I think it's underappreciated just how much human labour goes into annotating data fro autonomous vehicles. Scale is a fast growing start-up, with 35 employees and around 10,000 contractors, many based out of India and Africa, who do data annotation for Cruise, Zoox, Lyft, and Nutonomy among others.
Given that this training data basically gets loaded into neural nets, etc., can it be said that in some indirect sense these data labelers are actually driving these cars? Are we still just building the mechanical turk?
To be it shows how far we still are from biological intelligent systems. I don't recall having to point out what an apple is to my daughter more than a few times. Watching children learn is breathtaking. Even much simpler brains like those of rodents or cats are far beyond anything built so far in silico.
Generally, neural networks do not memorize their training data. If they worked that way, they would not do well with samples they've never seen before.
So no, I would not say that the labellers are driving the cars. They do play an important role in the process of building these systems, though.
I just find the mental gymnastics fascinating that they are talking about how these people are living in abject poverty, and yet in the same article have the gall to highlight how people in the US are losing their jerbs. Which is it? Are we helping people in poor countries have a slightly better quality of life by working in an air conditioned office vs agriculture, or are we taking US jobs? It's one or the other.
You obviously can't save the jobs in one country and also create those same jobs in another country.
The only way for both sides to profit (over the long run) is for either country to do the job it can do the most efficiently, i.e. comparative advantage. In practice, that's a real tough concept for people to come to terms with. People want to have their cake and eat it too.
There are enough non-zero-sum effects in economics that this isn't plainly apparent to me. It seems like quite a claim, really. For instance, I find it plausible that cheap labor from overseas may actually help subsidize the creation of an industry, which might grow enough to need more employees in the US for related work.
>You obviously can't save the jobs in one country and also create those same jobs in another country.
Who would want to create the exact same jobs that fit the needs of a labour market in the poorest region of the world in the US? What's the point of having someone in the US who went through at least 12 years of education to manually select pixels on a screen?
Getting those jobs to people in Nairobi is good for them because it lifts them out of dangerous and hazardous jobs, and it would be a horrible place to work at in a developed country.
Do you have a source for that? While still in definite need of improvement, the highest number for illiteracy I can find is 40%, and that is for adults exclusively.
The job of tagging data for supervised learning did not exist 10 years ago. Now that it exists, the first people to be employed in this job are in low-wage countries. These guys did not replace anybody in the first world. They are not "taking US jobs". Or do you have somehow in mind that the US workers should have first dibs at any new types of jobs?
Its not open source but this tool helps in managing labeling by your in-house team or contractual workers. It also provides you annotation software for most of image and text usecases. Here is link to the same.
I'm surprised that not everyone has figured this out. You're doing free labor for Google's Waymo. Why else would Google's re-captcha have all of the photos include cars, bridges, and traffic signs and lights?
2. After Google acquired the Captcha service they initially used it for Google Maps where you was always asked to pick street numbers and store names. Only recently has it pivoted to Waymo use cases.
- By sending dollars to Kenya it's helping reduce inflation for the Kenyan govt. The biggest problem for many developing countries is finding a way to get their hands on the global reserve currency.
- Its allowing private wealth to be created (by women !!), empowering women have very positive cascading effects in a country (fertility rate drops, women have personal freedom to do other things ). Private wealth directly in the pocket of Kenyans is one the most efficient ways the developed world can help the developing world.
Normally giving cash to the govt results in looting of public coffers.
- Teaching useful computer skills to the average kenyan.
I don't follow the "very positive cascading effects" line. How is low fertility rate and women being free to do other things a positive cascading effect? Historically, it makes families more unstable.
Edit: I'm not saying women working is a bad thing. It's good on many fronts. But I feel the ripple effects are not positive.
It's not only developing countries that benefit from increased female participation in the labour force. For Western nations, the increased proportion of women in the labour force was a core component of the rapid increased economic growth between the 1940's and today.
Low fertility is an interesting thing where it comes to development; for most countries it has quite dramatically declined as incomes have increased. The effects are likely bidirectional (increased economic growth -> lower fertility rate; lower fertility rate -> increased per capita income).
The drivers are complex and interrelated: high child mortality rates, low life expectancy, and high poverty are associated with high fertility. It's an unjust comparison to make, however leading theories share common features with K-r reproduction strategies in ecology (with increased survival / longevity, investing in a few highly educated/skilled children becomes feasible). Similarly, increased education / maternal health is highly correlated with economic development and child wellbeing. These are often highly compounding effects over time - a virtuous cycle in which increased economic freedoms, health, education, and opportunities (male and female - often females are relatively impoverished, so there's the potential for bigger immediate benefits) can lead to rapid development with the right institutional conditions.
Lower, not low. Having 5-7 children per family makes it a lot harder to increase your economic position and education levels. It's a good thing when developing countries start lowering their fertility rates towards those seen in developed countries.
It's not bad but why do they feel the need to eat sh? They could have said just that "we pay 4x the going day wage and we want to maximise our profits" instead of that BS: "we don't want to disrupt the labour market so we keep the wages for their own good."
My guess is that the guy they interviewed spends his weekends hanging out with NGO types. How else would you get invited to any parties in Nairobi? And they love this stuff... saying "everybody profits" makes you sound like a cold-hearted evil capitalist.
There have been spontaneous demonstrations among the workers voicing their joy and gratitude at our happy new way of life. /s It seems pretty naive to assume that a handful of individuals being paid a fraction of the going rate for western labor is going to have a meaningful impact on the country's economy or lead to sweeping social changes.
All of Asia was dragged out of poverty - over time - by doing stuff cheaper. China is a world power not from communism or protectionism or by some grand gesture. They took awful jobs for a tiny amount of pay (by western standards) and eeked their way up the GDP ladder.
It is a the worst kind of white, privileged attitude that taking a higher paid job won't improve things for the locals.
But lets give this a fair airing. Name a country that had a GDP increase in any other way. I'll wait.
Scandinavia and the entire Soviet Union spring to mind instantly. There's nothing particularly white or privileged about healthy skepticism of any system that hinges on labor arbitrage to accomplish it's goals. Go meditate on the econ definition of the word "exploitation" until enlightenment is achieved. Note, if you're planning on trying to muster McCarthyite arguments at this point you're going to have to overcome the fact that the first manmade object and the first person in space were both put there by communists.
Sure, but the argument is specious for all of the same reasons that philanthropy has a track record of exacerbating the social problems it intended to solve.
Yes, that title was baity and misleading, so we've changed it to a more representative phrase from the article. A photo caption, actually; it's surprising how often those make better titles.
>> "But one thing that's critical in our line of work is to not pay wages that would distort local labour markets. If we were to pay people substantially more than that, we would throw everything off. That would have a potentially negative impact on the cost of housing, the cost of food in the communities in which our workers thrive."
What a joke! We don't to pay them more because they will be able to buy decent house/invest in infrastructure or god forbid even start a small business...that would totally distort local labour markets, right?
Yes especially the last sentences reads as "our workers only thrive in poor communities and we are doing them a favor by keeping them poor"
Western food aid has distorted the price of food in African countries already and resulted in local farmers losing their job. The housing market is only problematic in cities that restrict new construction. All these companies are doing is hiding their greed. They don't care about the well-being of the workers at all.
With that said they appear to be paying people 4.5x more money than they had been making before, so it’s hard to get mad at them.
I’m sure they’re in it for money and glory too but bringing modern employment prospects to Sub Saharan African slums is hard and it’s a lot easier to snipe from a distance.
In fairness to the article, it's quite clear about this:
"Samasource targets those currently earning around $2 a day, or less, in the so-called informal economy of odd - or dangerous - jobs. Samasource instead provides a living wage of around $9 a day."
Only after that do we get the dubious justifications. And the guy they're interviewing probably spends an unhealthy amount of time hanging out in NGO circles, where I'm sure this kind of thing goes down much more smoothly than mutually beneficial exchange.
If the going wage of an employee with some skill is X and you decide to pay some arbitrary amount above X, how do you fill your ranks? What's a fair system? You have 10 equally qualified people going for 1 job.
One way is to make decisions based on non-economical conditions, such as attractiveness or familial connection.
Another way is lottery, which I admit is more fair, but it's still not fair to those that are willing to do the job for a lower wage than someone who got the job. It's okay to compete for a job by promising to work harder, but somehow negotiating a lower salary is unacceptable?
And the other big problem with paying above market wages is that it often introduces a middle man. Since you're swamped with applications, you hire a firm to sort everything out. The firm takes a cut as well and normally ends up bringing down the employees wage to the market wage. On the extreme, if the job is so lucrative, the middle man requires payment or a portion perpetual of the salary for the job placement. This happens in cruise lines:
> Long hours and subsistence wages are part of their contracts, as is the threat of being fired without notice or cause. Yet people from some of the world's poorest nations are so eager for work that some pay middlemen the equivalent of a month's wages to get these jobs, a fee that violates international law.
Exactly if you're disrupting existing power structures either by "overpaying" employees (see recent PSG football leaks) or unfairly cutting costs (Uber/Taxis or Airbnb/Hotels), you're going to make enemies of a lot of angry people.
I'm not saying Samasource does not profit from this situation, they clearly do, but Silicon Valley taught me that money spent paying developers through the nose ends up mainly in the landlord's pocket.
>> Exactly if you're disrupting existing power structures either by "overpaying" employees (see recent PSG football leaks) or unfairly cutting costs (Uber/Taxis or Airbnb/Hotels), you're going to make enemies of a lot of angry people.
This is pure BS. What "power structures" are you talking about? Well paid jobs bring money which is spent in the local economy. I doubt the nearby shop owner will be angry because now you can buy food instead to beg for it.
The Uber/Airbnb model is a totally different issue because they compete with the local market and enjoy unfair advantages which is not the case here.
Couldn't disagree more. If anything, more money empowers and helps them fight against injustice and perhaps even develop small businesses / create more wealth. More money paid into taxes improves the infrastructure(i.e. security) as well.
In the worst case scenario security can be bought too. Money gives you power, especially in poor communities.
Although a much less extreme example, this brings to mind GitLab's model of location-based pay [0]:
"Market rates for roles are different for different regions and countries. We pay market rates instead of paying the same wage for the same role in different regions."
Isn't this what happens in SF an SV? Big tech money certainly 'disrupted' the local housing market to name only that facet of the economy and social life.
So the article is right about the local disruption, but who are they to decide what's good and what's not for local people who deserve a fair compensation?
If that were the reason, then using the wage savings they could fund a local charity to run infrastructure projects - build schools, say; maybe wind/solar farms owned by the local communities.
That way the whole country will develop and they'll be able to pay their workers more without distorting things anymore than is the case elsewhere.
Depressing wages locally will just lead to economic migration and brain-drain, surely? Ultimately that seems much worse.
My heart sunk on reading that. Overall, the whole initiative is positive. But given the working conditions lightly touched in the article, this is amounting to pure exploitation
Is the person who assembled your phone or sewed your T-shirt exploited? Whether or not it is exploitation, it is capitalism, and it has worked this way since the beginning.
I was literally cringing when she spouted this nonsense. If that's the case, maybe they should do the same in the Bay Area as well. It won't distort the housing market etc.
Agreed, that is a particularly offensive line of bullshit. Labor arbitrage is a thing, sure, but it takes a special flavor of insipid shithead to try to unfurl that kind of tortured logic to handwave past the obvious exploitation taking place.
How is it exploitation? These are some of the best jobs available to people coming from the slums.
The real reason that they don't pay more is of course: They don't have to. That's not what the white media people (desperate for yet another "exploitation" story) want to hear though, so you gotta make up some "social" reason.
If these people were paid more, the whole business would just move to the next cheapest place, possibly in another country. That's the peril of unskilled labor everywhere. You can't fix it by fiat.
>That's the peril of unskilled labor everywhere. You can't fix it by fiat.
Actually you can. Forbid companies to outsource labor outside their target market country, or impose heavy tariffs, and voila.
That will mean less cheap gadgets in said market country (e.g. US), but more actual jobs, and a healthier middle class (and thus economy), and thus better access to necessities.
This will also force third world countries to actually become competitive in quality and delivery, not just throw sweatshop-like labor (including from children and in some cases, slaves) and cheaper dangerous working conditions at the problem.
There is an overwhelming consensus among people who study the subject that protectionism has a negative effect on economic growth and economic welfare [1][2][3][4][5][6]. Even interventionist economists like Paul Krugman strongly agree, and they have tried to explain why free trade is a good thing to an ignorant public. It is shameful that so many people feel obligated to opine strongly on a subject which they don't understand, causing bad policies to be enacted.
Protectionism makes society worse off. The policy you are proposing would make the world poorer.
>There is an overwhelming consensus among people who study the subject that protectionism has a negative effect on economic growth and economic welfare
Yes, establishment economists all agree on establishment practices are good and want more of them. News at 11.
It's not like economics is a science the way physics or chemistry are.
I think everyone agrees that outsourcing work to third world economies creates immense wealth in first world economies. We also know that burning fossil fuels is the cheapest way to make useful heat. That doesn't mean it doesn't have a lot of negative unintended consequences.
> I think everyone agrees that outsourcing work to third world economies creates immense wealth in first world economies.
It also creates wealth in those third world economies. In fact, they benefit the most.
> We also know that burning fossil fuels is the cheapest way to make useful heat. That doesn't mean it doesn't have a lot of negative unintended consequences.
Can you explain why you think this comparison is relevant?
Hollowing out the lower/middle class in those first world economies and forcing people into social welfare programs is a big unintended consequence here.
To be fair, it almost entirely transfers wealth to the extremely well off by any standards first world upper class, everything else is a rounding error. Outsourcing is colonialism by another name.
> To be fair, it almost entirely transfers wealth to the extremely well off
Wrong. Free trade benefits the poor the most. It is responsible for massive reductions in poverty at the global level, and is critical to the economic growth of developing countries. Trade barriers (particularly in agriculture) are a major source of harm to the world's poor:
Look, there's really no reason to call me closed-minded, we can both surely cite a lot of reputable sources that agree with our position, mine being that outsourcing causes a lot of problems and yours that outsourcing is a net positive on global gdp and increases incomes globally. I think we're both right! But what you may not be considering is - what do you think would happen if 8 billion people had a first-world carbon footprint? So what's the rush to get new labor pools into the global economy? Call me cynical but I think it is almost completely self interest on the part of the first world economies with only lip service paid to the well being of the poor Laotians.
I asked what you meant by colonialism, so enlighten me. Explain why you think outsourcing is "colonialism".
> But what you may not be considering is - what do you think would happen if 8 billion people had a first-world carbon footprint?
That's a discussion for another day. But you would have to reframe your argument as "Free trade is bad because the world's poor escaping poverty would be a bad thing for the environment."
> So what's the rush to get new labor pools into the global economy?
Simply put, extreme poverty.
> Call me cynical but I think it is almost completely self interest on the part of the first world economies with only lip service paid to the well being of the poor Laotians.
Of course it's self-interest. That's how economies work. Whether it's self-interest or not doesn't determine whether it's good or bad.
My entire argument is "there are a lot of unintended consequences" - naming an example is salient. I noticed that you haven't refuted any of my points so I'll assume that you are aware that global trade is worse for the environment and bad for the working class in the countries exporting their labor pool, but the benefits are so good for you that you will continue to argue in its favor. That's fine. We will clearly choose to look for different policies from our leadership.
As I've said elsewhere, the solution to pollution is more economic development, not less. This becomes even truer once demographic transition effects (like declining birth rates) start to kick in.
> and bad for the working class in the countries exporting their labor pool
Can you not read? The poor in exporting countries benefit the most from free trade. That was the main point I addressed in my comments, and the fact that you pretend it never happened makes it clear you're not arguing in good faith.
> We will clearly choose to look for different policies from our leadership.
You'll be making the whole world poorer with your harmful policy. And it's all due to an ignorance of the subject, an unwillingness to examine the evidence, and a resistance to changing your mind in the face of it. Shame.
That's not true, as I've already pointed out in my comments. The fact that you keep covering your ears and pretending this is not the case doesn't change reality.
All rich countries, except a few oil producers, have become rich by sweatshop-like labor.
It's the one way that actually works. By banning it, you are not forcing third world countries to find some other way, you are forcing them to remain in poverty.
>All rich countries, except a few oil producers, have become rich by sweatshop-like labor. It's the one way that actually works. By banning it, you are not forcing third world countries to find some other way, you are forcing them to remain in poverty.
Rich countries have also used feudal ownership, slave labor, and colonies. And all allowed child labor. Maybe we should allow those too?
The "way that it actually works" depends on what people tolerate and find acceptable. There's nothing written in stone, societies can drive their fate, not the other way around.
Why would "forc[ing] third world countries to actually become competitive" happen at all, if it had not happened before outsourcing was a thing? I would think these workers would just go back to "earning around $2 a day, or less, in the so-called informal economy of odd - or dangerous - jobs." If you think they are acceptable collateral, at least say it in as many words. If that's not it, please explain.
> Forbid companies to outsource labor outside their target market country, or impose heavy tariffs, and voila.
This is nonsense economics, straight out of Donald Trumps playbook.
It hurts everyone involved in the process. If unskilled labor becomes too expensive, it's simply eliminated. Business looks elsewhere. This hurts the people in the rich country less, there are more opportunities for people there. The people in the poorer countries stay poor, because they don't get the kind of foreign capital they need to advance. You know, the kind of money that buys things on the global market, the kind of money that the US just gets to print.
> That will mean less cheap gadgets in said market country (e.g. US), but more actual jobs, and a healthier middle class (and thus economy), and thus better access to necessities.
Not just "cheap gadgets" will be more expensive, everything will be, because the whole economy is intertwined in subtle ways. This makes everyone poorer, but it especially hurts those people that can afford the least. They'll be able to afford even less. You're not helping a middle class by raising prices on low-end jobs. Those jobs will disappear when they become too expensive. They can not become middle-class jobs by fiat.
> This will also force third world countries to actually become competitive in quality and delivery, not just throw sweatshop-like labor (including from children and in some cases, slaves) and cheaper dangerous working conditions at the problem.
If you have sweatshop-like conditions, which I don't think is the case here, then that is still a better alternative than whatever other jobs these people could've had instead - otherwise they would do those. Remember, the government can't just decide everyone gets to have a good job. Isolating these third world countries from the global market doesn't help them become more competitive. It's not like it hasn't been tried, mind you.
>This is nonsense economics, straight out of Donald Trumps playbook.
That's a facile response. The truth is that tariffs and similar restrictions have been used by every major economic power on its way to the top, the US perhaps more than others:
"Britain was the first country to successfully use a large-scale infant industry promotion strategy. However, its most ardent user was the U.S.; the economic historian Paul Bairoch once called it "the homeland and bastion of modern protectionism" (Economics and World History: Myths and Paradoxes, Bairoch)." [1]
Of course once they got there, they suddenly find protectionism "not working" anymore, and other countries shouldn't use it -- like someone who got to the top kicks out the ladder he used lest others use it too.
In other words, the interests of the 10% are taken as economic gospel (after all economists always belong to the 10% and cater to that crowd, especially anybody who's let anywhere near policy makers and top universities) -- and let the middle class and the bottom 30% be damned.
>Not just "cheap gadgets" will be more expensive, everything will be, because the whole economy is intertwined in subtle ways. This makes everyone poorer, but it especially hurts those people that can afford the least.
Everyone needs to be "poorer" when it comes to affording consumerism anyway (from smartphones to the tons of clothes [2]), and "richer" in affording rent, healthcare, education, job, and other such necessities, that is, the opposite of the trends for the last 30+ years.
>If you have sweatshop-like conditions, which I don't think is the case here, then that is still a better alternative than whatever other jobs these people could've had instead - otherwise they would do those.
The same could be said for child labor (better than the kids/families starving), and yet we outlawed that (at least in theory).
> The truth is that tariffs and similar restrictions have been used by every major economic power
...because it's politically expedient, not because it makes economic sense.
> after all economists always belong to the 10% and cater to that crowd
Oh, 10%. Why not 5% or 15%? What happened to the 1%? How do you even know that is true? I mean, you don't, but throwing around percentages and ascribing motivations to that is somewhat hard to falsify.
Seriously, let's say I'm an economist and belong to "the 10%". Why should I cater to them? Why not cater to the 1%? Why not cater to the 90%? What's so great about the 10%?
> ...let the middle class and the bottom 30% be damned.
So, the middle class starts at 90% and goes down to 30%, therefore they make up 60%. None of them are economists and no economist is catering to their interests either. Just to get a broad picture here.
> Everyone needs to be "poorer" when it comes to affording consumerism anyway (from smartphones to the tons of clothes [2]), and "richer" in affording rent, healthcare, education, job, and other such necessities, that is, the opposite of the trends for the last 30+ years.
Got it, poor people shouldn't be able to afford smartphones and tons of clothes. But how exactly does protectionism help them afford these other things you mention? Presumably these people are going to all have jobs making smartphones and sewing clothes (which are now so expensive they themselves cannot afford them). But why would their wages be higher? Remember, as prices go up, demand goes down. Without demand, jobs get eliminated.
> The same could be said for child labor (better than the kids/families starving), and yet we outlawed that (at least in theory).
Child labor becomes illegal only as soon as that is feasible, it requires a certain amount of economic development. Indeed, it is better for a child to work than starve, don't you think? Child labor (depending on how you define it) is still par for the course in underdeveloped countries, even when it may not always be legal.
The truth is, the annotation being done does not require much skill. If the cost of paying people rises too much, companies will just turn to mturk and get it done for cheaper (in academia we already do this).
Yes, and it's a shame because there's still plenty of value to be had from the article.
It's obvious that the end comment from the reporter is untrue, since by no definition could the workers be said to be becoming "AI experts" even if this is for some a step towards doing more challenging jobs.
It feels like it could be a combination of the reporters lack of knowledge, sloppy phrasing and getting carried away withthe enthusiasm with the situation.
While the main stream media is eager to portray fakenews as the gravest threat to free society, this type of clickbait headline writing essentially exposes their duplicity. So while the press like to pontificate from their ivory towers on their unalloyed noble purpose, they are ultimately human institutions, with all its ensuing fallibility, all its flaws.
"As it turns out, there is a utility in keeping us divided. As people, the more separate we are, the more politically impotent we become. This is the second stage of the mass media deception originally described in Manufacturing Consent. First, we’re taught to stay within certain bounds, intellectually. Then, we’re all herded into separate demographic pens, located along different patches of real estate on the spectrum of permissible thought. Once safely captured, we’re trained to consume the news the way sports fans do. We root for our team, and hate all the rest. Hatred is the partner of ignorance, and we in the media have become experts in selling both."
I understand the contention, but I don't think it's all that poor of a wording. In the end, who has more influence on the behavior of a neural network: The guy who wrote maybe a few dozen lines setting up its architecture, or the people who labeled the training data?
You should at least try to come up with an alternative headline that conveys to the average reader what is happening here, because "labeling data for machine learning" is not something that works either.
You should at least try to come up with an alternative headline that conveys to the average reader what is happening here, because "labeling data for machine learning" is not something that works either.
I have no inside info, but presumably it's a call-out to potential journalistic sources. The call-out emphasizes secrecy to assuage fears of retaliation by third parties not happy to see the news spread.
The reporters Edward Snowden reached out to almost missed out on the story because it seemed too much work to install and learn to use the secure communication tools Snowden asked for before saying what it was all about.
Click bait article like these are the reasons that Trumps branding of prestigious media houses as "Fake news" found resonance among the masses. Kenyans are not being paid to program driverless cars but to label data. There is a Himalayan difference in skillsets required for programming driverless cars and labelling data. BBC should no better. But why would they care? They got me to to click the article and increase their visitor count.
https://techcrunch.com/2018/08/07/scale-whose-army-of-humans...