Can Crowdsourcing Make a Dent in Unemployment? Ask MobileWorks

Jobs are the single biggest political issue of the day in the U.S., and rightly so. As of August, the official unemployment rate in the United States stood at 9.1 percent. That was down one point from the October 2009 peak of 10.1 percent, but still higher than at any time since the 1930s, with the exception of the worst months of the 1982-83 recession. And today’s real unemployment rate, if you include discouraged workers who have stopped searching for jobs and people who have settled for part-time positions, is much higher, at around 16 percent. That translates into 25 million Americans who need work.

That’s a terrifying number, because no one knows how the country might create that many new jobs. Let’s say President Obama’s $447 billion jobs bill were enacted in its current form (an unlikely prospect, given the levels of partisan obstructionism in Congress). The most optimistic estimates from economists are that the new spending in the bill would add only 2 million jobs to the economy in 2012. That’s nothing to sneeze at, but it’s a tenth of what’s really needed.

On top of that, technology-driven productivity gains are allowing American corporations to rack up big profits despite their trimmed-down workforces. So even if consumer demand were to magically return to pre-recession levels, companies probably wouldn’t hire back all the people they’ve laid off since 2008.

Could there a technological cure for unemployment? Ever since the 2005 introduction of Amazon Mechanical Turk, a Web service that pays users small amounts for completing tasks such as transcribing short audio recordings or recognizing an object in a picture, tech pundits have been talking about the benefits of digital crowdsourcing. It’s been portrayed both as a way for companies to get work done cheaply and as a source of supplemental income for casual Internet users with a little time to kill. But with the joblessness picture looking so dire, observers are now starting to ask if crowdsourcing technology could play a more central role in economic recovery. Could Internet-mediated “microwork,” as this kind of employment is being called, give millions of people a way to earn meaningful amounts of income?

Mechanical Turk, by itself, certainly isn’t a panacea. As a service, it’s not terribly inviting or easy to use, and Amazon itself has never expressed much ambition to improve or expand it. And by definition, Mechanical Turk workers need a computer to complete tasks, which leaves out a big slice of the unemployed population.

Here in San Francisco, though, there are at least four organizations taking Amazon’s idea and tweaking it to make microwork more feasible for broader populations. One is Samasource, a non-profit that distributes computer-based tasks such as data verification and audio and text transcription to workers in Haiti, India, Kenya, Pakistan, South Africa, and Uganda. Another is Servio, which creates crowdsourced editorial content for clients using a platform called CloudCrowd. Then there’s CrowdFlower, a kind of meta-platform that helps big, Fortune 500 companies with data management tasks by farming out the work to Mechanical Turk, Samasource, and other crowdsourcing engines. Finally, there’s a new player called MobileWorks; it’s a Y Combinator-backed startup that offers digital work to underemployed people in India and Pakistan, but with an emphasis on tasks that don’t require a computer and can be completed using only an Internet-connected mobile phone.

Yesterday I called up MobileWorks co-founder and CEO Anand Kulkarni to find out how people are using his platform, and whether the technology might offer hope for unemployed people here in the United States. As you might expect, he’s bullish on crowdsourcing’s potential for alleviating joblessness, whether in developing countries or developed ones. “It’s the only kind of work you can do anywhere in the world, without being limited by your education, your background, your gender, where you grew up, or who you are,” Kulkarni says. “It’s the kind of work you can do in a refugee camp, an urban slum, or a rural village, and it’s the kind of work that people can’t take away from you.”

The tasks MobileWorks sends out to its members are what Kulkarni calls “technical but unskilled”—tagging images, proofreading articles, reading handwriting that computers can’t decipher. The average MobileWorks member, he says, does about three hours of work each day and earns about $1.50 per hour. “Before joining MobileWorks they were earning an average of $2 to $3 per day, and after joining they are typically earning $4 to $5 per day,” Kulkarni says. “So it’s roughly doubling their income.”

As soon as the startup can clear a few regulatory hurdles and round up enough work assignments, Kulkarni says, MobileWorks plans to make its platform available in the United States. The types of jobs the site will offer to people here will be different from those it hands out in India and Pakistan, and the wages it offers will be higher—ideally at or above the minimum wage, which is currently $7.25 per hour. “We’ve had people here write to us to ask if they could join MobileWorks at the Indian wage, which really says something about the state of employment in the U.S.,” Kulkarni says. “Of course, we are never going to do that. Our objective has always been to make sure that we are paying people a fair living wage wherever we are deployed. So we want to make sure than when we do roll out in the U.S., we can pay people something meaningful.”

The MobileWorks founding team. L to R: Anand Kulkarni, Prayag Narula, David Rolnitzky, Philipp Gutheim

MobileWorks’ founding team of four, which also includes Philipp Gutheim, Prayag Narula, and David Rolnitzky, all hail from the University of California at Berkeley, where they originally developed the project as part of a workshop in the School of Information called Information and Communications Technology for Social Enterprise (ICT4SE). Kulkarni says the company’s special sauce is in its quality assurance algorithms, which monitor workers’ performance and assign new tasks to the workers with the best track records. “You don’t have to worry about who is doing the work or how you’re paying them or making sure they’re doing it right,” says Kulkarni. “That is the service we are charging for.”

MobileWorks has raised $20,000 from Y Combinator, where it was one of 63 companies to take part in the Summer 2011 session, and another $150,000 from Start Fund, the YC-focused investment fund set up by Yuri Milner and Ron Conway. Kulkarni says the team also just finished raising a “big seed round” from both social investors and traditional technology investors—watch for details soon.

My interview with Kulkarni, outtakes of which appear below, left me optimistic that the techniques social entrepreneurs have been developing to overcome barriers to employment in the developing world might also end up benefiting the advanced economies, where it’s become clear lately just how fragile prosperity can be. Of course, whether large numbers of Americans will be willing to do minimum-wage microwork is an open question. One study of Amazon Mechanical Turk found that users there earned an average wage of $1.38 per hour—which probably explains why the typical Mechanical Turk worker to date has been a 20-something person who uses the site to earn a little extra cash in his or her spare time.

But Kulkarni argues that even if the tasks themselves aren’t high-paying, crowdsourcing and microwork bring other benefits, such as the ability to decide when, where, and how much to work. “When people are doing crowd work, they get to set the terms of their employment in ways they cannot do with other work, which in itself gives a kind of dignity and freedom,” he says. Here’s the interview:

Xconomy: One obvious prerequisite to using Amazon Mechanical Turk and some of the other crowdsourcing platforms is that you need a computer. How is MobileWorks addressing that limitation?

Anand Kulkarni: When we started out, we were looking at why groups couldn’t use existing platforms like Amazon Mechanical Turk to get themselves out of poverty. We found that, first, people couldn’t understand the tasks, and second, they didn’t have access to the computing resources necessary to do this kind of work, which means that only the groups with access could do it. For us, it was an issue of reconstructing the work into a format where workers who only had access to phones could do it.

The reason we are called MobileWorks is because the mobile phone can go places a computer can’t. About 30 percent of our workers are working on low-end Nokia feature phones, paying $1 a month for a data plan. So crowdsourcing doesn’t have to be deployed only to technologically equipped, technologically literate populations. Even in low-income parts of the world, having a low-end feature phone is common, and it gives you many more capabilities than you might think. Nokia Series 40 phones, which is our standard example, are capable of light Web browsing. You might not be streaming video on them, but you can surf the Web.

X: Can you describe a few types of tasks that might be distributed on MobileWorks?

AK: The kinds of work we typically have our mobile workers do is, for example, handwriting recognition—similar to OCR [optical character recognition]. A piece of handwriting gets put on our website, we chop it into small pieces, and the work gets sent out to workers on their mobile devices. There is no technical requirement on their end aside from having data connectivity. A second example of the kind of work you can do on a low-end phone is spam filtering—taking a piece of content and deciding if it’s spam or not. The amount of data involved is relatively lightweight, and it’s really easy for a worker to understand what they need to do.

X: Another limitation with Amazon Mechanical Turk was that until recently, you had to be in the United States to get paid, though now Amazon can also pay workers in Indian rupees. How are you handling the payment issue?

AK: Our strategy has been to partner with local banks. When workers join our system, if they don’t’ have a bank account already they get an account. We have accounts with banks in three countries, and we do a bank-to-bank transfer. This is pretty efficient for us. It slows down how fast we can deploy into new countries, but once we are set up it works fairly well. It can all be done electronically. Of course, the best long-term solution would be mobile money transfers, but the rollout of mobile money has been a little bit inconsistent in the wireless industry.

X: Is MobileWorks helping users earn meaningful amounts of money?

AK: We know it for a fact—they write to us and tell us. We started from the premise that we would do two things: one, we would target populations that were historically underemployed, and two, we would make sure to pay them fair living wages, not purely as a charitable operation but because we think this is a driver of consistency and quality. When you pay people a fair wage, they are much less likely to spam you.

X: Are workers using this income to supplement their income from other jobs, or is it becoming the main source of income for some people?

AK: For some of our population, it is their primary form of work. It was never intended to be. It was designed to be supplemental, the kind of thing somebody could do on the way to their full-time job on the bus, or in transit. What happened was, we found that the kinds of people who were using MobileWorks to get work were the kinds of people who could not get any work outside the home. A typical example is a housewife in Pakistan. A substantial population in our system are folks who, because of various social factors, can’t go outside the home to get work. Unmarried women, or married women with children.

Initially we thought we would try and force folks to work only two to three hours a day. When we saw that people were working 8-, 9-, or 10-hour stretches, our first reaction was to try and figure out what was going on and try and limit them. But the workers made very compelling cases. They told us, ‘I’m not married, I am taking care of my elderly parents, I can’t work outside the home, it makes a difference in my life.’ We couldn’t argue with that. If people are using the system to support their families, that supports our mission as well.

X: Are there any big differences between India or Pakistan and the United States that might make it difficult to deploy MobileWorks here?

AK: It’s not the motivation to work—we have found that the same model we are promoting in other countries is extremely appealing here, because it’s the kind of work where there are no barriers to entry. Anyone can pick up a phone or a computer and start going to work. The main limitations for us here have been making sure we are compliant with U.S. labor regulations.

Not to call out our competition too much, but sites like Amazon Mechanical Turk get away with paying folks a wage which is below what you would make in any legal job. And it’s because they are not paying folks as employees, but piecemeal. Our objective has always been to make sure that we are paying people a fair living wage. So deploying in the U.S. means that we have to consistent with U.S. regulations, and U.S. labor law doesn’t know what to make of crowd work just yet. There haven’t been any defining cases on the matter. In the meantime, we are looking for ways to plunge ahead. For us, we are extremely passionate about deploying here in our own back yard, both in San Francisco where we did our first tests, and in harder-hit parts of the U.S. like Detroit, for example.

X: What about the companies that supply the work that’s distributed on crowdsourcing platforms like yours—what’s in it for them? Is it just that it’s cheap?

AK: Crowdsourcing isn’t just an innovation on the labor side. It’s also a new source of efficiency on the supply side, the customer side. Large or small companies need to contract work out to folks like you or I, and right now there is a massive transactional cost, in terms of paperwork. This cost is paid by everybody up and down the food chain. By lowering the transaction cost of creating work we tap into a large pool of work that is not being given to anybody today because it’s not efficient.

Part of the reason that Amazon has been so widely recognized is that you can get stuff done on Mechanical Turk for absurdly cheap prices. I don’t think that’s necessarily always going to be the case in crowdsourcing. But we offer prices that are extremely low, even while paying people fair wages.

X: I want to ask you a question about the quality of the work available to people on the crowdsourcing platforms. Much of it seems like stuff that, frankly, computers could probably do if the state of software improved just a little. Do you ever worry that the type of work you’re handing out is dull or demeaning for workers?

AK: It’s a great question. And I’ll tell you one more question that we think about: does this take them anywhere? Are there career development prospects if you do data entry work on a mobile phone for nine months?

There are two parts to the answer. First, in a way we might not appreciate quite so well in the U.S., in the developing world, digital work is meaningful. It’s considered good work. If the alternative for a person living in an urban slum in Mumbai is a factory job smelting plastic over a furnace and breathing toxic fumes all day, or sitting in an air-conditioned office processing images on a computer, they will believe the latter to be much better than the former. But it’s not just a question of comparing it to the alternatives. It’s that digital work is seen as an aspirational form of employment in these countries. Somebody who is working at a computer job or an office job is the new information worker in India. That job will have downstream prospects.

The second part to the answer is—remember that a good portion of our workers can’t leave their house for work. So for them having work of any form that can’t be taken away form them, that nobody can tell them they can’t do because they’re a woman or born in the wrong part of the world, is very valuable and very meaningful. A few workers have also told us it’s not just about the money, but they feel like they are learning something. The OCR workers say they get to practice their English, which we didn’t expect at all.

X: Still—do you think the typical unemployed worker in the U.S. would be willing to do this kind of work?

AK: The kinds of work we are going to offer in the U.S. are not going to be the same kinds of work we do abroad. The kinds of work we do here will be specialized for the nature of the U.S. market. So one person we have talked to was an audio transcriptionist for several years, and she lost her job. The company shut down and let her go. I don’t think any of us would say that audio transcription is a demeaning or demoralizing job. It’s certainly a respectable thing to do. We expect to involve her in MobileWorks once we have our US rollout, doing the same kind of work, but instead of going to an office she’ll be doing it from her own home.

X: Back to the big-picture question: Do you think crowdsourcing has an important role to play in reducing unemployment in the United States?

AK: I’m an optimist, and I’m a person who has dedicated his life to a crowdsourcing company devoted to tackling labor and employment issues, so of course my answer is going to be a little biased. But I am hugely optimistic about the potential. I think that crowdsourcing is enabling new kinds of work. It’s creating new efficiencies in the economy that are leading to new job opportunities that weren’t there before. So I’m optimistic that we can make an impact. I think that over the next three years, you are going to see more and more folks who say that they make a meaningful part of their income through a crowd work platform. People have viewed it as a technological mystery for a while, but at the end of the day, it’s just another form of labor.

Wade Roush is a freelance science and technology journalist and the producer and host of the podcast Soonish. Follow @soonishpodcast

Trending on Xconomy

By posting a comment, you agree to our terms and conditions.

12 responses to “Can Crowdsourcing Make a Dent in Unemployment? Ask MobileWorks”

  1. Alex Buran says:

    Add to this list Translation Cloud

    The Mechanical Turk-like app that provides work to unemployed translators through the Facebook.

  2. and add Seattle-based CrowdControl Software

    Crowdsourcing has been slow to take hold because the quality of the output is low and it doesn’t scale.

    CrowdControl Software tackles this problem head-on – all about high quality output regardless of data size, and at a low price.

  3. Gerald W says:

    There’s another startup ( that distributes predictive datamining problems to “data scientists” across the world.

    Rather than focus on simple tasks, the goal is to source the best of the best (the results so far support this admirable objective).

  4. John Glendale says:

    Wonder if they’ve heard of before

  5. Wade RoushWade Roush says:

    Alex, John R, Gerald, John G: Thanks for reading, and for mentioning crowdsourcing options like Translation Cloud, CrowdControl, Kaggle, and Gigwalk. I’ve heard of all of these, of course (and in fact I have an article about Gigwalk in the works). But I didn’t want this article to turn into a catalog, so I mentioned only a few of the crowdsourcing platforms, with an emphasis on those close to home here in San Francisco. Everyone — please feel free to name your favorite crowdsourcing options here.

  6. Alex F says:

    How would crowdsourcing to India lower unemployment in Chicago?

  7. Wade RoushWade Roush says:

    Alex F: As the article states, MobileWorks plans to adapt its platform to offer work to people here in the United States.

  8. Dana D says:

    Assuming the audio files or tapes that need to be transcribed and the materials that need to be proofread are in English, why outsource these tasks to countries where English is not the first language?

  9. Wade RoushWade Roush says:

    Dana D: Actually, I would pit the English skills of the average, mildly educated Indian or Pakistani against those of the average American any day. ;)