How Code for America Is Changing the Way Cities Run – Next City

Help us make our spring fundraising goal.

$1,865
$5,000 goal

How Code for America Is Changing the Way Cities Run

No App Will Save Our Cities, But That Doesn’t Mean There Isn’t Coding to Be Done

Story by Nancy Scola

Illustrations by Hawk Krall

Published on Jul 8, 2013

Jennifer Pahlka is playing devil’s advocate. “You can’t,” she says, “solve the world’s problems through apps.”

I’ve asked Pahlka to pinpoint what might be the most relevant criticism of Code for America, the widely praised San Francisco-based non-profit she leads. Pahlka is wearing a peach parka, despite it being in the mid-60s this early April day. She’s got a cold, she explains, with one leg of her jeans tucked under the other, her backdrop the single biggest monitor I’ve ever seen on someone’s desk. But being under the weather does little to diminish her intensity. Pahlka, 40-something with a blond-brown pixie cut, has been walking me through Code for America’s bet: That bridging the gap between the thriving world of technology and the limping world of government can restore the public’s faith in politics and civil servants’ faith in public service.

Contrarian is a role that Pahlka has come to play with gusto over the last four years, as she’s turned an idea hatched over beers at a family reunion into a multimillion-dollar operation credited with helping to transform the relationship between governments and the people they serve. But there’s only so long Pahlka is willing to entertain the idea that the discrete bits of software her fellows build on behalf of cities are just inconsequential bits of software.

“No, of course you can’t,” she continues, on the potential for apps to save the universe. “But you can start a dialogue. You can figure out ways that give people new tools and get them re-excited about government. You can give people opportunities to be the agents of culture change.”

If culture change is what you’re after, Pahlka has a point. Just a handful of years ago she was busy organizing tech conferences. In late April, U.S. Chief Technology Officer Todd Park testified to me that, “Jen and Code for America really pioneered the whole idea of tech innovators coming into government to code a better government.” And just seven weeks after our interview, Pahlka will be named a deputy chief technology officer of the United States. Serving under Park, she’ll be responsible for heading the White House’s government innovation efforts, and part of her portfolio will involve leading the Presidential Innovation Fellowships.

It will be a familiar assignment. The White House launched the program last spring and was, according to Park, “directly inspired by Code for America.” The Washington stint will last about year, Pahlka says, making it a kind of fellowship of her own.

And it all began with apps. Pahlka pinpoints a moment when Code for America fellows, dispatched to city halls across the country, demo the open-source tools they’ve whipped up. “Government can look like that?” she says, describing the reaction of someone confronted for the first time with the simple but powerful transparency of, say, Where’s My School Bus?, a Boston-based app for parents to track their kids’ rides home, or the direct engagement possibilities of a Textizen, a mobile tool, designed for Philadelphia, that lets a city government easily survey its citizens.

If the fellows are missionaries, their New Testament is likely the 2011 bestseller The Lean Startup. In the book, tech entrepreneur Eric Ries advocates for quick cycles of “build-measure-learn,” a model for turning out software that fits end users like a glove. Their Old Testament is perhaps the U.S. Constitution. A poster on the wall of the Code for America headquarters in San Francisco reads, “We the people of the United States, in order to form a more perfect union” — rendered in the untranslated 0s and 1s of binary code.

But spend some time in the Code for America universe, and it quickly becomes clear that getting from even the savviest apps to perfecting the democratic experience is hardly as simple as point-and-click.

Arizona Daydreams

For the 2013 cycle — the third round of Code for America’s city partnerships — more than 50 cities applied for fellows. Eight eventually signed up to pay $360,000 for the privilege of hosting a small team of technologists, though outside funders are expected to cover some of the tab. Nearly all involved say that Code has helped to coalesce notions of participatory government floating around in cities. In some cases, cities are so sold that Code alumni are hired on to continue their work. It’s a pattern that appears to be charging along. But looking back, Code for America’s moment of creation was marked, in a few different ways, by desperation.

It was 2009, and Pahlka was at a family reunion in Flagstaff, Ariz. She picked up an ongoing conversation with Andrew Greenhill, spouse of one of her best childhood friends and chief of staff to the mayor of Tucson. “Those were the years,” Greenhill recalls now, “that my city budget seemed to plummet.” He couldn’t pay for all the services the city needed, and had the vision for pieces of civic software that might fill some of the gaps. Perhaps his friend Jen could figure out how to get him help? “All those fancy developers that you have coming to Web 2.0,” Pahlka recalls the conversation with Greenhill going, “can you get some of them to come write apps for Tucson?”

Indeed, Pahlka had been yoked to the tech world since shortly after moving west in the early ’90s. She graduated from the rigorous Bronx High School of Science, then Yale as an American Studies major, and went to the Bay Area intending to qualify as a California resident and attend grad school. (Somewhere along the way she “forgot” to do this last part, Pahlka says with a laugh.) She found work with non-profits but hated the experience, finding them depressingly sclerotic and maddeningly hierarchical.

She took some time off and traveled in Asia. Once back in the Bay Area, she picked up a job organizing tech conferences. “Which was funny,” she says in our interview at her office, “because I knew nothing about tech.”

But by the time she and Greenhill were kicking back in Flagstaff in late aughts, Pahlka was a little disenchanted. The conference company she worked for had partnered with O’Reilly Media, headed by tech world luminary Tim O’Reilly, to launch the Web 2.0 conference — based on the idea that the Web was, at its most powerful, a collaborative medium.

It was exciting. At first. One minute, folks you know are pivoting their flailing gaming platform into a photo sharing community. The next minute, it’s Flickr and they’re millionaires. But, Pahlka says, it started to grate. It became all about building brands and winning at social media.

For a minute or two, there had been a flicker of hope on the opposite coast. Through conference organizing, Pahlka was at the heart of the so-called Gov 2.0 push. But that direct approach seemed, in some ways, born cursed. D.C. was an intensely vendor-driven place, and by the time the focus on connecting humanity bubbled through the layers, it had become so watered down that the State Department tweeting seemed a major victory.

Back in Arizona, Greenhill was worried about more practical matters. Talk that day, as both he and Pahlka remember it, turned to another time that a struggling city needed help — and Greenhill had been the one to step up. In 1991, Greenhill had been a member of the second corps of Teach for America, serving in Houston. Under the program, students from good colleges are placed as teachers in challenged schools for two years before, in most cases, moving on.

Huh. Pahlka says today that in her past conversations with Greenhill, she’d been operating under the notion that, “Nobody in San Francisco who is pitching a VC on Sand Hill Road is going to go to Tucson and write you an app, dude.” Well, not for the money at least. Teach for America, though, raised another possibility. “Maybe,” Pahlka recalls the thought, “people might take a year off and do it.”

“I was like, boom!” Pahlka says, laughing heartily but her voice raspy. “Beer. It involved beer.”

Code’s focus on cities, then, was something of an accident. But it has become central to the organization’s very mission. The fellows, Pahlka says, write often in applications of their interest in “the city as a life form,” representative of the cultural spotlight that cities are enjoying today. And on a more nuts-and-bolts level, it’s a happy accident. “A lot of the machinery of government is down here,” says O’Reilly, the tech figure who is today an emeritus board member of Code for America.

That is to say that much the direct contact that normal citizens have with government, day in and day out, happens locally, from taking the city bus to sending your kids to public schools to engaging with police to coping with zoning laws to, even, getting federally run but locally administered benefits like health care coverage. And it produces a lot of government data, the firewood that fuels civic tech.

More than that, though, it’s government on human-to-human scale. “We’re all about service delivery,” says Nigel Jacob, co-director of Boston’s Office of New Urban Mechanics and a Code for America board member. “At any given moment, people can walk in off the street and yell at us for doing a crappy job of something.” That intimacy, argue the folks at Code, creates an opening through which government can be reclaimed for the people.

“Culture virus” is the phrase you hear again and again in Code circles. The notion is simple: Start thinking differently about what technology can do in the public’s service, and other people will catch the idea.

And Code for America has, rather inarguably, made it far, far cooler to be a “government geek,” in the words of Dan Melton, a technologist who joined Pahlka as Code’s employee number two. The bar wasn’t all that high to begin with. Trying to do tech inside bureaucracies has, in the modern era, been something of an exercise in discouragement. Jacob channels the moanings of a typical pre-Code nerdy public servant: “The technology sucks. Procurement sucks. It’s hard to do anything good.”

Buried in that complaint is a critical point, if obscured by bureaucratese. Procurement sucks. Or, to put it another way, the means by which government buys and builds gadgets is broken.

Here, along with the validation of government geekdom, is perhaps Code’s other inarguable success. It has demonstrated that IT procurement can suck less.

In brief, the situation is this: Across the board, government in the United States isn’t set up to acquire or make many off the neat tech tools taking the rest of the world by storm. How’s that? In government-land, technology is something that generally comes pre-packaged and off the shelf, or made by a known, often quite large, IT vendor. The price is usually high, but with that comes the comfort of knowing that you’re buying a product that, in theory at least, captures a known solution to a known problem. More than that, it tends to come with a robust support contract. If something breaks, you call a number and someone smart comes and fixes it.

Compare that to the way that even the Web’s biggest companies and most powerful tools got their start. In many cases, curious geeks whip something up on the cheap and quickly push it out to see if other people like it. Someone may or may not be around to answer the call should something not work right. That’s not so much a failing as it a manifestation of an ethos. Go ahead and fix it yourself, goes the gleeful call of the DIY software world. Says the government, Come again?

“Procurement eats innovation for lunch,” bemoans Ries, also a Code for America board member, in The Lean Startup.

So by getting cities to opt in to a more exploratory approach to building software — “everything is sort of a thesis” when the fellows first get going, Pahlka says — they’re broadening the range of possible ways that government can innovate. Figuring out how to get Code as a line item in a city budget might be the start of a much larger and more transformational rethink of government procurement.

In Macon, Ga., for example, the Knight Foundation — a Code funder — paid the lion’s share for the three fellows brought in last year. Code puts the cost of each fellow at $60,000, including salary, benefits and travel, with the rest of the contract cost going toward sustaining the organization. For this cycle, Kansas City on each side of the Kansas-Missouri line paid $45,000, and is still looking for means of covering the rest. Santa Cruz, Calif., “didn’t have the money” last year, says Peter Koht, the city’s economic development point person. So surrounding areas were called upon to chip in for its business-centric Code for America engagement.

All that said, disrupting government IT procurement wasn’t exactly the beginning and end of the dream that got Pahlka and Greenhill so excited in Arizona that day. The ambition was, in large part, to use tech to make government of, for and by the people actually work better on their behalf. Can it?

Inside the Code Factory

That’s the question they’re busily trying to solve here at the Storek Building, a onetime leather factory and the site of Code’s headquarters in the South of Market neighborhood, a gentrifying but still-gritty slice of San Francisco. When I visit in early March, outside there’s a woman wandering around without pants on. Inside, though, the space is beautiful. Black columns run all the way from mahogany floors two airy stories up to timber beams from which hang industrial-cool light fixtures.

Maybe you’ve seen the episode of 30 Rock where Jack walks into his new Washington job. He’s thrilled to finally find himself in the seat of power. Except, his ceiling’s leaking. His boss assures him the issue has been studied and that the ceiling is, in fact, not leaking. One way of picturing Code for America’s headquarters is as a rejection of Jack Donaghy’s Washington. It is, by design, the physical embodiment of the idea that giving over your skills to your country doesn’t have to be quite so absurd, nor corrupted by myopic process, bad lighting, cheap carpets and unappealing cafeteria food.

Perhaps as a product of that vibe, the Code for America HQ has become ground zero for the civic tech movement. A graying man tapping away in one corner of the space turns out to be tech legend Tim O’Reilly. A few feet away from him, there’s a pod of former fellows from New Orleans trying to spin their app into a start-up. On one entry table, a bowl of self-affirmation stickers reads things like “Government Rockstar.” On another, a giant copy of The Human Face of Big Data. The space is humming. In a room to the back, comedian Baratunde Thurston is leading a session about being funny online over a lunch of sushi-like Korean rolls.

“It was a long road” from Flagstaff to the thrumming scene in the Storek Building, Pahlka says in her glass-walled first-floor office. Early funding was hard to come by. The 2011 inaugural crop of four cities was pulled together “by the skin of our teeth,” remembers O’Reilly. But this year, there’s money in the coffer. By its last tax filing, for 2011, Code had more than $2.3 million in revenue, a third of which came from cities. (Code says that cities now pay the bulk.) The rest comes from Google, the Omidyar Network, the Knight Foundation and other grantmakers.

Many of the fellows are gathered at long side tables, tapping away on laptops. This year there are 27 fellows, assigned to one of eight cities in teams of three or four. And worth noting is that, name aside, they’re not exclusively software programmers. Past and present fellows have included designers, urban planners, policy people and those with other skill sets. (Also worth noting: Just more than a third of Code’s fellows have been female. While there have certainly been female “coders,” many female Coders have been designers.)

The fellows serve for 11 months and receive a modest salary plus some perks, including a sharp blue-with-white-piping track jacket that brands them as part of the Code for America select. “They live on $35,000 a year,” Pahlka jokes, “so we have to feed them and clothe them.” One additional kindness is that during their one-month “February residency” on site in their city, they don’t pay for housing. The New York City fellows, to pick one squad, ended shacking up Real World-style in an apartment in the East Village, the reference apt if only for the fact that the two men and one woman hardly knew each other before the experience. When I sit down with the trio on the waning hours of their New York City immersion, they seem to regard it all as a whimsical adventure of a lifetime they’re happy to have over.

By early March, the fellows reconvene back in San Francisco. Ensconced in the Storek Building, they’re ready to enter the next phase: Actually building some sort of software of use to the city that hired them. The clock’s ticking.

Maybe you’ve seen the episode of 30 Rock where Jack walks into his new Washington job. He’s thrilled to finally find himself in the seat of power. Except, his ceiling’s leaking. His boss assures him the issue has been studied and that the ceiling is, in fact, not leaking. One way of picturing Code for America’s headquarters is as a rejection of Jack Donaghy’s Washington.

In Code’s first two cycles, Pahlka admits, what the fellows came up with tended to follow a script of limited scope. Step one: “Go find some data that people care about.” Step two: “Work with bureaucrats to explain to them why this data could be made more useful.” Step three: “Integrate it with other data.” Step four: “Create a simple and easy-to-use interface.” And, finally, step five: “Work with constituents to get it out there.”

Simple enough, and the process produced some neat apps. But there’s actually much buried in that formula. For one thing, it makes plain that Coders aren’t traditional IT consultants handed a punch list of hoped-for tech; ‘spec’ isn’t really in their vocabulary. They’re free to let their minds wander, hunting for problems, spitballing solutions. (Cities generally give fellows an idea of the area of government they’d like them to tackle, but they tend to be broad fields rather than specific challenges.) At the same time, though, they’re rather different from another brand of freethinkers often called in to figure out how organizations could work better. Unlike management consultants, Coders can’t get away with writing a diagnostic report and leaving the next steps to some other poor soul. They’ve got to build.

“That might really be the whole thing,” says Joel Mahoney, a year-one fellow in Boston who went on to serve for a time as a staff strategist and “evangelist” for Code for America. “When you can look at the screen and click on it and see how it works, that’s a much different thing than just talking about ideas.” Coders’ apps are a manifestation of the Lean philosophy of rapid iteration and user-centric design, existing as a way of conveying that approach without having to utter the words “philosophy” or “iteration” or “user-centric.” It’s not a new notion in tech. Steve Jobs used to talk about the power of the early PC as having less to do with it being a tool for practical applications than it did with suddenly having “a mirror for your thought process” sitting right there in your office. Instead of an Apple I, though, it’s an app for some city service.

The end result has been a small stable of fairly simple apps that cities might never have thought they needed, but turned out to be fairly useful and occasionally, well, transformative.

Honolulu Answers, for example, involved the creation of something like a parallel universe of the city’s clunky official website. The fellows used Google Analytics to figure out what locals were searching for, and often failing to find, on Honolulu.gov. The held “write-a-thons,” asking both city officials and everyday citizens to help draft plain-language entries on the most popular topics. Taken together, the Bing-like site provides a glimpse of what government could look and feel like if it weren’t so wedded to looking and feeling like government. On the original city site, someone searching for help with unwanted wild roosters on their property were treated to, no kidding, a decade-old city proclamation on the topic. On the Honolulu Answers micro-site, you get information you can use, and not much else: “Call Animal Haven at (808) 799-7791.” Done and done.

And then there’s Boston’s Discover BPS, a way of navigating the city’s complex maze of public school options. Says Jacob, “a lot of parents said that the way the old website worked is demoralizing.”

The gee-whiz factor is significant, but play with these apps and soon enough you start to wonder how sustainable they will be when subject to the daily grind of city bureaucracy.

Philip Kovacs is an associate professor of education at the University of Alabama at Huntsville, but I’ve rung him up because he’s one of the most vociferous critics of Code for America’s inspiration, Teach for America. Kovacs argues that, like TfA, CfA sounds great at first. But he’s got big worries. He warns that the “for America” approach threatens to do damage to the cities that play host to the programs, a risk obscured by the fact that they seem to offer short-term solutions to pressing problems. With Code in particular, Kovacs wonders if the “year of service” model neglects to consider that coders, and Coders, are getting a lot out of the deal, too — especially now that Code and outside funders are paying more attention to things like the Google-funded Incubator, Code for America’s new in-house hothouse for start-ups built on fellow-year projects, and an “Accelerator” for semi-established civic tech companies originating from outside the fellowship program.

“It’s almost a kind of techno-colonialism,” Kovacs says. “‘We’re going to help the natives, and we’re going to take some spice back with us, because, hey, we’re doing good work here.’” Or think of it another way. You’ve heard of corporate welfare, the idea that government is assuming making of the risks of entrepreneurialism while private interests stand to gain much of the benefit. Call it “coder welfare,” perhaps. Kovacs offers a solution. If an app finds an afterlife in the form of a start-up or commercial product, maybe the fellows should kick 5 percent back to the city that served as their laboratory. That, he says, offering a twist on Code for America’s motto, would really be “a new kind of public service.”

Kovacs acknowledges that he’s being dramatic to make a point. But it’s a dynamic familiar in New Orleans. After Code’s fellowship there, city Chief Information Officer Allen Square says he has struggled to figure out a long-term plan for keeping up the blight-tracking app the fellows had built.

Code’s fellows, Square argues, did New Orleans a major service. Indeed, the stand-alone BlightStatus app is given prime placement on the city’s official website. (Coders regularly build in programming languages that are considered more flexible and more fun than governments tend to build. That’s one reason their work often gets bolted on the side of existing municipal sites, rather than integrated into them.) Square even tried to get five of his six fellows to stay on in the city. But some of them instead decided to return to the Incubator. New Orleans is in discussions with the new post-Code enterprise, branded as Civic Industries. But the arrangement seems a little off to Square.

“I do believe,” he says, “that because of the sweat that I put in — I figured out what was the best usability, the best look and feel — our cost of maintenance and support should be different from what other cities would pay.” Square says that cities in the program would benefit from knowing up front what the nature of their relationship will be after the year is up.

Everything the fellows build is open-source, Code points out in response. The beauty of that is that anyone can take the code they’ve cobbled together and build their own version. Of course if New Orleans had been ready to build its own BlightStatus, there wouldn’t have really been a practical need for Code in the first place.

A Year in the Coding Life

In the waning hours of their month of on-location investigation, I’d sat down with the three 2013 New York City fellows in the Bikini Bar, a Tribeca surf shop-slash-coffee shop that served as something of a home base for them that February. Tiki music played, and they explained some of what they’d learned in their deep dive into the city’s criminal justice system, the assigned topic of their Code for America work. For one thing, New York City is still remarkably analog. “It’s a lot of paper,” says Ezra Spier, a 26-year-old web developer. For another, not knowing, admittedly, very much about either criminal justice or New York City government was freeing. “We can go and say, ‘Hey, what you’re doing is cray-zee,’” Spier says.

Just by doing the meeting-and-greeting that characterizes a February residency, the Code for America fellows are doing something unique in the municipal government arena. “The majority of people in local government,” Jacob puts it, “don’t talk to people.”

What happens in practice is that the fellows end up as something like the plus-one at someone else’s company holiday party. He’s invited, which is validating. But he can still get away with saying a lot that employees can’t. In Macon, for example, a group called the College Hill Alliance has worked to change that area’s reputation as an unsafe place. Crime data would help. But, says Nadia Osman, the group’s director of revitalization, they can’t seem to get it from the police department in any useable form. But Code’s 2012 fellows were able to go to the Macon PD, Osman recalls, and say, “Okay, here’s how to fix it. And if you want to, here’s how you can keep it working in the future.”

“That,” Osman says, “comes from a much better place than our office calling and saying, ‘Why is it so difficult for you to give us this information?’”

That said, coming in from the outside can pose an enormous challenge. Is it really possible to narrow in on one juicy, easily handled project when everything, from the way city government works to the nuances of, say, criminal justice, are new to you? Not without some struggle, it seems. When I sat down to debrief the New York City fellows after their month in situ, Spier told me that the three were really looking forward to the next several days in San Francisco. That’s when the fellows would chew through all that they had learned and come up with a plan of attack for building their apps. “Next week is going to be really interesting for us,” he said, looking hopeful.

But when I meet up with Spier five weeks later, it’s clear they’re still processing. Their app plan is very much in flux, and the amount of work still to be done is clearly creating stress. In truth, it wasn’t much of a surprise. Pahlka tells horror stories about a 2012 fellow who could regularly be found bawling in the bathroom and elsewhere, posing questions like, “Why do you people think I can do this?”

Some of the tension can come from the ‘team of fellows’ model Code employs. “It’s really tough,” says Joel Mahoney, the former fellow turned former staffer, “to ask three fellows to form a functional start-up team in a short amount of time, never having known each other.”

Indeed, fellows are told from the get-go that it’s ‘their fellowship,’ but it only seems to dawn on them as time goes on that (1) Code for America means it and (2) uh-oh. Neither the city nor Code is really functioning in a boss-like capacity. While that’s freeing, it also means that fellows don’t have much of a backstop. If something breaks, it’s on them to fix it. If, as it often happens, someone in a city is hesitant to hand over key data to these newly arrived fellows, best of luck to you.

Adding to that is the fact that the core step of finding people to test out their apps might be unfamiliar to the Coders. One thing separating coding for the government from commercial coding is that, in the former, you don’t have the luxury of hand-selecting your customers. Civic coders can’t stop at beta-testing their tools on their co-workers. To get a broad, representative base of users, they’ve got to do more legwork. “Users” = “citizens,” young and old, rich and poor. It’s not unheard of in the for-profit world for programmers to beta test their inventions on their suitemates. Civic coders need to work a bit harder, with success often boiling down to old-fashioned community organizing.

And knocking on virtual doors isn’t always easy. The number of people who actually use Adopt-a-Hydrant, the Boston app by which citizens can claim fire pumps they pledge to dig out during blizzards, “is fairly low,” says Jacob. He’s not surprised. “Simply giving people more information,” he says, “isn’t good enough.” But doing more requires resources — and timed luck — that fellows might not have. “We ran out of time and money,” Jacob says about getting buy-in on Boston’s shovel-out app. “And we really didn’t have any snow.”

Back in San Francisco in April, Team Louisville is struggling. The Coders are due back in Kentucky soon to test-drive their app, and they’re yet to build a working prototype. They’ve got one thing nailed down — they want to do a text-based system for notifying defendants when they’ve got court dates coming up. One hitch, though, is that they’ve yet to convince Jefferson County, which holds much of the cache of court records, that the proposed tool is worth releasing data for.

They’re unbowed, they say. If getting the data “involves us sitting in a courtroom and writing down people’s court dates and phone numbers,” says Laura Meixell, a 20-something fellow on the Louisville team, “so be it.” Their struggle is hardly rare. Inside government especially, having a hold on data can be a real source of power. But holding data closely can also be an admirable protection of the public’s trust. So is it curmudgeonliness or good stewardship? Fellows, Pahlka says, are encouraged to “push, but from a place of empathy and respect.”

If that sounds like a tough way to build software, it can be. But there’s an ace in Code’s pocket: Cities today often aren’t all that concerned about the apps they end up getting.

That’s a good thing, because otherwise the math can sometimes be tough. Dan Melton, that “employee number two,” distills a complaint heard especially in Code’s early days. “We’re paying you $250,000,” as that was the price back then, “and we got half an app out of it.”

Matthew Esquibel, technology director for Austin, Texas — a 2012 Code for America city — explains why that calculation doesn’t capture the big picture. “It’s not what they end up producing,” he says. “It’s demonstrating the process by which they produce it.” Amanda Deaton, a Macon city official, echoes the sentiment: “If I was looking for a product, I would probably be disappointed. But if I was looking for culture change, it was really, really worth it.”

And cities needn’t even wait the full 11 months to experience it. Code fellows are taught to look for “quick wins,” opportunities to demonstrate, through a bit of easily attainable shock and awe, the power of their approach. In Macon, the fellows were able to fairly quickly put together a map, built on an open-source platform available to all, showing neighborhood code violations — things like grass being too high, roofs in disrepair, trash in a yard — going all the way back to 1997. It wasn’t that heavy of a lift, but on the ground it still presented as a bit of a marvel. “I never knew,” says Osman of the College Hill Alliance, “this data could be pulled together so quickly.”

And those fast-and-dirty accomplishments can be enough to get people questioning bureaucratic processes that are no longer working very well for them.

Take Boston. The fellows, Mahoney says, had an idea for building a sort of “app store” for a range of education-based apps, but some in government were uneasy about handing over the data, considering that much of it necessarily had to do with children. So they cobbled together Where’s My School Bus?, both to show that the data could indeed be treated with respect and, per Mahoney, “to demonstrate to the city what could happen if they loosen up a little.” This wasn’t some months-long study or carefully crafted proof of concept. “Maybe ‘throwaway’ is too much,” Mahoney says, “but we built it over the weekend.”

But it did the trick. The more ambitious Discover BPS site followed. The city hired Mahoney, in fact, as a contractor to continue work on it. Mahoney has also been hired to rebuild OpenCounter, a slick site that streamlines the process of opening a business in Santa Cruz. And Boston is taking steps to refine the “throwaway” Where’s My School Bus?

Coders see those afterlives of apps as successes, but it points to a challenging fact: Cobbling together a quick app and sustaining useful software aren’t, in the end, the same thing.

Honed Code Versus a Sprawling Universe

In fact, asking what happens to the apps post-fellows leads us to a critical, if behind-the-scenes, debate about the future of Code for America. It boils down to whether the non-profit will be most successful if it focuses on refining its often rough-hewn apps or, instead, what seems to be its raison d‘être circa 2013: Rallying others around the government innovation flag.

Melton says that he gets why Code has, in recent times, focused on acting as a guide through the Gov 2.0 landscape or cheerleader of the burgeoning civic tech movement instead of, as he sees it, devoting itself to crafting truly awesome, innately powerful apps. For one thing, he says, one of the lessons from Code’s early days was that the latter was difficult to scale. Code’s inaugural model was putting a handful of savvy technologists inside city governments and then expecting the tools they build to themselves trigger meaningful, big-picture change. It was a lot to ask. Some city governments have more employees than Facebook and Google. And as terrific as the fellows might be, come on. (A useful reality check: Teach for America has 6,000 deployed teachers, the Peace Corps 8,000 active volunteers. There are, at the moment, 74 former or present Code for America fellows.)

Add in that “reuse,” or recycling code from city to city, is one of Code’s underlying functional promises, and the burden grows. In the heat of the fellowship, fellows are expected to document in open-source fashion the software they’ve built. But, protests Melton, “With what time? With what expertise?”

So, somewhat out of necessity, the fellows began to be seen more as evangelists than hard-core technologists. Melton gets why that happened — why, in addition to the Incubator and the Accelerator, there’s much focus on establishing a heretofore-struggling Peer Network of in-government allies and a “Brigade” of local volunteers. But he thinks that while producing great code might be more resource intensive and more of an uphill climb than expanding the civic tech universe, it ultimately comes with a bigger payoff. “Excitement only gets us so far,” Melton says, “and systematic change is what we’re trying to accomplish.”

But as others see it, it takes a village to build great apps.

From that perspective, asking generally young people with generally no civic experience to pull off powerful software in short order is, indeed, too much to ask. (Code, to be fair, dedicates the first month of the fellowship — the “January Institute,” in Code parlance — to training them in the ways of government.) No app is an island, and building technologies that meet the needs of real-live human citizens requires attention to the structural challenges that are going to greet any new bit of government IT.

“Excitement only gets us so far… and systematic change is what we’re trying to accomplish.”

Broadening the universe, argues Boston’s Nigel Jacob, expands the chances that the apps built by fellows, and other developers involved with civic tech, will actually change the way cities meet the needs of their citizens. “We’re still in the claim-making phase of civic innovation,” Jacob says. “We’re saying things and really hoping they’re true.” He echoes Melton’s complaints that Code’s technical repositories lag; for that matter, so does Pahlka. “They’re saying to us, ‘Can you get your house in order, technically?’” she says of those in the extended Code world. But here, too, Jacob sees potential in the capacity of non-Coders. Cities, he says, can help document the full “palette” trying to make use of new apps.

Moreover, the universe-building approach that, by all appearances, Code has adopted in recent days can prove appealing to the fellows that are at the core of the program. “It’s encouraging to see that there’s life after the fellowship in this world,” says Shaunak Kashyap, a Louisville fellow, pointing up to where several of last years New Orleans’ fellows, reconstituted as Civic Industries, sit. It’s an important point. Code fellows might not finish the year with much money. But they might have the seed ready to grow into a company.

But whichever focus Code puts on its work, Pahlka says she’s prepared for the calls to come soon asking for proof that the approach works. “Okay, what have you really done?” she says, playing out what she expects. “We’re constantly aware of that. What will we have done in five or 10 years? Will it be significant?”

With that in mind, she and others have their eyes on both New York City and Louisville, where the fellows are hard at work trying to figure out if an app might help crack open the criminal justice system. That sort of win can push the code versus coders debate a bit further into the background. And here, the fellows have an ally who knows her way around the inside: The former attorney general of New Jersey.

Beyond Decoration

“We need a guide,” Pahlka says. “We can’t do enormous interventions. We do small interventions, so we have to find small interventions that have a big impact.” Criminal justice, she argues, is a field that Code has yet to tackle, but it’s one where the stakes are high and there’s potential to create change. Right about the time that Pahlka and Greenhill were brainstorming over drinks in Flagstaff, Anne Milgram was serving as the lead criminal justice official in the state of New Jersey. She was dismayed by the system she found. “Yo, we had a robbery last week,” says Milgram, holding up an invisible sticky note and play-acting how some of her police department meetings went down.

After joining up with the Houston-based Arnold Foundation, Milgram approached Pahlka about working on some apps that might add tools and data to a system that affects millions of people and costs billions of dollars. “It seemed really politically charged and difficult,” Pahlka remembers of her initial reaction. But she soon got one board, and both Louisville and New York City agreed to serve as test beds for the new approach. That’s how two teams of three fellows have found themselves in Kentucky and New York this year, attempting to find ways to, well, hack the criminal justice system.

Team Louisville, you’ll remember, is struggling to find data. But Team New York, when I visit San Francisco in April, is still struggling to find a project. One early idea for a notification system for restraining orders in domestic violence cases was ditched when it turned out an upstate group of sheriffs was working on something similar. The team is now looking at warrant backlogs, but has run into some of the same trouble as its Kentucky counterpart: The desire for data hasn’t reached the tops of officials’ to-do lists. Failure is in some ways an option. “You don’t expect the first start-up in a space to win,” O’Reilly says. But the Louisville and New York fellows seem to know that eyes are on them.

“Innovation decoration” is a phrase that O’Reilly has picked up and defines, for him, what Code doesn’t want to be. “It’s not innovation decoration if you get people released without spending a couple weeks in jail for a trivial offense when they’re no harm to society,” O’Reilly says, referring to the criminal justice work.

In this, Code might be on the cutting edge. Where it is headed, however, is of a piece with where many hope the wider civic tech movement will go. Civic innovation, contends former Indianapolis mayor and former New York deputy mayor Stephen Goldsmith, needs to move away from “reporting potholes” and toward being “deeply ingrained in the management of government.”

Focusing attention on criminal justice is a start, as many see it, but there’s something else at work that makes the Milgram-led work significant. (The Arnold Foundation, for the record, is supplying a considerable amount of funding for the projects, some $300,000 in Louisville alone.) It’s useful here to return to that point of comparison: Teach for America. In its 20-plus years, TfA has held on to the idea that the quickest, cleanest path for disrupting troubled systems is from the outside in. And that’s certainly where Code for America started. As Pahlka puts it, “initially, the thesis was that practices and approaches from the tech industry could change government.” There was a definite technologists-to-the-rescue vibe to the early days.

That’s changed — with good reason, according to Pahlka. It was insulting to those already working to make change in government and it was, she says, “naive” about the real practice of government in the 21st-century United States. “They need Code for America,” Pahlka says about the civil servants she’s come across, “in only the sense that they need a tool for doing what they’re already trying to do.” Today, “instead of us being the doctor fixing them, the metaphor is more like we are a tool in their hands.”

“It’s not innovation decoration if you get people released without spending a couple weeks in jail for a trivial offense when they’re no harm to society.”

Other critiques arise from looking at Code through a Teach-for-America lens. Kovacs, the University of Alabama professor, argues that both programs risk robbing communities of workers who might develop, over time, useful experience and layered value. What’s more, Kovacs disputes the notion that Code’s existence is called for due to a shortage of people to fill science and tech jobs — or the idea that people aren’t going to Tucson to build apps. “The research says otherwise,” says Kovacs.

But officials from cities with widely different experiences agree that it’s challenging to get truly skilled programmers clamoring to join the ranks of civil service. Sure, the Coders don’t spend all that much time on the ground. (Most of their year takes place in San Francisco, where they can benefit from not only each other’s wisdom by the local tech scene; Jack Dorsey, co-founder of Twitter, stopped by recently to chat with the fellows.) But it’s something, and as Mahoney says, it means cities “can get people inside City Hall that they’re not going to be able to hire.” That’s true in Macon, according to Amanda Deaton. The Georgia city simply lacks the local tech talent pool from which to draw. It’s also true, it seems, in a city on the opposite end of the scale. Austin might be the Silicon Valley of the South. But that means, Matthew Esquibel says, “we have so much competition, we find it hard to attract the same kind of talent.”

The Metrics of Reinvention

“The model shouldn’t be cooked yet,” argues Nigel Jacob. It doesn’t seem to be. Code for America is itself iterating. That sort of constant rethinking, one can argue, calls for openness to new metrics of success.

At least, that’s what Eric Reis calls for in The Lean Startup, using the term “innovation accounting.” But the idea has support from a close observer of another program for emissaries into foreign cultures to spread new ways of doing things. The Peace Corps been around since the early 1960s and, argues Brookings Institution senior fellow Lex Reiffel, who has studied and served in the organization, we don’t even have great ways of understanding its value. “There are elements of human interaction,” Reiffel says, “that we don’t yet have the tools to measure.”

Pahlka, for her part, sees “signs that the institutions are different” in the wake of a Code for America engagement. In fact, a few cities have, post-Code, opened innovation offices of their own. “From our department heads down to our line employees,” Macon’s Deaton says, “they’re talking about new ways of doing things.” Mahoney, the Boston fellow, recalls running into a Boston school official after his fellowship had ended. She didn’t remember him and didn’t remember Code for America, but knew about Discover BPS. “You changed the way we related to parents,” he recalls her saying.

Mahoney also points to a new discussion in Boston about using algorithms to drive school assignments, in which he sees echoes of the Code approach. But, he adds, “detractors will admittedly be critical of these vague notions of culture change.”

It’s true. When Greenhill, now assistant to the Tucson city manager, argues that “it’s important to have systems-thinking people in government early in their careers, whether or not the person who does this work for a year becomes a city IT professional or lifelong teacher,” Kovacs scoffs at it as the same sort of fuzzy, feel-good thinking that has let Teach for America go unexamined. But Code has likely bought itself some time and space to define what, exactly, it’s bringing to the table, at least until the results from its experiments in Louisville and New York City are in. No doubt, its ambitions are great. At a salon last July, Pahlka described “shrieking with delight” as she read The Gardens of Democracy, a book positing that government is not a machine, but a garden — complex, adaptable and requiring of our care. “We’re in this time,” says Eric Liu, one of the book’s authors, “when the relationship between citizen and government is in a fertile place for reimagining.” Pahlka says that the notion explained to her “the why of what we do.”

Greenhill, whose desire to have “fancy developers” come “write apps for Tucson” sparked Code for America in the first place, now sees in Code’s potential something that, perhaps, is even more basic than the “garden” framing.

“With the changes that we’ve seen in cities over the last several years,” he says, “the way cities have been funded has changed fundamentally, therefore there is a need for cities to adopt new models and find new ways of doing business.” Some cities, as he sees it, are holding out for the day when they can return to the old ways of doing things. On the other hand, he says, the technologists he’s met are captured by “an optimism that the future is going to be better” than things ever were.

Maybe, then, that’s when Code for America can claim success: When marked by “an optimism that the future is going to be better” is something that can just as easily be said about cities across the United States as it can about Silicon Valley.

Our features are made possible with generous support from The Ford Foundation.

Nancy Scola is a Washington, DC-based journalist whose work tends to focus on the intersections of technology, politics, and public policy. Shortly after returning from Havana she started as a tech reporter at POLITICO.

Hawk Krall is an illustrator, cartoonist and former line cook known for food paintings that have appeared in magazines, restaurants and hot dog stands all over the world. In Philadelphia, he is known for his “factually creative” drawings and paintings of the city’s neighborhoods, most recently exhibited at Space 1026.