Open Source

Crowdsourcing Social Problems

Using distributed technology to tackle society's most intractable challenges.

|


engineroomblog / Foter.com / CC BY-SA

Recognize the distorted text to your left? That's a knock-off of CAPTCHA, the ingenious online system for verifying that you are a human user of a website and not some crawling bot. (The punny acronym is short for "Completely Automated Public Turing test to tell Computers and Humans Apart.") Developed more than a dozen years ago at Carnegie Mellon University, CAPTCHA has been a reliable firewall between sensitive databases and virus-like programs designed to pry into them.

Four years ago the computer scientist Luis Von Ahn, who helped develop the original technology, extended the same insight to the problem of accurately digitizing print books for online distribution. Ahn's reCAPTCHA, now owned by Google, uses the CAPTCHA interface to break up digitizing projects into two-word chunks of old scanned texts. Users-an estimated 10 percent of the world's population-are unknowingly helping to digitize around 100 million words a day, the equivalent of about 2.5 million books a year.

Ahn's latest crowdsourcing project allows people to learn a foreign language while simultaneously translating huge chunks of the Internet. Duolingo, launched in late 2011, teaches six languages (English, Spanish, French, Italian, German, Portuguese) by giving students short translations to complete based on their current language level. Every time one of these translations is completed successfully, a small part of the Web (say, Buzzfeed's English-language site) gets translated into the language in question.

In a TED talk, Ahn explained his approach to crowdsourcing with an historical analogy.

"If you look at humanity's large-scale achievements, these really big things that humanity has gotten together and done historically-like, for example, building the pyramids of Egypt or the Panama Canal or putting a man on the moon-there is a curious fact about them, and it is that they were all done with about the same number of people," he says. "It's weird; they were all done with about 100,000 people. And the reason for that is because, before the Internet, coordinating more than 100,000 people, let alone paying them, was essentially impossible. But now with the Internet, I've just shown you a project where we've gotten 750 million people to help us digitize human knowledge. So the question that motivates my research is, if we can put a man on the moon with 100,000, what can we do with 100 million?"

It doesn't get much bigger than digitizing human knowledge. Think of it. Technology and a clever business model allowing for the kind of large-scale coordination heretofore impossible in the annals of human history-without force, and much of it for free.

reCAPTCHA and Duolingo both represent a distinctly 21st-century form of distributed problem solving. These Internet-enabled approaches tend to be faster, far less expensive, and far more resilient than the heavyweight industrial-age methods of solving big social problems that we've grown accustomed to over the past century. They typically involve highly diverse resources-volunteer time, crowdfunding, the capabilities of multinational corporations, entrepreneurial capital, philanthropic funding-aligned around common objectives such as reducing congestion, providing safe drinking water, or promoting healthy living. Crowdsourcing offers not just a better way of doing things, but a radical challenge to the bureaucratic status quo.

Here are several ways public, private, and nonprofit organizations can use lightweight, distributed approaches to solve societal problems faster and cheaper than the existing sclerotic models.

Chunk the Problem

The genius of reCAPTCHA and Duolingo is that they divide labor into small increments, performed for free, often by people who are unaware of the project they're helping to complete. This strategy has wide public-policy applications, even in dealing with potholes.

In Boston, the city collects data on the driving habits of residents. Specifically, citizens volunteer to passively survey road conditions by opening an app called "Street Bump" during their daily commute. The resulting GPS data, combined with gyroscope readings, identifies potholes in time for intervention.

Boston's pothole problem might previously have required a small army of inspectors, managers, and relayed complaint calls. Now a citizen doesn't even have to report a problem herself. The local government can thus cheaply perform work that would otherwise rack up payroll. City officials Chris Osgood and Nigel Jacobs, the innovators who created Boston's Citizens Connect in 2009, call their approach microvolunteerism: empowering citizens to make small commitments to the public good, with a huge aggregate impact.

Microvolunteerism has proved effective the world over. The Kenyan slum of Kiberia needed maps. These would help citizens locate water sources and help officials plan future improvements. To map the slum, volunteers carried GPS units through Kiberia and marked landmarks such as water pumps and bathrooms.

Meanwhile, Finland's DigitalKoot project enlisted volunteers to digitize their own libraries by playing a computer game that challenged them to transcribe scans of antique manuscripts.

Governments can set up a microtasking platform, not just for citizen engagement but as a way to harness the knowledge and skills of public employees across multiple departments and agencies. If microtasking can work to connect people outside the "four walls" of an organization, think of its potential as a platform to connect people and conduct work inside an organization-even an organization as bureaucratic as government.

Change the Incentives

Wilmington, Delaware, was running out of landfill space. This wasn't surprising, considering the city's recycling rate of less than 20 percent. This is a common problem: Despite decades of public service campaigns, American recycling rates have hovered stubbornly around one-third, a far cry from Denmark's 69 percent.

In 2006, instead of continuing to spread its landfill costs among citizens, Wilmington decided to spread its landfill responsibilities. The city contracted with Recyclebank, a company that calculates, via RFID cards in trash bins, the volume that each family recycles. Recycling translated to points that customers could then redeem at local businesses.

Within months, Wilmington's recycling rate jumped to 65 percent, according to the city. The avoided landfill fees meant that Recyclebank was paying for itself. And city retailers churned up new business.

Instead of changing the culture, Wilmington simply changed the incentives. This is the premise of economics: In the right system of incentives, individuals shoulder the work of prosperity, be it the long-term prosperity represented by recycling or the short-term prosperity represented by performing better at work.

Incentives designed badly can destroy a market, like the once-thick Atlantic Cod populations that desperate fishermen have extracted almost to extinction. Designed well, however, incentives can preserve a resource, as has Alaska's system of Individual Transferable Quotas for salmon. Since the early 1970s, each licensed crew has been able to sell a quota of tons of fish to other fishermen, allowing some to hedge against the vagaries of bad luck while letting fortunate fishermen buy enough permits to exploit a big haul. The wealth of salmon continues.

Economic incentives can encourage people to produce wealth for the whole without using too much individually. The big picture just needs to be considered when setting up the incentive structure itself.

Decentralize Service to the Self

A young woman slices her finger on a knife. As she compresses the bleeding with gauze, she needs to know if her wound warrants stitches. So she calls up Blue Cross' 24-hour nurse hotline, where patients call to learn if they should see a doctor. The nurse asks her to describe the depth of the cut. He explains she should compress it with gauze and skip the ER. In aggregate, savings like this amount to millions of dollars of avoided emergency room visits.

Since 2003, Blue Cross has been shifting the work of basic triage and risk mitigation to customers. Britain's National Health Service (NHS) implemented a similar program, NHS Direct, in 1998. NHS estimates that the innovation has saved it £44 million a year.

Mobile technology enables the promising technique of mobile self-monitoring. Tools that help users solve their own problems funnel an expert's valuable time toward only those cases that require their particular expertise. The burden of basic services gets shifted from credentialed professionals to individuals empowered with technology, whether it's a new car owner printing out DMV forms at home or parolees checking in via ankle bracelets.

Self-guided Internet-based education, such as the Khan Academy, augments an autodidact's library by assessing student weak spots and confirming progress. A simple monitor, such as Nike's FuelBand, which tracks and gamifies fitness workouts, can inspire wiser behavior in individuals, for greater public health. Such technologies can place responsibility for our well-being directly into our own hands.

Gamify Drudgery

Finland's national library houses an enormous archive of antique texts, which officials hoped to scan and digitize into ordinary, searchable text documents. Rather than simply hire people for the tedium of correcting garbled OCR scans, the library invited the public to play a game. An online program called DigitalKoot lets people transcribe scanned words, and by typing accurately, usher a series of cartoon moles safely across a bridge.

In just a few months, 55,000 people helped digitize historical documents with 99 percent accuracy. A formerly tedious task was now fun enough to attract volunteers, and a library perpetually short of funds was able to achieve its goals for much less money.

Gamification-using game mechanics to engage participants in an activity-sets specific goals and rewards them. Platforms such as Rypple and RedCritter's Tracker encourage employees to post their goals and tasks, track their progress, claim skills, and confer virtual "badges" on coworkers. In 2009, the U.K.'s Department for Work & Pensions (DWP) deployed an "idea management platform" dubbed IdeaStreet that gamified the process of suggesting reforms. Within 18 months, 4,500 employees had generated 1,400 ideas; more than 60 of those evolved into successful implementations. Frontline staffers started solving problems that the DWP's central control center knew nothing about. DWP expects the savings from the new initiatives to reach £20 million by fiscal year 2014-15.

Gamification also holds the potential to vastly improve work quality. LiveOps, a virtual call center with 20,000 agents across the U.S., uses a gamified leaderboard: After gamification, some agents improved their sales by 8 to 12 percent and cut call times by 15 percent. Imagine such results at the DMV.

Gamification is simply a reworked form of performance review. You set goals, measure progress, and reward completion. It distributes the effort of enthusiasm from management to employees, and even to the service recipients themselves.

Build a Two-Sided Market

Road infrastructure costs government five cents per driver per mile, according to the Victoria Transport Policy Institute. "That's a dollar the government paid for the paving of that road and the maintaining of that infrastructure…just for you, not the other 3,000 people that travelled that same segment of highway in that same hour that you did," says Sean O'Sullivan, founder of Carma, a ridesharing application.

Ridesharing companies such as Carma, Lyft, and Zimride are attempting to recruit private cars for the public transit network, by letting riders pay a small fee to carpool. A passenger waits at a designated stop, and the app alerts drivers, who can scan a profile of their potential rider. It's a prime example of a potent new business model.

In two-sided markets, an exchange connects two parties-as an Xbox connects avid gamers with game designers, or AirBnb connects beds with travelers. Unlike a retail middleman, the two-sided market links the creator and consumer directly. It provides protections and standardizes transactions, reducing the cost of structuring a deal each time.

Successful two-sided markets tend to subsidize one side, giving the market a money side and a subsidy side. Search engines, for example, charge advertisers but not searchers.

With the ubiquity of mobile devices, two-sided ridesharing networks are thriving. Lyft, for example, with around 30,000 rideshares a week, has become a sort of de facto taxi service in cities like San Francisco, while Getaround and Relay Rides allow people to safely rent out their cars to strangers.

The beauty of the two-sided market is that customers power it. Where institutions become more unwieldy as they grow, a two-sided market becomes more effective with success.

Remove the Middleman

John McNair dropped out of high school at age 16. By his thirties, he became an entrepreneur, producing and selling handmade guitars, but carpentry alone wouldn't grow his business. So the founder of Red Dog Guitars enrolled in a $20 class on Skillshare.com, taught by the illustrator John Contino, to learn to brand his work with hand lettered product labels. Soon, a fellow businessman was asking McNair for labels to market guitar pickups.

Traditionally, the U.S. government might invest in retraining someone like John. Instead, peer-to-peer technology has allowed a community of designers to help John develop his skills. Peer-to-peer strategies enable citizens to meet each other's needs, cheaply. Peer-to-peer solutions can help fix problems, deliver services, and supplement traditional approaches.

Peer-to-peer can lessen our dependence on big finance. Kickstarter lets companies skip the energy of convincing a banker that their product is viable. They just need to convince customers.

Peer-to-peer strategies could help to retrain employees, finance new businesses, reduce traffic, or even review patents. In a trial program, the U.S. Patent office offered "peer to patent" services in which each patent application runs past the eyes of several citizens, often with science backgrounds, rather than distracting a lone bureaucrat.

The market is already unlocking the potential of peer-to-peer sharing. The question is how fast big institutions follow.

Release Your Data

"America is giving you billions and billions of dollars of data for free," Todd Park says. He means government data, like the kind that launched an estimated $90 billion GPS industry, fueling everything from Google Maps to the local weather report to the Garmin on your dashboard.

As chief technology officer of the United States, Park has the responsibility of releasing and publicizing enormous troves of government data-and encouraging private companies to use it. Dozens of health care apps have emerged from the Department of Health and Human Service's Datapalooza conferences, Health 2.0 Developer Challenges, and other data events. These tools match patients to clinical trials, trace the supply chain of food-borne illnesses, or link GPS data to inhalers to chart asthma triggers. Within a year of its 2011 launch, this drive in health care data innovations helped reduce costly repeat hospital visits by 70,000, according to the Department of Health and Human Services.

While big institutions once jealously guarded information, now they're finding that strategic, targeted release of data lets a jungle of entrepreneurs pick apart datasets and construct valuable services.

Open data enables more creative uses and the development of niche markets. Insurance companies have begun sharing data with each other to hunt fraud. Motionloft, in San Francisco, has offered the city its private estimates of hourly foot traffic to better deploy city services.

The Department of Energy has partnered with Siemens Corporate Research and more than 50 university geological departments and state geological surveys to share results of old well bores, hopefully identifying new sources of geothermal energy. Microsoft Research is rendering the test results into 3-D maps.

Many municipalities have released bus data, which private companies use to produce apps which guide users through public transit, thus reducing congestion.

Open data lets information and innovation flourish. It may yet deliver cures for genetic diseases-the Human Genome Project was an early open-data effort-or new sources of clean energy.

Make It a Contest

When NASA and the British Royal Astronomical Society wanted to better calculate the 3-D ellipticity of 2-D images of galaxies (a step necessary to detect dark matter), they were constrained by budgetary considerations. So they designed a competition. On the data bounty platform Kaggle, scientists vied to construct the most accurate method of analyzing astronomical data. Early leads emerged from a glaciologist from Cambridge and a signature verification specialist from Qatar University.

From radio call-in races to grade school reading challenges, people have long responded to contests. While traditional human resources collect minds with an accomplished record in one discipline, contests discriminate only by results, admitting fresh eyes from unrelated fiefdoms. Where patents incentivize only marketable ideas, contests reward scientific advances with less immediate commercial potential. This shows special promise in medicine, where drug research skews toward the ailments of the wealthy.

Sometimes publicity is what a service needs, as when the U.S. Veterans Administration increased access to its Blue Button Initiative that lets patients click a button to see their personal health data and share it with doctors or other health partners. The VA offered a $50,000 prize for the best way to spread the button to other health care providers. The winner, McKesson Corporation's RelayHealth, succeeded in spreading the Blue Button not to just 25,000 doctors, but to 200,000, while the losing teams also widely disseminated the technology.

The federal government has built Challenge.gov, an engine for government agencies to run their own contests. Hundreds of challenges have been posted on the site in the past few years. The competition platform InnoCentive has rewarded inventors for designing a solar-charged flashlight for use in rural Africa and has helped find a theoretical way to simplify the manufacturing of tuberculosis drugs. The $10 million Ansari X Prize spurred 26 teams from seven nations to invest more than $100 million in space flight, investments that may accelerate development in technologies ranging from rocket fuel to seatbelts.

X Prize director Erika Wagner emphasizes that prizes are a supplement, not a replacement: "Prizes aren't good for stimulating basic science, and we need to have a strong science infrastructure in this country," she told Next Big Future. But prizes expand that infrastructure's capabilities, distributing a problem across borders and brains.

Distribute Your Solution

Science-fiction writers used to worry that robots would automate all the work and disemploy everyone but robot owners. Instead, technology is permitting us to coordinate work on a larger scale than ever before. Through microvolunteerism, economic incentives, games, contests, sharing both data and peer-to-peer expertise, and helping people address their own needs, we can distribute the effort of great undertakings.

Building the pyramids and harnessing atomic energy required an elaborate structure of human resources. But that was before the Internet. The challenge of organizing talent is no longer the burden of bureaucracies, but a consequence of systems. These systems can divide big problems into digestible chunks, and spread the labor so thin that it doesn't even feel like work.