Chasing Pennies – My time as an Amazon Mechanical Turk Worker

A look into my two years working as an Amazon Mechanical Turk Worker and my three months as a requester.

Advertisements

Update: Any time I attempt to log back into Mturk, I swiftly exit the page. Writing this article put much into perspective for me and cemented my view that crowd work is no longer a good fit for me. Best of luck to anyone out there whom still wants to give Mturk a try!

—————————————————————————————————————————————

For the past few years, I have invested more time than I probably should have into working for crowdsourcing services. Amazon Mechanical Turk, Clickworker, Rev and User Testing are the primary platforms I used, but several of these turned out to be dead ends.  The latter three, I barely used and never fully established a foothold.  Mechanical Turk, however, worked quite well for my purposes.

For the uninitiated, Mturk is, at its core, a crowdsourcing platform: companies or individuals can request work (data entry, receipts, etc.) to be done. These are called human intelligence tasks. Each hit can be offered an award amount. This can be anywhere from nothing to whatever the requester wants to pay. I have taken $20 surveys previously on the platform, but these are like Captain Ahab’s White Whale in rarity.

Most hits only pay pennies. After a while, it adds up. Especially, if you know how to use a combination of scripts, the community, and other tips and tricks to maximize your gains on the platform. Unfortunately, it’s quite addictive. Several online resources exist to help nascent Turkers in their quest. However, the secrets of the high-earners are never revealed. Also, despite the ready availability of Mturk resources, there will always be that one user who decides to post a question about the following (already covered) topics:

  1. How to get started.
  2. How to make more.
  3. How to get approved.
  4. Where to find the best HITs.
  5. How to report Mturk income on your tax return (trust me; you should report it).
  6. And much, much more!

I dove headfirst into the world of MTurk in March of 2014.  In the time since, I have become quite adept at the tedium that being a Worker on the platform represents. I first learned about Mturk from my research into crowdsourcing in 2014. I became interested in the idea of microtasks and, to better understand it, I decided to participate. Gonzo journalism at its finest. I think Hunter S. Thompson and Jon Ronson would be proud.

Diving into the world of crowdsourcing proved to be a fruitful endeavor. I learned how to use scripts on Mturk for maximum efficiency. With Rev, I managed to make a little bit of money doing transcription work, which proved to be tedious. I quit after a week.

Usertesting is a good source, but work is very rare. It pays $10.00 per successful test, making it one of the higher paying sources I have found. Clickworker/URSE and Prolific Academic are excellent alternative to Mturk, but do not provide nearly as much available work in my experience.

The sources for Mturk are much more detailed, friendly, and useful than one would think, based on the competitive nature of being a worker. Users over at Reddit, Turker Nation and Mturk Crowd make things much easier for new Workers and are, for the most part, incredibly welcoming, and friendly areas.

In my own experience, Mturk was a way for me to make some extra cash to help build my home studio. I started out chasing pennies, finding whatever I could to build my numbers and make little bit of money. Later, I discovered a HIT I enjoyed, wherein the user would draw a box around a section of an image and then write a sentence about the image.

I did thousands of those HITs, which I later found out were part of Stanford’s Visual Genome Project, until my eyes were sore. I made some money, bought a few studio components, and paid a bill or two.

I listened to Frank Zappa’s “Does Humor Belong in Music?” and the music of R. Stevie Moore, among others (tons of Weird Al, The Arrogant Worms and Dr. Demento to be specific). I worked, toiled, and relished the sweet, sweet pennies, nickels, dimes, and occasional quarters coming my way.

Did I mention I once snagged a $20 survey? That made my entire Mturk career. I’ve heard of working for change before, but this is ridiculous!

I gained weight, became detached from my peers and my health took a nosedive. Mturk became a fixed, toxic cloud hanging over my head. Upon awakening, I would immediately fire up my computer, set up my scripts and turk away. At one point, I also began using the requester side of the platform, in addition to being a worker.

I treated turking as a second job, with myself as a supervisor. Let’s just say I came to despise my supervisor after a time. The burn out set in after the first year, but I continued to turk up until recently.

I had to ask myself, is the extra money worth it?

For some, I’m sure it is. If you venture over to Reddit (https://www.reddit.com/r/mturk/), Turker Nation (http://www.turkernation.com/), or Mturk Crowd (http://www.mturkcrowd.com), you may find some who enjoy the work. What you’ll mostly find is a wide variety of people turking for an even wider variety of reasons.

Some of us do it for extra spending money, others for all their income. It varies based on circumstance. What some of the articles always seem to “get wrong” is their focus. The play up the sad stories, or the exploitative aspect of the service, without really delving into multiple sources.

In my own research, I decided to use Mturk for research studies and other projects. I pay more than a slave wage and will only reject if necessary.  After an extensive lit review and reading a few online guides from other researchers, I determined a methodology and payment scale that only resulted in one poor review on Turkopticon, the distinct-from-Amazon review aggregate service that workers use to gage whether to do a certain HIT or not. Most of my information came from an excellent resource by Michael Buhrmester, PhD, located here: https://michaelbuhrmester.wordpress.com/mechanical-turk-guide/.

Pew Research Center conducted a research study on Mechanical Turk in 2015, where they reported that most Turkers make less than $5 and hour. That seems accurate in my experience, though a perusal of any forum will tell you there are exceptions. A full copy of that research study is available here: http://www.pewinternet.org/2016/07/11/research-in-the-crowdsourcing-age-a-case-study/.

The way a work day goes, depends on the worker. In my own experience, my thought process would go something like this:

I have enough in my Amazon Mechanical Turk account right now to make a small payment on a credit card; maybe to buy some groceries or put a tank of gas in my car. I earned these funds over the past three weeks, working anywhere from two to five hours each day, every day of the week. Sundays have been the least productive, while Mondays and Tuesdays have seen major increases in number of HITs approved on my personal account.

I am dreading the prospect of working anymore HITs this week, but at the same time, I feel compelled to do so. This is helping pay for my education and a few other expenses.

This reflection doesn’t change the fact that I have been staring at my main account page for the past half an hour, trying to decide if the few dollars I’ll make today are worth the time investment. Cogitating upon whether to work more HITs or to do something else is exhausting. Ultimately, if any of my favorite requesters – chiefly Audiokite – post, I will decide to stay and work.

I strongly believe that if Audiokite posts today, it just might be worth it to grind a bit more. To grind is to earn. On Amazon Mechanical Turk (MTurk), grinding – or working as many HITs as possible in a short period of time – is a way of life. It reminds me of many a High School afternoon (and evening, going into early morning) grinding for levels in “Final Fantasy VII.” Only this time, I’m not on a quest to prevent the end of the world; I’m on a quest to earn as much money as possible, one penny at a time. Although, I suppose Chocobo farming is probably worse than anything MTurk could throw at me.

Audiokite is posting multiple HITs today, ranging from a $0.10 payout for 40 seconds of listening and a short review to a whopping $0.18 for two minutes of listening.  I can knock out the entire batch and receive a $5.00 bonus. There are over 300 HITs in this batch. OK. *clicks mouse and opens new Mturk tab.*“

There is no doubt in my mind that most Amazon Mechanical Turk Workers (Turkers) know what I am talking about. Of course, my experience is unique, as is the experience of anyone working on the platform.

Some workers, only do HITs for a week and abruptly quit the service, like Jeremy Wilson did for his 2013 Kernal article (http://kernelmag.dailydot.com/features/report/4732/my-gruelling-day-as-an-amazon-mechanical-turk/ )

How the media sees crowd work, compared with how the actual workers see crowd work an interesting comparison. Sometimes, it surprises me to see just how skewed some of this information is. Other times, it will be completely accurate, to the point where I’ll say, “wow, that’s just like my experience!” It’s an interesting take that is worth further study and much more intense scrutiny.

Some stories try to present a view of Mturk as a virtual sweatshop (http://www.vocativ.com/410794/are-virtual-sweatshops-the-future-of-work/), which is accurate to a point, but in my experience, turking is what you make it. The article outlines a few individual stories, which are outside the norm, and didn’t bother doing any research within the actual communities.

Journalists should double and triple down on their sources, with supporting documentation. Reading the article gives one the sense that these journalists were only looking for the sensational scoop that would get them the most page views and traffic.

Sure, I get it; sad stories make for more page views. That makes sense, and there’s an entire book about it called, “Trust me, I’m Lying,” by Ryan Holiday. It’s worth the read. In the book, Holiday discusses the concept of valence with the reader. Valence is what determines whether a story is shared. Per Holiday, sad stories get shared the least, but polarizing stories are shared the most. Hence, the content is intended to anger readers or upset them, so they share it more often.

That isn’t to say that Mturk doesn’t have some major problems with equal pay and worker rights. For instance, a requester can simply decide to reject a submitted HIT, resulting in the worker not being paid for their work.

This is a vicious cycle: the worker does the work, the requester rejects it, requester gets free data, worker gets a strike against their account, more rejections lead to blocks, block lead to losing your worker account. It’s a harsh domino effect that leads to a loss of income. For some workers, that can be a major blow to their livelihood.

In this article from Tech Republic: http://www.techrepublic.com/article/inside-amazons-clickworker-platform-how-half-a-million-people-are-training-ai-for-pennies-per-task/, the reader learns a bit more about what makes the service run, what it’s used for and features accurate, in-depth interviews with real turkers.

The difference between the two articles is stark. The Tech Republic article tells the story with ore impartiality than either The Kernal or Vocativ stories do. It also does something the others don’t: it reaches out to the foremost authority on being a Turker today: Kristy Milland.

Milland founded Turker Nation and has authored several papers and performed speaking engagements for crowd work. She can be found here: http://kristymilland.com/. She is quite active in the community and has done much to help workers.

Turker Nation maintains an excellent sub-forum with tremendous resources for crowd work in the media: http://turkernation.com/forumdisplay.php?56-Crowd-Work-in-the-the-Media.

Nothing truly sets me apart from anyone else. My experience is unique, but I am not. I must compete in much the same manner as my peers and choose whether to contribute advice to the community; I haven’t really done much in that respect.

So, the question is, do I keep grinding away at Mturk to make extra money or do I find something else entirely (spoiler: I quit Mturk for the time being)? The gig economy is not the best, nor easiest way to generate income. However, with automation concerns, market saturation and economic downturn, the gig economy may be all we have left at the end of it all.

The saddest thing of all: some of the work we do on Mturk may be training the artificial intelligence and automation systems that will replace our jobs in the future. Will the robots take over the job market? Maybe (https://www.theguardian.com/technology/2017/jan/11/robots-jobs-employees-artificial-intelligence ).

But, maybe not (https://www.ted.com/talks/david_autor_why_are_there_still_so_many_jobs).

Perhaps, Elon Musk will save us all by convincing the government about the validity of his ideas (http://www.cnbc.com/2016/11/04/elon-musk-robots-will-take-your-jobs-government-will-have-to-pay-your-wage.html)!

 

(###)

 

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s