The Tragedy of the Directories: Towards the Co-Production of a 21st Century Safety Net

By Greg Bloom, instigator of the Open Referral Initiative [http://openreferral.org]

drafted for The Maintainers, at Stevens Institute of Technology, April 8th 2016

Abstract:

Governments, non-profits, and other civic entities offer a vast array of resources to people in need. However, this 'safety net' — which includes diverse sectors of health, human, and social services — isn’t shaped by the same forces that drive the flow of information in a typical marketplace. This is because, broadly speaking, when an institution isn’t paid by the people who consume its services, it may lack incentives to attract more consumers. Given this and other perverse incentives, it is a tremendous challenge to keep track of which institutions provide what services for whom.

For this specific purpose, an entire field of research and activities known as ‘information and referral’ (I&R) emerged in the mid-20th century. Having evolved from library card catalogues to enterprise call-centers, I&R now faces the same kinds of disruptive forces that have already transformed almost every other information-based sector. Over a dozen referral service startups have emerged in recent years, typically promoting themselves as the ‘Yelp for Social Services’ (or Uber, etc). These startups are typically web-based, often for-profit, place-less, and designed to meet the needs of particular types of users; they are competing with the legacies of non-profit, localized, public-service referral providers. Yet this wave of entrepreneurship has not yet solved the core problem around which the I&R sector formed: each referral provider still manages their own directory, in incompatible, redundant data silos. Without intervention, such innovation will intensify patterns of fragmentation and privatization.

If you're interested in an institutional diagnosis and hypothetical prescription for this problem, read here. For a technical analysis, read here. Narratively-minded readers, read on.


Introduction: Shitty Google Results and the Matter of Justice

My sister recently became a public defender, which has given me an entirely new vantage point from which to gaze upon our failing safety net.

Public defenders serve indigent clients who face criminal charges. Most of whom are people of color. In other words, those people for whom mainstream America has the least sympathy. Yet who are nevertheless rightfully entitled to legal representation.

To truly receive effective representation, however, these defendants don't just need legal counsel; they often also need food, places to stay, childcare, even transportation, etc. If they don't get these things, it's difficult for their lawyers to effectively defend their interests in court.

As one of my sister's colleagues recently explained to me, this means that a critical layer of her job is essentially social work: helping people help themselves. She was not trained to be a social worker. However, as a lawyer, she has been trained to sort through large quantities of unwieldy information. So every day or so, when she sets aside her legal casefiles and turns to Google to look for other sources of help for her clients, she is surprised to find herself daunted.

"I normally completely trust the internet. If I need something, I can look for it and find it -- no problem. Except for this stuff: information about shelters. Information about mental health counseling. The things needed by people who are in actual trouble. When it comes to that stuff, Google is a ghost town.”

Why?

This story is familiar to me. I've heard hundreds like it -- from librarians, from teachers, from nurses and doctors, even from cops. The context varies but the pattern is the same: "it's hard to know where to send people to get help."

Perhaps you've heard versions of this story as well.

If not through personal experience, perhaps you’ve attended some kind of 'civic hackathon' or other such events dedicated to 'technology for social good' -- in which case, you’ve probably heard some version of this story along with some version of the following proposal:

“So we’re going to build an app that helps people in need find all the programs that are available to them.”

Sometimes the app is specific to certain kinds of people — veterans, people experiencing homelessness, family members of people entangled in the criminal justice system, etc. Sometimes it’s specific to certain kinds of resources — job training, food, shelter, etc. But the gist is typically the same:

“We need ‘a clearinghouse of community resources.’ A ‘one-stop shop for human services.’ A ‘Yelp for help.”

Perhaps you're even familiar with some of the many things that subsequently go wrong with these projects. Or, perhaps, you've got one of these ideas yourself, and you're already working on wireframes, a pitch deck, and a capitalization table.

In either case, in this essay I'm going to share some stories that suggest why there isn’t, perhaps won’t be, and maybe even shouldn’t be, a ‘Yelp for social services’ — so that we can begin to imagine how else we might address this wicked problem.

My story

As the staffer responsible for 'communications' at one of Washington DC's pre-eminent anti-poverty service organizations, I came across this issue — "how do people know where to go to get help?" — in many different forms.

First and foremost, I'd regularly get phone calls from people who were looking for basic information about the services we provided and how to access them.

When fielding such calls, I realized that even I found it to be a struggle to keep track of important details about my own organization's services. For example, a response might look like:

"The food pantry is open 10am to 4pm. But if you want to become a new client of the medical clinic, you have to call the clinic right at 9am on the day before you want to make a visit. And if you want to schedule an appointment in our legal clinic, you'll have to go through intake -- which is from 1-4p on Monday and Wednesday at our Northwest Center, or 9a-noon Tuesday and Thursday at our Southeast Center. Oh wait, maybe that's changed…"

These things changed all the time: funding streams would come and go, staff would come and go, programs would be re-organized. Any such occurrence would affect operational details about how people actually accessed those services. Even within the organization, such announcements would be made ad hoc -- by email, or a meeting announcement, or a flyer. Whatever: people would figure it out. This constant shuffling of programs and information was just one layer of frustration among many that afflict people on both sides of the service provision relationship.

Over time, I noticed another pattern of requests for our information: calls (or just automatic emails) from groups who maintained directories and wanted me to update our organization's record for them. Some of these directories were printed booklets; some were for official-sounding 'hotlines.'

The first time I received this request, I diligently complied.

I had to double-check information from our website (about four departments and associated sub-programs, at two different facilities) and then edit it to match the way they asked their questions; altogether, it took me a couple of hours.[1] A couple months later I received another request from a different directory, and I complied with this too; it still took me an hour to check all of that information was still correct .

When I fielded a third request for us to update our information in yet another directory, I asked our management and staff about it. What I heard back surprised me:

Maybe we don't need to bother?

“These directories are all over the place,” a longtime staffer told me with exasperation. There was the city’s 2-1-1 call-center service, but our social workers said it hadn’t been updated in years. There were various printed directories and websites — maintained by volunteers at a church or a library or their peers at other more narrowly-focused organizations like the domestic violence hotline or the housing assistance guide. Nothing was particularly comprehensive or reliable.

“New ones are always popping up, and none of them stay updated -- so I just stopped trying to stay on top of all of it.”

I want to reiterate here that this organization is really one of the best of its field. Our staff was known to be exceptionally diligent and compassionate. This disinterest in the request to update our own information in all of these directories was not indicative of irresponsibility, but rather of sophisticated despair in the state of the systems that are supposed to help people find help. We knew that most other organizations wouldn't comply with the requests either. We already had more people coming to us than we could reasonably serve. There simply wasn't a good reason for me to spend my time updating all these various directories; that time could instead be spent doing things that might raise money so that we could provide more services in the first place.

In the meantime, people would just figure it out. They'd keep calling our number or just walking in our door.

It wasn't a great situation, but in the broader scheme of things — given the fact that there just weren't enough resources out there to meet people's needs in the first place — it all seemed like a secondary problem.

The Obscurity of the Safety Net

But I would eventually see this problem rear up yet again -- and it would then seem like less of a persistent nuisance and more of a systemic failure.

During the midst of the city's fiscal crisis in the early phases of the Great Recession, the mayor proposed to cut the budget for many agencies — with the large majority of the cuts coming from safety net programs. I got involved with the campaign to push back against these cuts.

Advocates wanted to sound an alarm that this kind of austerity, in the midst of a severe recession, would increase the poverty rate in our community. Yet we had no way to describe how many (and what kinds of) safety net services already actually existed. We lacked a baseline of data about who did what for whom; instead we had anecdata and broad generalities about the gaps between the needs in our community and the resources available to meet them.

Over the course of the budget season, we managed to prevent some of the worst of the budget cuts. Yet the issue remained abstruse: we knew that more people were facing poverty, and we knew that safety net services were being scaled back, but we couldn’t be specific about the extent of the latter, so we were unable to be specific about what kinds of investment were really needed in response to the crisis.

And so the full wickedness of the community resource directory data problem revealed itself to me: Because there isn't a single reliable answer to the question of "which organizations provide what services to whom?" — and instead there are a bunch of fragmented, redundant, ad hoc directories which all add up to far less than the sum of their parts -- the result is not just discouraging for people who need help, and not just frustrating for people trying to provide help. It's also debilitating for people who are trying to understand the state of our safety net, in order to advocate for its improvement.

DIY Databasing

My story has a curious bright spot. That bright spot actually someone else's story though. As it turns out, my organization was maintaining a resource directory of many if not most of the services available to people in our community -- and I discovered that many other organizations believed ours to be the best source of information.

That was yet another type of call I would receive about this information: from other organizations (non-profits, academics, even technology vendors) who wanted access to our database. For these calls, I referred people to a social worker named Stacy, our resource directory maven.

Stacey has been working at my organization for about a decade. Stacey is small, calm, discerning and patient. She hates to send people out our doors without some kind of helpful information about where they can go next.

Over time, Stacey developed an intimate sense of which organizations would be able to help which people in what situations. While calling around, she'd regularly find new information about these organizations.

So a few years into her job, in between a caseload of thousands of clients on one hand and two kids at home on the other, Stacey taught herself how to use Microsoft Access. She designed a database in which she (and the social-workers-in-training whom she oversaw) diligently kept track of this knowledge. They update the database with new information every week.

By now, Stacey has a personal relationship with people at many of the organizations listed in the database. This makes a big difference in their willingness to share information. Often, she’s one of the few people that organizations reach out to regarding changes to their own program details. In contrast to the automatic emails and cold phone calls that I learned to dismiss, many people at organizations across DC want to make sure that Stacey knows what they did, because they personally trust her to effectively deliver that information to thousands of people.

In turn, Stacey was willing to share this aggregated data with others who asked. "If someone else can benefit from all this hassle," she said, "well that would be great."

Of course, Stacey knew that this process of database duplication yielded a bunch of decaying directories on websites and in other organizations' servers around the city.

There had to be a better way, I said.

Stacey sighed and shrugged. Perhaps. But she doesn't have the time to try to figure it out. She'll just keep doing what basically works.

The Field of Information and Referral

What about those call centers, though? The ones that kept calling me to ask for our information to update their database -- what's their story? Eventually I met enough of them that I feel comfortable sharing with you a (fictional) story that represents a composite description that may be quite similar to the person who manages the database maintained by the 'resource referral help line' in your own community.

Let's call her Susie. (They are almost all women). Susie is the resource database administrator works for the call center (which is probably run by a non-profit organization) that operates the 2-1-1 phone service in your county. Susie has a degree in information management, and she also has some experience volunteering at her local crisis hotline. Keeping track of all this information is tedious, but she feels like it's an important service for her community.

Susie uses an information system that is the bane of her existence. It's software purchased from a distant vendor who doesn't respond very quickly or proactively to her issues -- but her organization doesn't have the budget to try anything else. They struggle enough already to win contracts from the county, the state, the federal government, to get grants from local foundations -- whatever they can get their hands on. They even compete against the other call centers for this funding. It's frustrating.

She spends her days poring through their resource directory database, calling down a list of about 400 organizations across their region -- from A to Z and back to A. She tries to make it through more than once each year.

Organizations have such elaborate criteria for who is eligible for what program, and how they're supposed to access these programs. This stuff changes all the time, and the websites aren't much help. She sometimes spends two or three hours on a given organization -- looking the scant details on its website, calling the number and figuring out who she really needs to talk to, and convincing the person that this is a valid and important request. Sometimes, when she's really built a relationship with an organization, they'll actually call her to let her know when something changes (or they'll submit an update on the 2-1-1's clunky webpage).

But even then she still has to check carefully, because organizations will say "we serve the entire community's needs" and list all kinds of services when she suspects that they're really just referring people right back out the door to those services. Getting clarity on these things takes time and care.

So when Susie hears about other organizations are maintaining their own databases -- and it seems like there are new ones that pop up every year -- it frustrates her. She's a professional who puts pride in her work. Why can't they just use our website? she wonders. Sure, their website is not pretty. It can be hard to search if you don't know precisely how to find what you're looking for. But if everyone would use it, maybe it would be easier for them to raise more money to make it better.

Susie regularly fields requests from other organizations who want access to her data. This is frustrating. People don't realize how hard it is to stay on top of this data, and they always just want to take her hard work and run off with it. Some groups are just parasitic, and benefit from her work without helping them raise the resources necessary to support it. So she hears all this talk about 'open data,' and she shakes her head. If everyone had easy access to her data, how would she keep getting paid to produce it?

Susie knows that something needs to change, but it just seems like the world is getting faster and less connected and she doesn't know what to do.

Whither Collective Impact?

Another composite story: consider the public health researcher.

She is working for a local foundation in support of a 'collective impact' initiative.  Through this initiative, a previously-uncoordinated network of nonprofits and other civic institutions is brought together to cooperate towards shared goals.  It's her job to corral dozens of stakeholders through a series of meetings and projects, to gather data along the way, to analyze and support stakeholders' analysis of this data.

When she facilitates the first of the group's 'deep dives,' she asks them to identify the key questions that they want to answer together. Two of the most popular questions entail information about community resources: "How can we be more effective at referring people to helpful services?" and also "What are the unmet needs in our community?"

She notes that the room tenses up when these questions arise. The people at the table have dealt with this issue before, and there seems to be reluctance to speak openly about it. "Can we get 2-1-1 to share their database with us?" someone asks. There are murmurs. "Who else collects this data?" Some names get written on the board. "Couldn't we just build our own resource map? We can get interns and college students to do the work." Someone sighs. Someone else explains that this has been tried before, "and it didn't go well."

Our researcher is confused and frustrated. After the meeting, someone emails her with an attached file and a request for this information to be withheld from the group. "Just so you get a sense of what we're dealing with."

Now she sees: It's been structured for a call center, which needs things that are pretty different from what she needs. Things like the difference between 'program' and 'service' -- both ambiguous phrases, but funders generally think in terms of 'programs' (funding streams) and this data seems to think in terms of 'services' (activities) yet it's not clear from the data which is which. Other things like funding sources are barely addressed.

If she could have real-time access to this data, maybe it would make sense to try to make refinements and additions to meet her needs. But it doesn't seem like the 2-1-1 wants to provide that. So maybe she should just get a bunch of interns and build her own…

Innovators and Dilemmas

One final story, also composite: this time about the entrepreneur.

Most people who manage community resource directories are women, working in fields that are mostly populated by people of color. But the entrepreneur is almost certainly male, and almost certainly white.

He received a business degree at an elite university, and he has participated in various 'innovation leadership programs.' He now has a well-paying job at a health IT startup, but he also wants to use his skills to do good. To give back to the community. While volunteering as a ‘business analyst’ at a local non-profit, and he was shocked at how rudimentary their tools are. He knows he can make something better.

So he puts together a whitepaper about the tool that he'll build: a one-stop-shop community resource locator. It'll provide a single place for all the community's resources. It will enable people to leave feedback about services. And it will enable pull reports about the best services in their communities.

He pitches the idea at a 'social good hackathon.' The crowd really loves it. A young engineer approaches him and says he thinks he could build a prototype within a few days. Great! Now he has a technical co-founder. They look on Susie’s 2-1-1 website, and right away they spot a dozen modern design elements and features that it lacks. So they go ahead and write a script to scrape this data from the 2-1-1's site; this takes about half a day.

They build a demo site the next day: a simple search field, big text, lots of white space, a map interface, sharing features, and all of the data. The categories are pretty messy, and searches turn up some confusing results. But it's a proof of concept.

They reach out to the email listed on the 2-1-1’s website, to talk about partnerships. Susie receives their email doesn't respond. (Susie doesn't know what to make of their request, but she can tell that they're not social workers by the look of how they presented their data. She doesn't know if what they've done is illegal, but she thinks it should be. She forwards their email to her boss, who says just ignore it and this group will eventually just fizzle out.)

They know an investor who's interested. He thinks that there are some health insurance companies, and some hospital networks, maybe even a city agency, who would pay serious money for this kind of tool. He'll make some introductions. They should talk capitalization. It's all very exciting.

The next day, they write a better script that automatically crawls the 2-1-1 site as well as three other sites, and cleans up the data automatically. They think they could arrange some algorithms to match this against Google Places and elsewhere. Pretty soon they think they could have the best data around. They're going to move fast, they're going to disrupt this market, and they're going to help a lot of people while making a good amount of money too. It's an exciting time to be alive.  

In conclusion: a co-produced world is possible

Yochai Benkler famously observed the phenomenon of 'commons-based peer-production,' in which networks of people collaborate on the development and maintenance of complex software and other digital media, despite having no formal relationship with each other (or even ever having met). This mode of production is entirely different from markets and firms, the two primary modes previously recognized by mainstream economics.  

And yet this mode was not, however, entirely new. It was not native to the Internet any more than it was discovered by Benkler. Commons-based peer-production is but a subtype of labor within a broader class identified by Elinor Ostrom simply as 'co-production.' (Shortly before her death in 2012, Ostrom's work on common pool resource management and co-production won the Nobel Prize in Economics.)

Ostrom first observed co-production in an unexpected context: community policing. Having noticed that the increasing size of police departments did not actually correlate with decreasing levels of crime (if anything, there was a slight inverse correlation), Ostrom studied incidences of low-crime neighborhoods with lower levels of police presence. There she discovered a telling pattern: in neighborhoods where police officers walked around their neighborhoods (rather than patrol in cars), they also tended to know more residents by name. They had relationships with people. And when police had relationships with people, the people were more actively involved in the process of keeping their neighborhoods safe. From simple eyes on the street to outright mediation, people who felt recognized and respected by their police officers became more involved in co-producing the conditions of order.

Ostrom and her emerging field observed this phenomenon of co-production in many different contexts. From lobster fisheries in Maine to rainforests in the Amazon, when people who use a resource are also involved in the process of producing that resource (or, as it were, the process of maintaining the conditions that sustain the resource), the resource tends to have a better shot at surviving the various pressures that otherwise threaten the resource through overuse.

This is to say, Garrett Hardin's infamous essay "The Tragedy of the Commons" was quite wrong. When Hardin conducted his thought experiment — about a pasture inevitably depleted by the uninhibited selfishness of farmers with their increasing number of grazing cattle —he mistook the concept of a ‘commons’ for an unwalled and unregulated open access resource pasture where nobody tries to communicate with anyone else about the rate of resource depletion. Hardin's thought experiment did not go so far as to conceive of the possibility of such a thing as co-production. It turns out that non-tragic commons very much exist, and often serve their users' interests much better than the corporately-privatized or government-regulated alternatives; in these commons, you'll probably find people working hard to communicate with each other in accountable ways about who should do what, where, and how -- and if that fails, what should happen as a result, etc.

In the realm of networked information technology, Benkler noticed a similar phenomenon: from Linux to Wikipedia, when certain kinds of digital resources are able to be not just used but re-produced and even adapted by their users, they tend to flourish to an extent that surpasses their proprietary counterparts.

Or, at least, it's possible. Ostrom would be the first to warn everyone: co-production can sometimes work out better than private production or regulated behavior, but it's no panacea, and it's not easy. It requires a lot of work in communication, rule-making, monitoring, mediation, rule-revising. In other words, the maintenance of trust and shared understanding.

So this, then, is perhaps our question for the future of information resources, such as community resource directory data, in which value must be produced by someone's costly labor, while many other someones stand to freely benefit from that production. In such circumstances, how can our energies towards *innovation* — towards developing new tools, new platforms, new means of realizing the great levelling and elevating promise of information technologies — be leveraged specifically for the production and maintenance of trust. How can digital tools enhance (rather than replace) systems of human relationship and accountability? What would be possible, if we expected our tools not to release us from work and its attendant concerns, but rather to enable us to do more work worth doing, together?

The very process of asking, and seeking answers to, these questions is itself a task suited for co-production. People, working together, can figure it out.


[1] Now, by the standards of our field, I think our website was quite good. However, the primary purpose of almost any non-profit's website is to generate donations. Toward that end, ours got results: it *was* one-click easy to give us a donation, or to subscribe to our email newsletters (which are designed prompt you to make a donation). But if you wanted to find out the intake hours for our medical clinic? That would take at least a couple of clicks through our site, if you knew where to click. So most people just called.