Life of a Data Labeler
scale.ai
scale.ai
Although pointless tedium is not the most important factor in rating a job, I think data labeling could well be along the most pointlessly tedious jobs ever created.
The broader point, though, is while people in the West bemoan outsourcing to poor countries as exploitative etc (and it can be!), even these "shit" jobs can be transformational. BPO has propelled literally millions of people into the middle class in both India and the Philippines.
Yeah, I mean I sort of get it. But if I really had to choose between the mental tedium of labeling data (which I end up doing quite a bit of myself) vs. backbreaking outdoor work, or something like repetitive food service prep... I’d pick the latter every single time.
At least for me, data labeling is just brutally mind numbing and draining.
In developing countries, status is an important element which people in developed nations may not think about. For example, how many derogative terms can you rattle off for people who labor in the heat of the sun all day? In the U.S. we have "redneck." There are worse terms for people of other races and nationalities in the region. In the U.S. these terms are a bad joke, in the Philippines, this is life. In the Philippines, skin whitening products are a big seller because dark skin is associated with the poor and laborers (the "Mestizo" has light skin which traces back to relatively well off Spanish heritage.)
You might choose outdoor work, but you have to understand the culture to understand why their choices might be different than yours.
Are you saying that the tedium itself is engineered to be present? I mean, the work is definitely not pointless. It is as pointed as it gets, as significant as a miner in a mine, a farmer in the field. Monotonous, tedious, yes. Pointless? No.
Now, the idea that the tedium is engineered in would be interesting. That would suggest it could be engineered out. Which is very interesting. You solve that problem, I submit there are some HR folks who would like your contact info.
I’m not sure how I would compare it to traditional blue collar low skilled labor jobs. Farm hand work, for instance, in addition to being physically difficult, is also surprisingly tedious contrary to what some people may imagine. But I think it is easier to understand why it is important work. If you don’t pick the berries, the berries die and nobody gets to eat them. If you’re a machine learning data labeler, you may believe that it’s creating a path to super intelligent computers. Or you may believe it’s a big waste of time. My acquaintances who did Alexa data labeling tended to lean towards the latter.
One interesting side to this is citizen science; websites like Zooniverse where you can get random members of the public to do your labelling. Technically there's little difference to what Scale does, just that it might be pictures of Pelicans or galaxies rather than cars on the road. In my experience what you're paying for with Scale is consistency. Zooniverse relies on the public being... Gaussian in quality - we needed something like twenty annotations per image I think.
People spend hours going through data and they don't get paid. On Zooniverse most of the data is boring, but the fun is the prospect of finding an unusual image. Some of the datasets are genuinely interesting (like the old ship logs). Many people say they enjoy the idea that they're contributing to science.
> Although pointless tedium is not the most important factor in rating a job, I think data labeling could well be along the most pointlessly tedious jobs ever created.
I don't want to pass judgement either way, but it's an interesting debate. These jobs have been around forever - data entry, for example, used to be (still is?) a fairly common unskilled job in first world countries. And tedium aside, you can at least work from home if you have a computer.
But where is the line? A lot of work I've seen on MTurk is transcription, for example. What about translation/subtitling services? These are labelling tasks in some way or another.
Let's talk about Venezuela for a second. ~75% of the population lost >19lb in body weight in a year according to this survey: https://www.upi.com/Top_News/World-News/2017/02/19/Venezuela... It's unbelievable that we haven't figured out how to prevent people from living like that in the 21st century. I think it's totally deplorable and honestly an affront to humanity.
You know what I think the most effective way to combat large scale poverty is? Not by working for an aid agency (we've all heard the horror stories) - instead, how about making those people economically valuable? The internet is an amazing way to reach those people - and guess what - Scale is actually doing that (evidence: these stories). If we continue to grow, we'll be doing that even more.
To be clear, this isn't the main mission of the company - but I don't see how anyone could think it isn't a great side effect. Hence the title of the blog post - positive externalities.
The outright dismissal of OP's criticisms/comments doesn't allow for a discussion in that data entry is a very tedious job along with maybe there are people who are unhappy with the work but very happy with the pay (which I think is perfectly okay).
The comment reads as "I'm going to tell you off" with the `- and guess what -`.
But I do see a lot of data around company engagement.
There’s basically no chance this level of positivity is ubiquitous for them
I guess what you're asking is - how many of these people find their job tedious? I will say it's a lot less tedious than you might imagine as a first impression - e.g, if you see something weird in some data, you talk to other people about it; if you get good, you train people; when doing a new project, you're learning from your coworkers; if you get really good, you might be asked to help develop training materials, etc. So there's a lot of interacting with other people, and a sense of community. For me personally, I honestly find it meditative to label a lot of data - it feels kind of like tending to a large garden, maybe even fulfilling some deep OCD/obsessiveness desire. Some of our labelers find it meaningful that they're contributing to robotics / self driving cars - e.g, the last interviewee in the blog post.
Back to your question though - how many of these people find their job tedious? I'm not sure how to ask the question to them in a way which gives a satisfactory answer. e.g, I'd expect if we just asked "do you feel like your work is tedious?" the answer would be dominated by people's realistic alternatives, and wouldn't have much to do with the job itself. (so we'd get similarly positive responses) If you can think of a better way to frame that question, I'm happy to ask it and post the responses here :-)
Given that all of Scale's open positions, and current FT employees presumably, are in San Francisco (and not remote), a pessimistic view would be that the data labelers are subsidizing Scale's lifestyle choices.
To clarify, there is no hard numbers around the quantity of people performing this task, for how long they are employed, their average working hours per day, the median satisfaction rating, the median pay per employee, etc. It feels as if only some happy examples were picked.
If you think about bias in ML models that we've seen in recent years for example: if you let people label 980 images of old white people and 20 diverse ones that's still a heavy bias, even if they spot and label the 2% outliers correctly.
https://medium.com/kiwicampus/how-kiwi-empowers-students-in-...
What's the median hourly pay for a Scale data labeler? What fraction of their employees enjoy working for Scale? What's their six month retention rate? Amazon's Mechanical Turk has a reputation for grinding people up and burning them out; what makes Scale better?
The labeler side of scale is hosted at "remotasks.com" and they run IP checks to try and ensure that only people from certain countries can sign up to label data. Also a Facebook account is required for signup...
... while of course not providing with a stable regular job, and not necessarily doing much to strengthen the local economy in a fundamental sense.
I tried to pick the answers which were most well written, not the ones which were most positive/negative. Personally I do think this is representative of the labelers who've stuck with us, but at this point don't have the energy to argue this; hopefully my other comment is convincing (i.e, there isn't a high bar to making people happy when their next best alternative is a lot worse)