Is Amazon Mechanical Turk Moral?

Making money online with Amazon Mechanical Turk (Mturtk) is the hard road to take. If you’ve read some of my posts, then you definitely caught some of my ambivalence about Mturk.

In some ways I see Mturk as being a modern version of the sweatshop. Instead of a thousand workers being housed under one hot, steamy roof with tyrannical pit bosses with a whips walking up and down the aisles ready to “incentivize” the less productive, you have lots of workers all over the world in their own private sweatshops -- sans pit bosses.

In an earlier post, I told you the results of a survey conducted with Mturk workers. One of the questions in that survey is really the best framing device for answering the question as to whether Mturk is moral or not.

The question asked:

Is the money you make on Amazon Mechanical Turk your primary source of income?

Only 12% of the U.S. respondents answered yes to this question. Workers from India responded with 28% affirmative.

So, in the U.S. no one is really making you work for Mturk.

But does that justify the shamefully low wages offered by requesters on Mturk? Because no one makes anyone work on the HITs, it’s okay to pay a someone a sub-standard pay for work? The bottom line: Is it okay to exploit people that allow themselves to be exploited?

Certainly the vast majority of workers on Mturk could walk away from Mturk and find a way to earn similar or better wages. This does seem to true. Sure, just about anyone who works on Mturk could go out and find job, but in this economy, some people need the money they get on Mturk to make ends meet and jobs aren’t the most plentiful things out there.

In the free market world of capitalism, anything goes I guess. Freedom means the freedom to be exploited or starve. It’s all about about choices. Some choices just aren’t as enjoyable as others.

Maximizing Earnings Per HIT

As I’ve said many times before working for Amazon Mechanical Turk (Mturk) is not an easiest way to make money online. In fact, it’s slow and tedious. And if you’ve read any of my past posts, you read between the line (or maybe I’m more obvious than I think), but I’m not a big fan of the low-paying/high volume strategy.

I was reading some posts on the Turker Nation forum where some Mturk providers discussed their average reward per HIT. Here’s some excepts from the forum:

“My income per HIT: $0.270.”

“$4595.53 / 40550 = 0.113”

“183.16/796 HITS = .23 per HIT”

“$3,604.75 / 243,920 = 0.014”

On another post, I read of a poster who advocated a low-paying/high volume strategy. This poster like doing a horde of low-cost/low-thought HITs. Here’s an except from his post:
“All told, in the span of almost a year, I've done about 25k hits and made about $530. I try when possible to do the 3, 5, 10 cent hits (to me, those are "big money"), but the one centers are my bread and butter -- literally!”

My operating philosophy for Mturk is diametrically opposite to this poster. I take almost exclusively writing HITs and I like maximizing my earnings. To illustrate my point here are my numbers from the past 30 days:
Past 30 days -- 216 HITs = $348 -- Average per HIT: $1.61

Now, I certainly can’t quantify my time spent compared to this poster’s time spent, but I can point out that I made nearly 2/3’s of what this poster made in one year in only one month of work. I have averaged around two hours a day working on Mturk and made over $10/day. I don’t know who has the more tedious job, but I can say that I learn something from my work on Mturk -- along with making money.

HITs to Avoid on Amazon Mechanical Turk

The world is hard enough making money online and with the low wages on Amazon Mechanical Turk (Mturk) it’s even more difficult. To add insult to injury, there is a small contingent of requesters that want to exploit the workers of Mturk in the worst possible sense of the word.

But how do you know who to work for and who to avoid? Well, below are simple guidelines to follow.

You should avoid HITS that involve:

• Secret shopping
• Testing a website that requires that you enter personal information
• Testing software - I recommend you avoid any HIT that asks you to download and test software. The risk is too high (you don’t know what you’re getting) and the reward is to low.
• Free trials that ask you to sign up for a service
• Surveys that request you input personal information (Although request from academic sources and have a human subjects disclaimer are usually good to go.)

The above HIT examples to avoid are usually ones posted by scammers. There is a big risk that you will be opening yourself to any number of problems, plus you very well might not get paid at all for taking on this risk.

If you want you can review my two previous articles on Turker Nation, a forum for Mturk workers, and Turkopticon, an online database with reviews of requesters by Mturk workers.

In many cases, these types of bad HITs should not be on Mturk at all since they violate Amazon Mechanical Turk’s Terms of Service, but review is lax and enforcement is almost non-existent. My recommendation is that you should avoid any or all of these. The risk/reward ratio is just too skewed to the risk side.

Writing for Dollars on Amazon Mechanical Turk

When it comes to making money online with Amazon Mechanical Turk (Mturk) there are definitely different strategies and philosophy for making money. In an earlier post I talked about a low impact strategy of taking on low reward HITs in high volume. In this post, I will flip this strategy on end and discuss high reward/low volume strategy with a focus on writing.

Two Criteria
The first two criteria for taking on writing HITs on Mturk are:
One - you should be able to write. That doesn’t mean writing Pulitzer quality material, but you can string a few sentences together and make sense
Two - you have to want to write. If you want to work for this low of wage, you should at least enjoy it.

To SEO or Not to SEO
There are a multitude of writing HITs on Mturk. Many are ones that you require you to write SEO (search engine optimization) targeted material. You’re basically given a topic and have to include a phrase about that topic two or three times in the article. These articles usually have to be between 350 and 600 works, but sometimes requesters want 1200 words. Rewards vary on these and sometimes you can make $3.00 on these HITs.

There are also HITs that aren’t SEO related. These work pretty much the same as the SEO articles, but aren’t as stringent in the phrasing of phrases and words.

General Writing Assignments
I tend to prefer general writing assignments. If I can get into a groove, I can knock out three and sometimes four articles in an hour with average pay between $1.70 and $1.90 per article. If I do hit that sweet spot, I’m making over $5.00/hour.

In the past month alone I made over $300 taking on writing assignments in my spare time. And I mean my spare time. I usually write an hour in the morning before work, then over my lunch hour and maybe do a couple in the evening. If I can get six completed in a day, I can average $10 a day and that gets me to $300 a month.

Speed is the key to making this work. You have to be able research and write quickly.

Research
If you get bogged down in the research, you’ll be shooting yourself in the foot. I basically do a quick Google search of the topic and then find a primary source to work from. That source needs to substantive. If there’s only a couple sentences you’re going to run into trouble stretching the story to the required word length.

Conversely, if you have too much source material or the material is dense and complicated, you can lose efficiency. I’ve abandoned many HITs because the topic was just too complicated to summarise quickly.

Writing
Once you’re ready to write you need to get the words down quickly. Don’t spend a lot of time looking for the right word. Remember you’re not being paid enough to win an award, just to get the facts down in a logical and readable fashion.

Plagiarism Protection
Do not plagiarize material. Do not cut and paste from your source. Most requesters run the writing assignments through plagiarism checkers and your work will be rejected. Even worse, you will run the risk of being banned from taking anymore HITs from that requester. Take a few moments to run what you’ve done through a plagiarism checker before you even submit it.

I’ve used a plagiarism checker in the past and have caught some of my original writing that was too close to an already existing source. When this happens I have had to just do some small re-writes to make sure my work doesn’t seem too much like something else that is already out there.

My favorite plagiarism checker right now can be found at dustball.com.

Final Proof
If the article is short, I recommend that you give it one last proofing before you hit the submit button on the HIT. I’ve caught more than fair share of mistakes in my writing that just needed small tweaks.

Writing my not be your thing, but it works for me and I think that if you compare it with the low-reward/high volume strategy, writing will be pay better dividends.

300

No, it’s not the movie with all those Spartan warriors. It’s my first month working for Amazon Mechanical Turk (Mturk) in which I have eclipsed the $300 mark in earnings. Woo-woo, for me.

In the past, I was a pure hobbyist when it came to Mturk. I dabbled around taking HITs enthusiastically for a couple days, then lost enthusiasm, going through periods of waxing and waning with productivity with Mturk. It was only recently that I decided on a more systematic and professional approach to Mturk.

I set a goal of earning $10 a day and averaged that throughout the month even though I had a couple days in which I made very little. To offset those low earning days, I put in a couple long weekend days in which I pushed my total earnings to over $20 per day.

Now, if you’re envisioning me sitting hunkered down over my computer from sun-up to sundown, then you’d be wrong. I have a fruitful and busy day job. I do Mturk to make a few extra bucks in my spare time. Yes, on the high earning days, I did spend between three and fours a day working on Mturk HITs, but most days have me working 90 minutes to two hours. I tend to get up early before my family is awake and work for one hour on Mturk. Then I do some HITs over lunch. If I meet my quota, I may stop for the day, but if I have a few extra minutes in the evening, I might fit in one or two HITs before I go to bed.

I’ll be writing an article soon that discusses my strategy in more detail. For now, I just hope the attractive and well paying HITs just keep coming and I can maintain my new found discipline.

Who Works for Mturk and Why - Part IV

Here we are back in the rough and tumble world of making money online -- little bits at a time with Amazon Mechanical Turk (Mturk). If you’ve been following my series Who Works for Mturk and Why, you’ve now come to my fourth and final installment. In my last article, I wrote about data collected by researchers from Mturk workers, focusing on the marital status of workers and how much time they spent working for Mturk.

In this article you get the big WHY for workers on Mturk. What motivates people to slave over a “hot” computer for hours for just pennies an hour.

Well, there are a lot of people with a multitude of reasons out there. For the purposes of this article, I’m going to attempt to stick data from U.S. worker as much as I can because the differences in standards of living when comparing the U.S. to other countries are striking and the reasons are pretty obvious why foreign workers do work on Mturk.

The Big Why
I sometimes wonder why anyone would work for Mturk, but then I just look in the mirror and understand.

The research I’m using as the basis for this article put some probing questions to Mturk workers and found out some interesting facts. Nearly 70% of U.S. workers stated that they found working for Mturk to be “fruitful way to spend free time and get some cash.” But conversely, they also said that tasks on Mturk were not fun. So, what I read between the lines is that Mturk is not a lot of fun, but it’s not too challenging of a way to make some extra money. Many (60%) report that Mturk is a good way to kill time and make a little money.

Here’s comes a telling figure and I will contrast U.S. and India data this time. The question was: Is the money you make on Amazon Mechanical Turk your primary source of income?
India -- 28% U.S. 12%

The follow-up question asked if Mturk was a source of income for less than essential purchases.
India -- 38% U.S. 61%

So, it’s not hard to glean from that that a lot of U.S. workers are using their Mturk money to buy “toys” or other non-essentials. While workers from India actually see Mturk as a substantial and essential income.

Thus ends my series on the “who and why.” Feel free to let us know your why and who?

Who Works for Mturk and Why - Part III

Workers trying to make money online with Amazon’s Mechanical Turk (Mturk) are not few in numbers. I’ve always been curious as to why someone would choose to work for Mturk and who they are. Fortunately for me, there is someone out there doing research on these very issues. I hope to take the insights of these hard-working and generous researchers and share them with you. By the way, I will spend most of time focusing on statistics of U.S. workers for the purpose of this article.

In two past articles, I talked about who the workers for Murk and why they do this tedious and low paying work. The first article covered where workers lived and in the second article I discussed some demographic information including age and educational ranges and income levels of worker. In this article, I want to share with some of the findings in terms of the marital status of Mturk workers and how much time they spend taking HITs.

Marital Status of Mturk Workers
The vast, vast, vast majority of Mturk workers are single. That category hits 35%. But around 13% of workers do report being married and a similar amount reported being married with children. Also, the majority of these married workers reported having two kids. My take is that married folks with kids are feeling the pinch of the economy and are looking at places to earn a few extra pennies around a busy life and work schedule.

Time Spent on Mturk
As far as time spent working for Mturk, the biggest response was 4-8 hours spent a week working for Mturk with a response rate of around 25%. 2-4 hours was the second most reported response at 21% and 8-20 hours not far behind at 19%.

I thought I should report my details in these areas. I would be in the married response area and also would have to select the response of 2-3 children. As for time spent, my time does vary. Lately, I’ve found HITs I can tolerate and can make tolerable earnings with. So, I would say that I spend about 10-12 hours a week working on Mturk. Now, if you asked me two months ago, I would have to had reported 2-3 hours a week at best. I’ve found a groove and am mining it for all I can get. At least, for now.

I will wrap up this series next time with an article on the motivations why people work for Mturk. Hope to see you back.

Who Works for Mturk and Why - Part II

In my introductory article on who works at Amazon Mechanical Turk (Mturk) and why, I just brushed the surface of this topic talking about where Mturk workers live and some basic rationales for why some of these folks work on Mturk along with talking about the gender distribution for Mturk workers.

With this article, I’ll go more into depth and discuss the age distribution of works, their educational levels and income levels. If the data provides this info, I’m going to focus on U.S.-based work data.

Age Distribution
Not surprisingly, there are a lot of 20 and 30 somethings work on Mturk. Roughly about 40% of Mturk workers hit this range. These folks are in that age range that are computer savvy and might be in a position where they need some extra income.

BTW: About 4% of the folks on Mturk are 60 plus.

Educational Levels
There is somewhat surprising data here. The vast majority of Mturk workers have college degrees. Over 30% of workers report having a bachelor’s degree with around 12% having a Master’s degree. Those having some college, but no degrees came in at around 25%.

My gut would have told me that most of the workers would not have degrees, but obviously my gut can’t deny the hard data.

Income Level
Remember, I’m focusing on U.S. worker numbers here. Still I found it somewhat surprising that the two largest income ranges reporting that they work for Mturk are $25K to $40K and $50K to $60K. My suspicion would have been that lower income folks would be doing this work, but maybe they are busy making a better hourly wage. My other thought would be that workers in the two most prolific income ranges could be a tell about how tight the U.S. economy is during these recessionary times.

Look for Part III of this series when I discuss the marital status and the time spent working for Mturk.

Who Works for Mturk and Why - Part I

When trying to capture the essence of the people who work for Amazon Mechanical Turk (Mturk), you quickly discover that a lot of different people work for Mturk and for a lot of different reasons. This article will the first in a series of article I present on research looking into the people who work on Mturk and why.

It’s reported that many of the people working for Mturk are from overseas. India and the Philippines are two likely sources for workers for Mturk. The cost of living in those countries are so much lower than in the United States. So much so that some of these workers can make a decent money.

That’s not so true with the U.S. Mturk worker. Of course, that depends on your definition of what “decent money” is. For those out of work and need some income, then just about any source of income is decent.

I’m by no means and expert on this subject and I have only done just a small amount of research on the topic and I really owe much of what I know to others like the people at CrowdFlower.com. They are a labor-on-demand provider and also study entities like Amazon Mechanical Turk.

CrowdFlower conducted a survey and their findings were interesting. Now before I go into that I must say that their survey sampling seemed somewhat low with only 67 respondents.

Now with some of their findings:
> India placed first as to the number of people who work on Mturk at 46%
> U.S. Workers were close at 42%

Another researcher completed his own survey of Mturk workers in early 2010. His findings were similar, but did show some differences and also provided more details.

This second survey showed that U.S. workers dominate Mturk and not workers from India.

> U.S. workers 46%
> India workers 34%

This second survey also showed that in the U.S. females made up the majority of workers in Mturk at over 60%. In India, males made up 70% of the Mturk workforce.

The researcher posits that the reason that females dominate the U.S. Mturk workforce is because most of the U.S. Mturk workers look at Mturk as a supplemental source of income. In many cases, they are stay-at-home moms who can take small amounts of time and devote it to working for Mturk.

Tune in next time when I talk about some more in-depth data on Mturk workers. Maybe you will find yourself in there somewhere.

Taking HITs: The Low-Paying/High Volume Strategy

Making money online with Amazon Mechanical Turk can be a time consuming drone -- especially on the low value HITs. The work can be grueling and tedious at times. Some workers like to take on a smaller amount of higher paying HITs while others like to have a mass harvest of a long series of the same type low-paying HIT. I like to call this strategy the Low-Paying/High Volume Strategy.

This strategy basically consists of finding a repetitive HIT that has a large number to pull from, learning what it takes to complete them, and then putting your head down and churning them out.

I’ve done this before. In about 90 minutes one day, I completed about 120 HITs and was $1.20 richer for it. Woo-woo But that’s not really my cup of tea, though.

This article will focus on that strategy. For some people, they really don’t look at Mturk as a place they want to put on their thinking cap. They need or want to earn a few extra bucks and don’t really want to have to work up a mental sweat.

Knocking down a long series of low-paying, uncomplicated HITs can be relaxing or so does one Mturker report. He said:

“For the most part, I like simple, plain, straightforward tasks. The repetition relaxes me, in a sense, but at the same time drives me to see how many I can do before they run out.”

This strategy might work for someone who has sedentary job in which they might have to monitor something on a computer or watch an area like a night watchman. It’s sort of like double-dipping because you’re making a wage from your job and also making money on Mturk.

There are some downsides to this approach. First, if you have a requester who rejects your long series of HITs, your rejection rate is going to skyrocket. Also, if you mistakenly get something fundamentally wrong with the HIT, you’re setting yourself up for a whole slew of rejections.

The other downside are slow payers. For the Mturker who likes this strategy of low-paying/high volume working, slow payers can be quite frustrating.

Anyway, for quite a few workers, this strategy is the only way to go. Set up the cheap HITs and they will knock them down.

With Amazon Mechanical Turk, Avoid Rejection - At All Costs

Rejection in real life, be it for a job or asking for a date, can be painful, but rejection in the world of Amazon Mechanical Turk (Mturk) can have some serious effects to your earning potential. That is if you’re serious about working for Mturk for the long haul and trying to make some money online.

If you working on Mturk for any amount of time you’re likely to encounter a HIT that makes you think twice. It could be badly worded such as including contradictory instructions or be poorly worded. This should be your red flags to steer clear.

You see, your earning potential on Amazon Mechanical Turk is tied to your Acceptance rate. It’s really better correlates to your Rejection rate. To be able to access the greatest amount of HITS on Mturk, you need to have a high acceptance rate. If you drop below 95%, you will see a dramatic decrease in the amount of HITS available to you.

It turns out Requesters do have standards after all. They don’t want just anyone working on their HITs.

My example of this is a sad one because it happened to me. Early in my Mturjk days, I found a quick and easy series of HITs and while they didn’t pay a lot, I could plow through them at a speedy rate. I read the instructions and while I felt there were some contradictions in them, I proceeded anyway.

Big mistake.

The HIT did have contradictory instructions. I was categorizing some-such thing and they wanted the worker to select a categorization in an “either/or” fashion, but they didn’t design the question criteria in the correct way. It was sort of like asking a question about a flower and saying it is either red or blue --- or green or white, but it can’t be both, but you have to provide answers to both. So, with each subsequent question, I was building a slew of contradictory responses...and rejections.

My acceptance rate dropped below 95% and I saw a precipitous drop in the number of HITs available to me once I dropped below that 95% line. It takes doing a lot of HITs over a lot of time to move that needle back up above 95 depending on how far you’ve dropped down. It’s sort of like tanking your first term in college with a C, 2 Ds and an F. It’s going to take moving a mountain for a long time to get get that old GPA back up into a range where your parents will send you beer money again. If ever.

My advice is that you carefully read all instruction for a HIT. If you come across a red flag, read them again, this time very carefully.

I’ve taken a very conservative approach to low-pay, mass-issued HITS. I only do a couple and then submit them. Then I wait to see if they are accepted or rejected.

If I lose out on some HITs, I guess it’s worth the risk avoidance. That's just my humble opinion, though, but enter iffy HITs with caution.

Low Impact Strategies for Amazon Mechanical Turk

So, you want to make some money online and you’ve decided to become apart of Amazon’s online version of a modern sweatshop? And you know what you’re in for? (Read previous posts to learn more.)

There are two ways to approach Amazon Mechanical Turk (Mturk):
1) You can look for low impact/low-cost HITs that require little or no thought and pump them out one right after the other in a near-mindless state of conscious (or as you watch TV which is may be the same thing)
2) You can look for more involved, higher cost HITs that require some concentration, time, research, and eventually effort -- like writing assignments.

This article will focus on the low-impact approach.

Okay. How can you do this work in a low impact way that doesn’t cause you to work up a sweat and allows you to continue watching your reality TV show with your laptop in, well, your lap?

First, determine if the HIT is actually a low impact HIT. This may take some concentration, so wait for a commercial, open the HIT, and read the instructions. And that means reading all the instructions. The devil is in the details and some of the low cost HITs that look low impact may not really be.

For instance, I just reviewed at a $.10 HIT which initially looked quite easy. Get an address and hours of operation. Easy, right?

No, not really because the Requester also wanted the Worker to provide a half dozen key words on top of something other information.

So, it’s important to read all the instructions before committing to doing a slew of mindless HITS. If you get an important detail of the instructions wrong and then complete a bunch of HITs wrong, you’re in line for a whole slew of rejections. Rejections are really bad for your AMT profile because a high rejection rate can lower the amount of HITS you can have access to. (I know this all too well as I let my rate dip below 95% which excluded a great deal of HITS from me.)

If you do find a low impact HIT and you understand the instructions thoroughly, then you can go to town and in an assembly line fashion, pump those HITS out, one right after the other and watch the pennies pile up. And that’s what happens, you make pennies, but a lot of pennies can add up.

You have to be very fast and efficient to make this approach work for you. If you get bogged down, then you will see your effort transition to low impact to drudgery with a very low rate of return.

My advice is to set a (low) hourly wage floor (say, $3.00/hr. -- this isn’t rocket science and you shouldn’t expect rocket science pay). This is the lowest rate you want to work at and then do a few of the HITS. Then do the calculation as to whether you can hit or exceed that hourly wage floor by doing dozens of these HITS in an hour.

There are many people out that make this approach work for them. It is a good way to get lots of HITs completed and make a few dollars. It’s not the way I approach Mturk, but I’ll talk about my approach at a later time.