Dawn of the Digital Sweatshop | Hacker News Reader

Dawn of the Digital Sweatshop

eastbayexpress.com

47 pointsanandkulkarni21 comments

I think the real problem here is that the systems (being built by people making far more than minimum wage) and beneficiaries of those systems (also making far more than minimum wage) haven't yet been able to extract enough value from HITs to pay a decent wage. If a worker can only produce $3/hr in economic value, there's no way he can or should be paid $10/hr.

Once a work system is good enough to provide 5 minute highly qualified answers to questions like "is this CT indicative of cancer?", there will be a lot of people making good wages on task marketplaces. I think we're just in a transitional period right now.

Most of it boils down to trust.

The main problems of these systems are that if you can get $3/hr of economic value from a single turk, it's hard to trust that you are getting accurate results without checking by repetition. This means that you could end up repeating the task 2, 3 times or more to get accuracy.

Scammers and bad workers are a big issue on any platform like this. I'm not sure how easily this can be solved - the workers only incentive to not cheat the system is the possibility of the requester rejecting their work and them not getting paid as well damaging their rating.

Whilst this seems like a good-enough way, there are enough lazy / bad requesters who don't bother rejecting work which skews ratings of workers. Also, ratings on mturk are lifetime ratings, rather than last n days, which isn't that indicative of recent worker quality.

I'd argue the solution to trust is appropriately matching work to reliable workers for whom the work is fairly paid, not simply raw repetition.

At $3 an hour, someone in (say) Lithuania would be highly motivated to do the work right, even if someone in the US wouldn't particularly care.

This is the approach we took at MobileWorks, and it works fairly well.

A possible solution could be for users to have an accuracy rating (perhaps short-term and long-term), and for the Turk API to take an optional minimum accuracy rating for tasks to be visible to a given user. (For all I know this is already in place; I'm not a Turk user.)

I would almost argue there is a digital food chain.For example by moving from MTurk towards Odesk towards VWorker towards Elance and so on. It really really all comes down to the individual skill level and with the wealth of information available online it comes all down towards the level of dedication one is willing to invest into acquiring a marketable skill.

The problem I see with these articles is, that they try to paint MTurk in a bad light and I would hate to see it disappear just because it becomes an image problem for Amazon.

Economic value is nothing more than what someone will pay. It is meaningless to say that if people will only pay $3 then there's no way anyone should pay $10.

On slave-owning plantations, the owners were paying $0/hr in wages, but that didn't mean that there was no way slaves can or should be paid more than $0/hr.

Once a work system is good enough to provide 5 minute highly qualified answers to questions like "is this CT indicative of cancer?", people would be paid $3/hr for those answers.

Scarcity and demand, not ability determines wages.

I agree that ability doesn't determine wages. But I would like to qualify your statement about what does determine wages. In a free market, wages are determined by scarcity and demand. However, a free market is a theoretical construction that often fails to model reality. For example, free markets do not include coercion, deception or collusion, all of which are ancient and universal parts of human life.

For example, the $0 pay rate of a field slave is not because nobody wants cotton-picking, or because there are too many people who will pick cotton for free.

Indeed. That's why equating pay to ability is pretty silly.

MobileWorks was one of the few companies that joined YC with an explicit social agenda tied up with our commercial goals. Our idea's always been that you can get much higher-quality results than Mechanical Turk by treating people better, paying more fairly, and focusing on the worker -- the opposite of Turk's anonymous high-volume approach.

Still, it's funny to see how surprised crowd workers are when they first find a crowd platform that takes their interests into account more than Turk, even through small steps like maintaining a friendly community of workers and a gentle onboarding process. One guy told us that "going from Mturk to MobileWorks is like going from a seedy motel to a 5-star hotel".

Pretty awesome that YC will fund businesses which both have a commercial goal and a social agenda. IMO that's more useful than pg et al donating a percentage of their carry to charity (although that's great too). Of course, the businesses have to make sense commercially as well, but MobileWorks doesn't seem to have a problem with that.

I expect they're used to this. Successful businesses in YC often have hugely positive social outcomes even if it's not part of their initial value proposition: Airbnb has helped folks earn enough to avoid losing their homes, Ridejoy reduces atmospheric pollution, etc.

For us, it was fortunate that the commercial opportunity was directly linked with our social mission -- doing the right thing by workers makes more business sense in crowdsourcing than treating people anonymously and paying them lousy wages.

Why is an article like that (quality long form journalism) published in a local alt-weekly paper vs. something national?

EBX does great work. But more broadly, local alt weeklies are much more willing to fund this sort of thing than the big papers, and that's been true for at least fifteen years.

No one wants to buy advertising next to an article like that, so it's a tougher sell. Meanwhile, local alt. weeklies are often run by people who think it's cool to publish articles like that.

I think advertiser's content preferences are oversold. They care about the size of the audience and their willingness to buy, and maybe how much trouble they'll get in if they're caught funding the KKK. Almost nothing else.

Local alt weeklies have two problems: they're local (small audience), and alternative (not buying your mass market crap).

Yeah, I was thinking GQ, Vanity Fair, or Rolling Stone -- still counterculture (sort of, at least for RS), but wider circulation.

Yet another indication skill development is the key to a decent life. Just like nobody will pay much for unskilled labor, nobody is going to pay much for unskilled button clicking.

People who depend on laws to extract more than their labor is worth are swimming against the current.

Here's one way Facebook has benefited from this type of model - http://gawker.com/5885714/ - "Inside Facebook’s Outsourced Anti-Porn and Gore Brigade, Where ‘Camel Toes’ are More Offensive Than ‘Crushed Heads’"

It's hardly a "sweatshop", since performing online tasks is not really physically demanding.

From wikipedia (http://en.wikipedia.org/wiki/Metaphor):

"A metaphor is a literary figure of speech that describes a subject by asserting that it is, on some point of comparison, the same as another otherwise unrelated object. Metaphor is a type of analogy and is closely related to other rhetorical figures of speech that achieve their effects via association, comparison or resemblance including allegory, hyperbole, and simile."

Reply on news.ycombinator.com