Data Is Not the New Oil. Rethinking Data as a Reusable Resource, Not a Finite Commodity With Hal Stern


In this episode of Leading with Data, we discuss rethinking data, problem-solving, and why it is important to not treat data as “the new oil.”

Joining us is Hal Stern, VP and CIO of R&D at The Janssen Pharmaceutical Companies of Johnson & Johnson.

Hal spent over 20 years at Sun Microsystems and worked with them throughout the dot com boom and bust before moving to Juniper Networks and then Merck. Hal realized, however, that he was much happier designing solutions than designing products.

This led to a dramatic switch from trying to solve networking problems, to solving societal problems around the mechanism of disease at Janssen. Hal joined Jason Dorsey in an episode packed with insights around data analysis and warnings about treating personal data as a commodity.

Hit play to listen to the podcast episode or read on to find out more.

Hit play on the player above or read on to find out more.

Download arrowDownload the Mp3


Solving Problems With Data

Hal has spent over 20 years working on solving problems around networking. While he loved dealing with these “exceptionally large” problems, he was frustrated that optimizing networks or optimizing bandwidth were problems that couldn’t be solved perfectly.

When he got the call from Johnson & Johnson with the opportunity to work on some really hard problems, and in life sciences, Hal jumped at the chance.

This was also an incredible opportunity for Hal to work on complex problems that had a real societal impact.


Embracing the Outlier Perspective

When dealing with big problems, being able to think differently is key.

Hal has always seen himself as an outsider. He describes life growing up as the grandson of Ukrainian immigrants in New Jersey with the added complication of having a hatred of socks!

Yet, Hal’s “weirdness” is something he believes has helped shape his career. He is proud of being different, and it’s not just his hatred of socks. It’s his unique perspective. He believes that the way he approaches problems is due to his different way of seeing the world.

“The guy who hired me at Merck used to call us the ‘Island of Misfit Toys,’ because he collected people who didn’t have a better place to be.”

Without these misfits and weirdos, however, we wouldn’t have people like Bert Sutherland. Hal worked with Burt at Sun Microsystems and without his unconventional thoughts, we wouldn’t have the computer mouse or graphical computer displays.

Weird is not only good, but is essential for driving innovation. Hal argues that having a mix of different skill sets is fundamental to creating a strong team. (Hal recommends David Epstein’s book, Range for more insight on this).


How Can We Make Better Decisions From Data?

Hal explains how it’s easy to train machines to differentiate between a spider and a starfish. But the real advancements, the big decisions that can come from data, need a lot more work and a lot more training.

“I can take any image recognition systems and find adversarial inputs for them, whether its fried chicken and Labradoodles or chihuahuas and blueberry muffins. You can fool a lot of systems because they haven’t been sufficiently trained with a broad enough input.”

Hal sees disease as an adversarial input to the immune and endocrine systems of the body. To really understand the mechanism of disease, we need to evolve our data analysis capabilities and grow our learning systems.

Hal says there’s a temptation to say – that’s it – we’ve sequenced the genome, get machines to work out the answers. Unfortunately, disease evolves, and our machine learning systems need to evolve too.

“We don’t understand enough, and the combinations are so computationally complex and so incredibly large in terms of the input space, that we have to be careful not to jump at the bad science.”

Hal warns that we must try to resist looking for correlation without strong scientific causation.

Does he think machine learning will replace scientists? Never. We will only ever be able to amplify what scientists do, says Hal, but the technology will get us there faster and hopefully more easily.


Is Data the New Oil of the 21st Century?

Absolutely not. According to Hal, we all need to stop thinking of data as a commodity. Data is not a commodity like bacon or orange juice or oil where you use it once and it’s gone. Data can be recycled.

“Data – we use it, we reuse it, and we remix it to find new value in it. It’s music.”

Just as the early days of hip hop were about sampling, remixing, and drawing on musical legacies, data analysis also needs to bring the old and new together, cross genres, to drive discoveries.


Should Personal Data Be Treated as a Commodity?

We need to stop thinking about data as something that is owned by one person, says Hal. The real value of data is in how you can combine it with other data.

Hal explains how standalone data, like your location or health, is not that interesting. However, if you can put data together and remix it – things become very interesting.

While people are relatively happy with giving away their data for specific things, once data is mixed, there are ethical considerations that need to be made.

Hal predicts that we will probably move towards managing personal data like the creative commons approach to copyright in the publishing world. Perhaps our personal data can have attribution rights? Or we could put limits on the number of times it is shared?

Whatever the future holds, Hal believes data has a lot to learn from the music and publishing world.

To find out more about Hal’s insights into data analysis and leadership, listen to the full interview on Leading with Data.