This Column Was Crowdsourced By Servio

Find out what happens when a journalist crowdsources a piece about a crowdsource company to the people who work there.

From DARPA soliciting the masses to design aspects of next generation combat vehicles, to Kickstarter crowdfunding projects by artists, musicians, and filmmakers, crowdsourcing has gone mainstream. The FBI has already crowdsourced part of a murder investigation, asking the public's help in deciphering two encrypted notes found in the pocket of a victim dumped in a field. And Iceland has crowdsourced its entire constitution. Even the news is being crowdsourced now: The Washington Post, The New York Times, and others asked readers to scour an ennui-inducing trove of Sarah Palin emails written when she was Alaska governor. The Guardian crowdsourced the expense reports of British politicians. Spot.us asks users to crowdfund news stories. There are plenty of other examples. Ping me on Twitter (@penenberg) and I'll add them to the list. (See how I just crowdsourced that?) 

All of this got me thinking: Why not crowdsource my next column? I'm tired of working so hard, conducting background research, identifying then interviewing subjects, sifting through information, writing, editing, cutting ... adding ... changing. In the time it takes to do all that I could be curled up on a chaise lounge, reading Snooki's book A Shore Thing ("by far the best assisted suicide novel I have ever read," wrote one Amazon reviewer—a member of the crowd, as it were).
 
Then I received an email from a publicist working for Servio (formerly CloudCrowd), a company that breaks up work-related tasks into mini-projects and distributes them to a legion of workers eager to complete them for pennies. I ignored it, like most PR-related come-ons, but on Twitter joked that I should crowdsource the research and writing of a profile about the company, and in subsequent tweets laid out a strategy. The company rep egged me on to do it, so I told him I'd pay their usual rate for the project and came up with 20 questions to be divvied up among the Servio 120,000-member crowd. 

I'm not the only journalist tempted to cut corners in this way. There's a veritable rogue's gallery who've done far worse. Jayson Blair couldn't be bothered to leave his apartment when he filed stories with out-of-state datelines that contained falsified sources and facts, and, while he was at it, he even plagiarized—a trifecta of lies. Stephen Glass, aka the New Republic's master fabulist, didn't interview a single subject when he conjured in part and/or in whole two dozen stories. Nevertheless, hiring a team of freelance reporters to research, report, and write a story on my behalf is not as clear-cut an ethical violation as you might think. After all, newspapers and magazines often hire reporters to file notes that later appear under someone else's byline, and TV news has always borrowed heavily from newspaper stories without any attribution or acknowledgement. So I asked one of Servio's freelancers, Rick Gershman, to contact media watchers on my behalf and ask: Was it unethical to crowdsource the research and writing of a piece to people who were paid by the company I was profiling?

To my surprise, Kelly McBride, an ethicist with the Poynter Institute, wasn't completely aghast. In fact, she was downright helpful in pointing out little things I could do to make my column pass ethical muster. My crowdsourced sources, she said, would have a natural bias, so I'd have to compensate for that, and I'd need to find a way to credit all these people because a byline isn't simply for a writer's ego but also provides accountability. To this end Eric Deggans, a media critic at the St. Petersburg Times, suggested I employ a journalistic firewall, which he dubbed "The Ultimate Adam" (I may have to trademark that) which would be the final arbiter and watch over the crowd to be sure what they report is not only accurate but not overtly biased. Even then, there's a risk: "If there’s a question that didn’t get asked, a fact that gets dropped, or a perspective" that is downplayed or not included—"probably from an honest place—Adam may not know enough to ensure that it gets in, or correct that decision." 

Then there was TechCrunch columnist/author/critic Paul Carr, a walking/talking cautionary tale , who said asking him about journalistic ethics was like asking Ben Kingsley about hair gel. Still, he didn't think it was any less ethical than "the myriad journalists who will gladly rehash a company's press release, changing a few verbs and nouns around, and calling it 'breaking news.' And at least there's a meaningful commercial incentive for [Servio] to do a good job, unlike most professional journalists for whom the only chance of getting paid is by cramming their 'work' full of clumsy SEO keywords." Carr hoped that Fast Company was paying me more than I was paying the crowdsourced workers. If not, he suggested I auction off my byline to the highest bidder. (Great idea!)

Most of what I asked Servio to supply was boilerplate stuff: Interview the founders, find out company revenue, look at the business landscape, explain exactly how the crowdsourcing works, scope out competitors, locate some Servio workers to ask how they like it and why they do it, blah blah, the usual baseline research every tech and business reporter has done hundreds, if not thousands of times. After I had a chance to scroll through the research Servio's worker army compiled for me, I could see that certain answers were handled more deftly than others. It was clear they gleaned basic information from the company's website, through basic Google searches, and in lengthy telephone interviews with the company's founders.

Basic facts were accurate; anything that required interpretation, however, was ripe for abuse. They simply avoided the questions I submitted that asked them to describe the company's greatest weaknesses and to critique its competitors, and I never did find out what the company's revenues were. Anything having to do with the company's cofounders Jordan Ritter and Alex Edelstein was painfully fawning. They were described as "hip, young businessmen" with "boyish good looks," so much so that "it is not difficult to imagine how they have become so successful in the hip world of Internet business." What's more, Ritter, with his "casual elegance," seems like "someone everyone could know," and "when he speaks, his low voice emphasizes his straightforward intelligence. Mr. Ritter is a man who certainly knows what he is talking about." As for Edelstein, with his "long, wavy, blond locks" that "graze his shoulders" and "blue eyes" he "looks more like he stepped off of Project Runway." He was also pegged "a classic California boy" while his manner "is enthusiastic and charming, while holding that slightly intimidating intellectual charisma that can captivate an audience."

Ah, so this is why journalists shouldn't write about friends, relatives, or bosses.

The article that Servio's minions crafted is below. I assigned them 400 words but like most writers paid by the word they had trouble squeezing it into 1,132. You can judge for yourself on whether you think it's good or not—I'd say it's roughly akin to an assignment filed by a first-year journalism student in college—or whether you believe writers like me are even necessary. Maybe companies should just write about themselves and skip the journalists, the Sarah Palin approach to media. At any rate, the article below, along with pages and pages of Servio research I didn't end up using or needing, cost $222.92 and net me $177.08 for my efforts.

While it was fun, it hardly seems worth it. From now on I think I'll do my own work.

Servio
By Servio
 
SAN FRANCISCO - At this very second—regardless of whether you're reading these words 4 p.m. or 4 a.m.—there are at least several hundred people working for Servio. It could well be several thousand. These are freelancers from around the world, hard at work on tasks that pay from a single penny to more than $20 at the company’s online workspace known as CloudCrowd.

Let's pause for a second. Did that sound too promotional? Consider the source: These very words were composed by one of Servio's "crowdsourced" writer-editors. He was paid by the company to assemble this article from other Servio writers. But Adam will tell you more about how that. Let's talk about Servio: When a company has a need for a project such as indexing websites, editing documents, or writing press releases, it can submit the project to Servio, where it will be separated into several smaller jobs.

The startup was launched in April 2009, and it released CloudCrowd as a Facebook application in October 2009. Since its inception, the company has expanded to include about 120,000 workers, and that number is growing by about 2,000 each week. Alex Edelstein, the Chief Executive Officer, founded the company with Chief Technical Officer Jordan Ritter. With his short, dirty blond hair and groomed five o' clock shadow, Jordan Ritter looks like the typical college fraternity brother. Alex Edelstein looks more like he just stepped off the set of Project Runway, with wavy, blond locks that graze his shoulders. [Editor's note: Edelstein has cut his hair since the piece was filed.]

Both men were born in Los Angeles. Edelstein’s family moved to the Silicon Valley when he was 5, and he grew up in Palo Alto. Edelstein studied engineering at Harvard, then was hired by Microsoft, where he worked on the original version of the Exchange application and what would become Outlook.

Uprooted from L.A. at age 2, Ritter spent his childhood in Texas and his teenage years in Florida. His passion in high school was music, while his aptitude was for computer science. He later would major in both at Lehigh University in Bethlehem, Pa. Appropriately, he went on to co-found the online music file-sharing service Napster and act as the Chief Technical Officer for Columbia Records in Tokyo, Japan.

In spring 2009, the pair teamed up to create Servio, which was originally called CloudCrowd. It could have just as easily been named CrowdCloud. "We were looking at both names," Edelstein recalled. "We wanted it to communicate both the concept of using a large workforce and the concept of the Internet."

Why CloudCrowd, and not the reverse? "It’s pretty simple," Edelstein said. "The guy who was squatting on (the name) 'CrowdCloud' wanted $100,000 for it, and they guy who was squatting on 'CloudCrowd' wanted $1,000 for it."

The term "crowdsourcing," or using collective intelligence to complete a project, was coined by Jeff Howe in a 2006 Wired magazine article. Crowdsourcing allows a large task can be completed in a few hours, or even a few minutes, because numerous people can work simultaneously on individual portions. Servio prefers the term "widesourcing," suggestive of the worldwide breadth of this type of workforce.

The concept has many applications. The Christian Science Monitor is using crowdsourcing to create a new cookbook, an idea suggested by New York Times food columnist Amanda Hesser.

Ritter, who coded the CloudCrowd application, was interested in merging cloud computing with the human capital available worldwide. Launching the company required a different type of capital: Edelstein originally funded the company with $1 million and later added another $500,000. In August, the company raised $5.1 million in venture capital funding.

Finding workers wasn't hard, the founders said. The idea of getting paid every business day for work you can do from anywhere—the company compensates workers through PayPal Monday through Friday—quickly attracted a large workforce.

The majority of Servio’s business comes from companies such as Target and Healthline that need high volumes of written content from product descriptions to articles for their websites. Other business comes from simple tasks that would be tedious if they had to be done all day and around a widesourced version of traditional proofreading, writing and translation services.

CloudCrowd's user interface is friendly, but workers must grab the work when it appears. On average, a worker has about 60 minutes to complete each task. CrowdCloud.com has assignments for everyone, but higher-paying ones are available only to those who pass qualification tests.

Pay rates vary broadly depending on the task. Editing tasks typically pay between $1.25 and $1.75 per section. Writing tasks can range from $1 for short tasks to more than $20, with translation work paying in a similar range. Workers without specialized credentials can do lower-paying URL search and categorization tasks, generally ranging from a penny to 40 cents.

If your task is approved through peer review, you get paid. If it's not, you don't. If your task submission is rejected, you will also lose credibility. You gain credibility by doing good work.

Servio could be considered another entrant in the "crowdsourcing" field that includes Amazon's Mechanical Turk, oDesk and many others, but Ritter says it's evolutionary, overcoming the lack of automated quality control, for example, that is typical of first-generation platforms.

Evolution is big at Servio, which recently updated its client-side operations into a service which provides "one-stop shopping" of a sort for customers who want to harness the crowd for various business and academic needs, including editing and translation projects.

While Servio’s growth has been explosive, it is not yet releasing revenue numbers, said VP Marketing, Mark Chatow.

Gaby Merediz, an American stay-at-home mom, talked about her first weeks with CloudCrowd: "For some tasks, I only earned a penny. I justified the time and the pay by reasoning that if I was just sitting around the house doing nothing, I wasn’t getting paid at all." She has since moved on to far higher-paying writing, editing and translation work.

When RentCycle CEO Tim Hyer needed to categorize more than 20,000 rental business websites, he turned to Servio: "Before I knew it, I had a clean spreadsheet of organized data that was now properly validated, categorized, and sub-categorized—with time to spare before my deadline."

The system doesn’t mesh perfectly all projects. As Chatow explained, "Some companies need a solution that's akin to hand-modding a car. We cater to clients who want a Mercedes off the factory line."

Several workers, however, said flexibility is one of the main reasons they enjoy the platform.

"As a stay-at-home mom, I can check tasks when convenient and work around my daughter’s schedule," said Jessica Fresco, who lives on the U.S. west coast. "There are a variety of short tasks… so I don't run the risk of leaving a job incomplete when I have to attend to my daughter's needs. When she's asleep, I can work on the longer tasks."

[Image: Flickr user francistoms]

Add New Comment

4 Comments

  • Mark Nowotarski

    For those interested in doing their own
    kickstarter project, there is an entertaining and informative video
    at http://bit.ly/kikpat4. It's
    based on a talk I gave to the CT Inventor's Assocation.

  • Mark Chatow

    I am Servio's VP Marketing, and worked with Adam to help coordinate the story. While it was a lot of fun to work on the piece, we never expected that we'd best a writer of Adam's caliber.  We were curious to see how our high-volume factory-oriented process would work with something as hand-crafted as a story like this, but we don't purport to be a replacement for a journalist writing a feature story.  

    Ironically, because the piece was written about us, we actually had less ability to tweak the output than we normally would have to ensure quality. Because there was already inherent bias built into the piece, we were very careful to have as little direct influence on the final results as possible.  We didn't tell the writers to lean one way or the other, and we were just as amused as everyone else when we first saw things like the descriptions of our co-founders.  

    Adam mentioned that our writers ignored his questions about revenue, weaknesses, and competition and went significantly over the requested word count.  We really struggled with how to handle these questions. As a privately-held startup we don't release revenue numbers, and we couldn't make an exception in this case.  The question on weaknesses was ironically answered in part by my quote that we're good at delivering a "factory line" result and not at "hand-modded" content--for example, pieces that involve significant amounts of iterative work, in-person interviews, etc., like a feature profile
    . Adam's original question on the topic asked our writers to find clients of ours who were unhappy with our work.  Our writers don't generally hold back, and if there clients out there who weren't happy and wanted to air it in public, they'd have been glad to write about it. 

    Regarding the competition question and word count: Adam had originally given us 20 questions, many of which called for detailed answers.  We actually had copy in our original materials on our competition, but we were already going far beyond the initial word count.  The CloudCrowd editor working on the piece made a call that it was a profile on Servio, and that if something had to go it should be the non-Servio related content, so he cut the discussion of our competitors.

    Servio's model has evolved since Adam first tweeted the seed idea for the story back in August 2010, and since then our focus has been on what we call "industrial-strength content engineering".  
    Our sweet spot is creating and enhancing high volumes of content for companies like Target who need tens of thousands of product descriptions or category articles written.  They can't scale it in-house or with freelancers, and they can't offshore it while maintaining their brand voice.  They need solid writing, but nothing at the level of a writer like Adam.

    The bottom line is that we appreciate great writing and we enjoyed Adam's article about the process as much as anyone.  The world is a better place with people like Adam writing the stories they do, and we are pretty sure that isn't going to change any time soon.

  • Gordon Gooch

    I agree, the true concept of crowdsourcing in this article is not the same as piecemealing the questions up and then assembling the article from the questions and answers. True crowdsourcing, and one that is far more efficient, involves challenging individual writers to submit complete articles. Not bits and pieces.

  • Anji

    This is why "crowdsourcing" model needs people like you Adam :-)
    Such an article cannot be crowdsourced, at least right now.
    What we call "task design" is really important to get the best quality in content.
    Not everything can be crowdsourced. In this very case, the result would have been better when only one person had made researches and write about it.