07.16.14

The FDA’s Open Digital Sweatshop Initiative

Posted in Culture of Lickspittle at 2:40 pm by George Smith

Marvel at the promotional video uploaded by the Empire of Bezos to showcase “Amazon Web Services.” It’s awesome in that it has the FDA’s Chief Health Informatics Officer, Taha Kass-Hout, going on for four minutes about the miracle of “turning manual submissions from the public into machine-readable information with 99.7% accuracy” without once mentioning Mechanical Turk or that the work is performed by digital sweat-shopping.

Instead Kass-Hout relates how the FDA had a “19th century problem” of backlogged paperwork, adverse drug reactions submitted by the public, health professionals and doctors, paper that needed transcription into machine-readable form. There’s a stream of jargon and techno-gobble about “the cloud” and database language coupled with pretty digital icons of documentation whizzing by with not a single admission of the reality that the job is not implemented by machines at all, but by the equivalent of human hamsters running the wheels at Mechanical Turk for pennies.

It’s truly Orwellian, releasing a stink of vague obfuscation so that people who don’t know a thing about what’s going on in the background are led to believe it’s just another marvelous technical wonder on the road to the glorious future.

Taha Kass-Hout ought to be ashamed of himself although that’s a bit much to hope for.

The Obama administration has put on a public populist face, one that chides the Republican Party and corporate America for allowing inequality to balloon and the compensation of workers to flat-line. And here is the man from the FDA, talking about a technical work-around that simply relies on paying people virtually nothing for record transcription work.

The blog mentioned this previously, in connection with Mechanical Turk, and here is a recap:

At a time of great unemployment, poverty wages and increasing inequality, the Food and Drug Association has committed to employing Mechanical Turk digital sweat-shop labor through a private sector sub-contractor. Pure and simple, it is the use of taxpayer money in the nullification of people for the siphoning of the money to corporate America. Think of it as anti-stimulus. (What percentage of Mechanical Turk workers are in the food stamp program? Rhetorical. There are no statistics as the service and the businesses that use it are non-transparent.)

Plus human beings working for twenty or thirty cents a job are more reliable and so much cheaper than crappy optical character recognition software …

At a time when the economy is not producing jobs or a living for many Americans, the government response should not be to fill a labor need by leveraging desperation digital sweat shop labor.

This is wrong. The US government, specifically the Food and Drug Association, should hire Americans and get the job done, not resort to machine-like digital chiseling through a third party because it is allegedly swamped by a work load. Alternatively, it can use tax dollars to buy more automation and keep the work within the agency.

There are many paper shuffling and data entry jobs in the US government, all performed by civil servants. And a lot of that work, without labor protections, could simply be turned over to digital crowd-sourcing in network sweat-shops.

The government must still pay workers according to some set of civilized standards. And in no cases can the federal government refuse to pay civil servants if it doesnít like the cut or result of their work on any given day. Yet that is the model put in place when a federal agency transfers data transcription to labor on Amazonís Mechanical Turk. On Mechanical Turk, below subsistence pay workers can be denied mere dimes if their work is deemed sub-standard. And there are no appeals.

This isnít where the government should be leading. And whoever came up with the idea at the FDA needs some bad publicity and brush-back.

And I’m not sorry to say I will supply bad publicity when able.

Captricity, which is the firm that does this for the FDA, is a small firm in Berkeley.

Reads its website:

Extract structured data from paper forms. Even handwriting! Fast. Secure. 99%+ accurate

Ninety-nine percent accurate. For pennies on the job and the ability to deny payment through Mechanical Turk if the sub-standard is turned in, I should hope so.

Here is the Captricity blog.

You may especially enjoy, if only in a perverse way, the entry entitled “Evidence-based Research to Combat Global Poverty.”

And do tell us, how do you combat global poverty by using workers paid at astonishingly sub-poverty levels to transcribe administrative information and research on populations in poverty in endemically poverty-stricken countries?

Boy, that’s a brain-twister. But I’m sure they have an answer in the Silicon Valley.

Comments are closed.