meeting the Abundance Challenge
Given my penchant for referencing Ezra Klein’s “Everything Bagel” metaphor, it’s no surprise that I’m fully onboard with the Abundance movement and his latest book co-authored with Derek Thompson. Their argument, directed at the left, boils down to: the accretion of well-intentioned requirements—in everything from affordable housing, clean energy deployment, even basic scientific research—is now so sprawling and costly it blocks the very goals it was meant to serve.
On how to deliver Abundance the duo is more circumspect. As an Urban Technology Fellow at Cornell Tech, I became convinced that there was an opportunity between the speed of low-regulatory jurisdictions and the slow, methodical work of reform. Diving into federal, state, and city approval processes, I found not only ripe opportunities to drastically accelerate and improve every step of existing procedure, but also governments focusing on incremental fixes unsure of what larger questions to ask of technology.
That’s where automation enters. An Abundant Government defaults to applying technology and automation to all its day-to-day processes. Only through the wholesale adoption of technology in government is a third way possible that preserves the intelligence of proceduralism but matches the speed of a looser, hands-off deregulatory approach.
Why Deregulation and Procedural Reform Both Miss
The Deregulatory Tradeoff
In places like Texas and Florida, deregulation is appealing because of its speed and lower upfront cost. Those states’ aggressive growth and simultaneous discounting of long-term threats to their communities and economies embodies the inherent tradeoff. Texas leads both in clean energy and oil production. Houston continues to massively add to its population while only further deepening its bleak financial outlook, because sprawling infrastructure outpaces tax revenue. Florida is connecting its largest cities with speedy rail, significant portions of which are certain to be underwater in coming decades.
Waiting for Outdated Reform
The only things slower than lengthy review processes are efforts to reform them. Abundance case studies keep returning to three slow-motion fixes:
Carve-outs. These reforms make for splashy headlines—“Solar farms get a pass,” “Bike lanes exempt from review”—but only arrive after years of legal wrangling. Rather than streamlining, they turn the rulebook into an ever-thicker patchwork.
Time limits. These are statutory clocks, e.g. two-year caps on NEPA documents or 90-day mandates for agency sign-off. These offer welcome certainty, yet rarely come with extra staff or streamlined steps. Deadlines slip, lawsuits follow, and the calendar bloat returns.
Method tweaks. Policy wonks love new metrics: California swapped traditional Level-of-Service delay counts for Vehicle-Miles-Traveled. The change advanced good goals but layered on complexity that has discouraged other states from following suit.
Each reform targets a different lever, yet all share the same flaw: they maintain enormous manual workloads. Re-tooling methods without seriously shrinking timelines means governments can’t deliver abundance.
The Third Way: Automating Proceduralism
Caught between calls for blanket rollbacks and proposals for additional layers of safeguards, I grappled with these competing pulls as an Urban Tech Fellow, fired up by headlines like Have we found the regulation that's strangling the U.S. economy?
I interviewed current environmental review practitioners across sectors and found the bogeyman of Abundance & YIMBY nightmares. Proceduralists benefitting from the status quo, institutionalists requesting more bodies to throw at the problem, staffers frustrated by the status quo, but boxed-in by overly risk-averse rules.
I eyed New York City’s environmental review process, CEQR, which can easily stretch to two years (if the project doesn’t get abandoned!). Coming from the startup world, I had assumed that the larger challenge was one of matchmaking, between agencies, applicants and the advanced methods being refined in the tech sector.
Poring over the CEQR Technical Manual’s 800-page intricately detailed methodologies, I had the dawning realization that CEQR analysis was not a compendium of expert judgment and advanced modeling, but instead a litany of rote and formulaic steps. Compiling population data, multiplying traffic factors in Excel, or exporting sixty variations of buildings’ shadows—these were tasks delegated to the most junior consultants.
Frustrated, I built a demo to show analyses could be digitized, streamlined to take seconds not weeks. Between the technical analysis chapters, such as transportation, shadows, and emissions, I estimated, overall, that 75% of the methodologies spelled out by the City were significantly automatable. Given the persistent drumbeat of announcements of new AI capabilities—LLMs demonstrating multimodal city image and geospatial understanding, and specialty systems interpreting and flagging building plans—this is likely an underestimation and 90% is feasible.
What Technology Can Shoulder
Tasks Ripe for Code
Inside CEQR I found nearly every step sitting between “collect the facts” and “exercise real judgment” is largely rote. Feed in the size and components of the proposed project and the manual tells you explicitly whether to open chapters on noise, air quality, energy, and so on. The same decision tree mechanistic approach is true for the tiered analyses prescribed within each chapter.
The paperwork that wraps those calculations is overwhelmingly boilerplate, and distributed in massive, identical PDF packets to City agencies and community boards. Even the weighty work of choosing how best to mitigate impacts becomes formulaic: staff pull from a standard menu of noise dampeners, tree plantings, or ventilation upgrades and slot the costs into a template. This long chain of lookups, thresholds, templated prose, and file shuffling is textbook work for machines to handle.
Where Judgment Still Rules
The remainder of the CEQR process falls into four human domains:
1. The literal physical disruption of the environment—soil borings, archaeology, hydrologic sampling, or contaminant extraction—necessitating an on-the-ground technician.
2. The expert, nuanced weighing of findings: deciding whether an impact is meaningful in its local context and how policy should interpret it.
3. Real-time public engagement—listening, responding, and adjusting in front of an audience.
4. The final act of rejection or approval, balancing social, economic, and political values.
These will always reserve space for human judgment because priorities shift and reality is imperfect. But even for the first three categories the windows for automation are expanding. Drones and small robots collect field samples. AI copilots sift through technical results and surface anomalies. Chatbots translate and summarize public comments while measuring sentiment. Like with the automatable tasks above, the aim for technology adoption should be to clear away busywork while enhancing the quality of decisions and government responsiveness.
Applying Automation to (Abundant) Government
Once you see the hidden cogs and morass behind CEQR approvals, it’s impossible not to spot it throughout government. A building plans examiner comparing CAD files to tables in the Building Code; an interconnection engineer plugging prospective rooftop solar energy into a load-flow model. These plus benefits portals, housing lotteries, and innumerable government functions all run on the same plumbing: gather fixed inputs, march through prescriptive rules, deliver a compliance verdict, then push a packet to the next division or agency for their piece of the deterministic approval process. No one is exercising higher-order judgment until the results are compiled—yet the intervening weeks or months pass in unseen slow motion because those inputs are compiled by hand.
Flip the default towards automation and those waits collapse. Sensors, databases, and secure APIs can feed raw data—traffic cameras, air quality monitors, site surveys, census tables—straight into pre-approved rulesets and algorithms. Automated pipelines can then flag exceedances or spit out ready-made text as fast as the numbers arrive, while seasoned professionals maintain “human-in-the-loop” precisely where discretion, public trust, and empathy matter most.
The payoff is speed and equity. When a zoning check or a grid interconnection study runs in seconds, with transparent results, small nonprofits and mid-market developers suddenly play on the same field as firms that can afford armies of consultants. City staff, liberated from the formulaic and menial, can focus on the genuinely hard calls and tradeoffs: Are 20 permanently affordable units worth increasing a playground’s time in shadow by 10%? Is the best use of this stretch of curb a loading zone, outdoor dining, bike parking, EV charging, or a rainwater garden?
Automate the rote and the state can act at the pace—and scale—our moment demands. That is the state capacity the abundance agenda requires.
Building Faster, Guarding Better
The Abundance movement says we must build, deliver, and measure ourselves by real outcomes, shifting away from a mindset of scarcity. But I saw in New York how a doom loop of accumulated analyses and procedures leaves an invisible graveyard of public-benefiting projects alongside a dejected and distrusting populace.
Too often we’re told to pick a side: tolerate burdensome regulations and delays or scrap the rules and risk abuse. But technology lets us satisfy both demands to uphold environmental and social values and maintain an efficient, business-friendly system. Yes, it’s a heavy lift to digitize millions of pages of code, train staff, and integrate workflows, but that effort is the entry ticket to Abundant Government.
We already have the tools—what’s needed now is the commitment to wholesale digital transformation within government. If you’re a reformer, rather than asking whether you can ‘shave a week off this step’ or ‘improve on this method’—the answer to these is almost certainly ‘Yes’—instead, ask the deeper question: ‘Does this step require a person at all?’ If the honest answer is no, prioritize the work to automate it and reposition staff to best make public-serving judgments. Once the rote is coded, then we can interrogate the codified methodologies using continually-measured outcomes.
Automate the 90% of procedure and process, and building homes, grids, and tomorrow’s infrastructure becomes as fast in New York as in Texas—without ditching the guardrails that keep us safe. That is Abundant Government’s promise.