Thursday, September 29, 2016

Google takes pushes digital Assistant to the next level

Check out this article talking about Google's next move with Google Assistant. This article talks about how Google are seeking to stay relevant in a world with declining web searches. Thanks to Ana Maria Sanchez for sending in the article!

Amazon's battle for grocery dollars

John Martell shares this article about Amazon's fight in the grocery space. Despite an influx of funds for grocery delivery startups and lots of room for growth, traditional brick & mortar groceries have proven very difficult to displace.

Thursday, September 22, 2016

How Microsoft competes with Apple & Google in product connectivity

April Baker shares with us an interesting article about the inter-connectivity of products among three big tech firms, Apple, Google, and Microsoft. We found the visualization of product connections particularly interesting. Have things changed much in the past year since the article was posted?

Learning from Video Games: Micropayments and Unbundling

Thanks to David Hong for sharing an interesting article that discusses “micropayments”—not as a new idea, but an increasingly viable alternative to advertising and paywall revenue models. Specifically, how can digital content providers combat the squeeze in profits that’s resulted in Facebook and Google’s dominance in advertising? Freemium pricing models within the gaming sector offer a lot of promise.

Tuesday, September 20, 2016

Deep and Cheap Learning: Why does it work so well?


Henry Lin and Max Tegmark have a fascinating new paper arguing that the success of deep learning in so many domains has deep connections to the fundamental laws of the universe.  Both take potentially enormously large sets of possible data sets and simplify them to tiny set of outcomes, governed by just a few parameters. Luckily, they both simplify to much the same tiny subset.

As the authors put it:

We will see in below that neural networks perform a combinatorial swindle, replacing exponentiation by multiplication: if there are say n = 106 inputs taking v = 256 values each, this swindle cuts the number of parameters from v n to v×n times some constant factor. We will show that this success of this swindle depends fundamentally on physics: although neural networks only work well for an exponentially tiny fraction of all possible inputs, the laws of physics are such that the data sets we care about for machine learning (natural images, sounds, drawings, text, etc.) are also drawn from an exponentially tiny fraction of all imaginable data sets. Moreover, we will see that these two tiny subsets are remarkably similar, enabling deep learning to work well in practice.
 

Monday, September 19, 2016

Machine Learning's potential impact on future jobs

An interesting article on machine learning which gives a great perspective on how this ties into the success of Google, Amazon and the likes plus links to specific programs that companies are running to train their employees accordingly. 

Thanks to Christian Umbach for bringing this article to our attention!

Sunday, September 18, 2016

Estimating Demand Curves at Uber, and Elsewhere

Swati Rao points out that Steve Dubner interviews his co-Freakonomist, Steve Levitt about "Why Uber is an Economist's Dream".  Among other things, they talk about the new paper that Jonathan Hall, Steve Levitt and others wrote seeking to estimate consumer surplus from Uber, which starts with an estimate of the (short-run) demand curve for Uber.

It's a great discussion, though I'm a bit surprised he said "nobody ever had really actually estimated a demand curve" because its been done numerous times, sometimes with controlled experiments.

Thursday, September 15, 2016

An Analysis of the Labor Market for Uber’s Driver-Partners in the United States


Jonathan Hall, Uber's chief economist, has argued that Uber has limited ability to change the hourly income of its drivers because higher prices just attracted more drivers, who then spend more time waiting or driving idle. He says the equilibrium is when their Uber earnings roughly equal what they could earn elsewhere.

But what do we know about typical Uber driver earnings? Here is a paper by Jonathan Hall and Alan Krueger on the labor market for Uber drivers

"This paper provides the first comprehensive analysis of Uber’s driver-partners, based on both survey data and anonymized, aggregated administrative data. Uber has grown at an exponential rate over the last few years, and drivers who partner with Uber appear to be attracted to the platform in large part because of the flexibility it offers, the level of compensation, and the fact that earnings per hour do not vary much with hours worked, which facilitates part-time and variable hours. Uber’s driver-partners are more similar in terms of their age and education to the general workforce than to taxi drivers and chauffeurs. Uber may serve as a bridge for many seeking other employment opportunities, and it may attract well-qualified individuals because, with Uber’s star rating system, driver-partners’ reputations are explicitly shared with potential customers."

Wednesday, September 14, 2016

The "Great Debate" continues



Bob & Erik Open TED 2013
I talked to Scott Lanman and Daniel Moss about the future of growth and they published my thoughts in Bloomberg, interspersed with wisdom from my friend Bob Gordon under the title "The Great Debate: Can Technology Transform the Economy Again?"  As with our TED talks, I'm the optimist in this one. 

Asymmetric Information: The Market for Lemons




The Economist has an interesting article summarizing some of the key insights about the asymmetric information, including George Akerlof's classic about the Market for Lemons.  It's part of their six part series on seminal economics ideas.