We Must Prevent Data Pseudoscience Before It’s Too Late

We Must Prevent Data Pseudoscience Before It’s Too LateTimothy BowmanBlockedUnblockFollowFollowingMar 19Until recently, it was possible to choose to live outside of the reach and influence of “the Big 5” technology companies.

You could use a old flip phone to talk to friends and family, pay with cash at a locally-owned mom-and-pop store, read and write with pen and paper, navigate using a giant map (then lose your mind when attempting to refold it), and use the USPS to communicate with friends and family.

But, as tech bloggers are increasingly discovering, those days are rapidly coming to a close.

Retailers are moving away from cash, the lifeblood of commerce.

Not only do 95% of recruiters use LinkedIn to source candidates but, of January 2019, there were nearly 27 million more US workers with LinkedIn profiles than are active in the civilian labor force.

Even the very creators of the digital tools themselves are struggling to find a way to unplug.

While the above examples are what make the headlines, what lives just below the surface is the bigger risk: the unseen and often unknowable algorithms that leverage the data and metadata generated from consumer-facing tools and services to make decisions that materially impact the day-to-day lives of humans (health, financial, news, social feeds, etc).

These trends beg three very important questions:How do we ensure historical inequities and cognitive biases aren’t “baked in” to these algorithms, whether deliberately or unconsciously, in ways that cause material harm?How do we ensure that junk science, non-reproducible studies, and non-statistically significant polls aren’t weaponized by those with ulterior motives?How do we ensure the US, or any country, remains competitive and innovative in the digital economy while also protecting and respecting privacy?A Minimum Viable Path ForwardThis is not the first time we’ve found ourselves at the intersection of ethics, technology, and commerce.

When such conflicts have occurred, two types of solutions have been used to create transparent, trust-based markets: top-down regulations and bottom-up movements.

A good example of a successful top-down approach is the automobile.

As that new technology became sufficiently advanced as to be lethal when used irresponsibly, we collectively decided (over the objections of industry) to define the rules of the road: we set speed limits, defined standard signs, established licensing requirements, and required certain safety features like seat belts and air bags.

The process is still evolving as new consumer safety technologies, such as back-up cameras, are proving more effective.

The top-down route, however, only makes sense when three conditions are met:The need is obvious to everyone.

The solution is easily communicated and understood by non-experts.

The rate of change is slow enough for legislators to gain a working knowledge of the technology itself before making the rules.

When such conditions aren’t met, the experts and practitioners themselves must come together to mutually and publicly agree upon what defines the “minimum viable ethics” necessary to balance the safety of consumers with the profit motive of enterprises and creativity of entrepreneurs.

The need for for a bottom-up solution has been recognized since Antiquity.

Doctors have the Hippocratic Oath.

Engineers have the Archimedean Oath.

It’s time for Data Scientist to have an oath of their own.

If done right, the oath will increase trust in institutions, technology, and algorithms, eliminating friction to adopting and using digital ecosystems so that all people can benefit from the digital revolution.

Trust is the new currency when it comes to digital technology.

— Satya Nadella, CEO of Microsoft via SmartCompanyAfter all, if we can’t see how the sausage is made, even the most diligent of us can fall victim to manipulation and misinformation.

A Beacon of Hope: The Hypatic OathHypatia at the Haymarket Theatre, H.

M.

Paget via Wikimedia CommonsHypatia was a philosopher, mathematician, and astronomer who lived in Alexandria, Egypt around 400 CE.

While much of her legend is most likely apocryphal, it is believed that Hypatia’s death resulted in the burning of the Library of Alexandria and end of classical antiquity.

Beyond her contributions to math and science, Hypatia evangelized two key tenets that make her uniquely qualified to assume the mantle of an oath for data science:It is the responsibility of those with power and knowledge to lead with high moral standards and act for the benefit of their fellow humans.

The root of all “evil” is an absence of light.

To jump start this bottom-up movement, there is a first draft of the Hypatic Oath of Data Science on Change.

org (reproduced below).

By signing this covenant, I do hereby commit to upholding the long tradition of Hypatia, Copernicus, Sir Isaac Newton, Emilie du Chatelet, Jonas Salk, Madame Curie, Sir Francis Bacon, Mary Somerville, Dorothy Hodgkin, Nate Silver, and the countless other scientists and mathematicians, known and unknown, who refused to let the powerful manipulate and exploit those who weren’t.

I, of sound mind, believing in the principles of science, truth, and justice, do hereby swear to only use my abilities for the benefit of humankind and do no harm with data.

I will conscientiously object to using my mathematical and technical abilities or fruits of them to willfully and intentionally:1.

Misrepresent or aid in the misrepresentation of the “truth” by participating in the creation, analysis, or dissemination of non-reproducible and non-statistically significant studies and polls.

2.

Enable discrimination of or cause disparate impact on protected groups by treating predefined data types or data sources differently than others.

3.

Violate and/or ignore the fundamental human right to privacy by using, selling, or storing data or metadata acquired by unethical or illegal means.

Turning on the Light, TogetherImage by Skitterphoto via PexelsIf you’re a practitioner of statistics, mathematics, or data science or work for an organization that employs statisticians, mathematicians, or data scientists in either the for-profit or non-profit sector, sign the oath and publicly encourage others to do the same.

While it the road ahead is long and the forces we’re going up against are incredibly powerful, doing nothing is no longer an option.

.

. More details

Leave a Reply