Tech Economics Newsletter
What's new from economist Andrew Chamberlain, Ph.D.
I hope your summer is going well. I'm here with a quick update from the world of tech, economics and data science.
Fun With the News: I've been busy with media this summer. I'll share two favorites. Here's a profile in The Atlantic of how technology is changing my role at Glassdoor. And here's me on Marketplace explaining why more Americans should quit their jobs.
Stream This: There's a terrific new NBER working paper exploring data from Swedish music streaming giant Spotify. It shows songs being added to weekly recommended playlists has a huge impact on artist revenue. It's the first time I've seen Spotify data used in a paper -- hopefully a sign they're opening up to researchers.
Gender Pay Gap in Gig Economy: Uber has been a pioneer in sharing data with researchers. Here's a terrific new paper revealing a large gender pay gap among Uber's rideshare drivers. The paper shows the gap isn't due to discrimination -- it's mostly due to women bearing the burden of family responsibilities. Nevertheless, this paper is at risk of being misinterpreted by some readers -- a real-world risk every company that shares data with researchers eventually faces.
Do We Really Need Deep Learning for That? A team of researchers at Google have a new paper on the power of deep neural networks to make accurate health predictions from electronic health records. That's great. But there's a fun twist in their appendix. They show a simple regularized logistic regression performs about as well as deep learning -- a gentle reminder about the practical value of simple methods.
Is Deep Learning Just Polynomial Regression: Here's a provocative new paper that argues deep neural networks are basically just the wolf of old-fashioned polynomial regression in sheep's clothing. Needless to say, the paper has generated controversy. Regardless of the paper's merits, it has sparked a useful debate about what's happening inside the "black box" of deep learning algorithms -- a healthy conversation for data science practitioners.
Use Maps Like A Boss: Two years ago Uber developed it's own mapping software to keep tabs on millions of riders and drivers. It's called keppler.gl and now it's open source and available for free. Here's an intro from their engineering blog, and here's the GitHub repo. When economists say traditional surveys overstate inflation in today's tech economy, "free" services like this are what we're thinking about.
A New Economist Joins My Family: Finally some personal news. My wife and I are expecting our first baby next week -- it's a boy -- so this newsletter may be on hiatus until the fall. Wish me luck!
Thanks again for staying in touch -- I wish you a safe, happy and healthy summer.
Andrew Chamberlain, Ph.D.