When GDP became the dominant measure of economies in the 1940s, the internet was still a half-century out. Today, the internet is responsible for a major chunk of economic activity, but GDP misses much of it. This has widened the gap between the closely watched metric and actual economic health.

• Kaveh writes: A global effort, reaching back decades, is picking up steam to either change or supplement GDP as the go-to gauge for understanding a country's — and the world's — level of prosperity.
• The efforts seek to account for relatively new industries, plus intangibles like income inequality and clean air and water.
• But the dizzying pace of technological advances may be enlarging the gap even as they work to close it.

Why it matters: The problem with GDP is not merely academic — for instance, reliance on the economic benchmark deserves part of the blame for the 2008 financial crisis, according to a 2009 report from a commission led by Nobel laureate Joseph Stiglitz. If economists had been watching other metrics, too, like indebtedness, they would have had an earlier warning of trouble ahead, the commission wrote.

Among the proposed alternatives:

• In the early 1990s, a "green GDP," developed at the U.S. Bureau of Economic Analysis, suggested that the mining sector's positive impact on the economy was being overstated. A congressman from the coal mining state of West Virginia led the successful drive to squash this attempt, its former director told NPR in 2004.
• In 2016, a measure of contentment, akin to Bhutan's "gross national happiness" score, was boosted by USC economist Richard Easterlin.
• In 2018, an "extended GDP," suggested by University of Maryland's Charles Hulten and Leonard Nakamura of the Philadelphia Fed, would have captured more spending on the internet.

The latest proposal comes from a group of economists led by MIT's Erik Brynjolfsson. In a forthcoming working paper, they propose a measure called GDP-B, adding in the benefits of free digital goods and new technology.

They estimate that hidden benefits from Facebook alone have added 0.05–0.11 percentage points to GDP every year since its 2004 launch. That would have totaled 1.54 percentage points from 2003 to 2017 — a result "too large to be ignored," according to Hulten, who was not involved in this research.

• How it works: The economists ran experiments in which they offered people varying sums of money to stay off of Facebook for a month, then extrapolated its total value based on the average amount people accepted. This is the willingness to pay test, a method for determining the hidden value of something that’s free.
• Right now, when an old-school product gives way to a digital substitute, only half of the trade-off shows up in GDP, says Avinash Collis, an MIT researcher who co-authored the GDP-B paper. The decline of landlines is captured, for example, but the rise of its replacement — social media and free messaging apps — is not.

What's next: GDP has solidly hung on despite its critics, but Hulten, who advises the Bureau of Economic Analysis, says the body is now again considering alternatives. "Agencies are very aware of the need to change," he says, but are burdened with a busy schedule — plus many decades of inertia.

A new supercomputer to be built at Argonne National Laboratory in Illinois will be the most powerful in the country when it begins operating in 2021 — more than five times faster the current leader, the Summit supercomputer at Oak Ridge National Laboratory in Tennessee.

Axios Science Editor Andrew Freedman reports: The Energy Department, Intel and subcontractor Cray Inc. announced Monday an agreement worth "more than $500 million" to provide Argonne with the country's first "exascale" computer system.

Why it matters: The transition to exascale computing involves a thousandfold increase in computing power from the petascale systems installed during the past decade, and it promises to open up a broader array of applications, such as precision medicine and AI

Details: The Aurora computer will have the performance of one "exaFLOP," which is equal to a quintillion floating point computations per second, according to a press release and briefing from Intel. The potential uses for this computer include:

• Complex cosmological simulations to better understand the universe.
• Precision medicine, such as testing new approaches for drug response prediction to treat cancer and other diseases.
• Climate and extreme weather prediction.
• Mapping the human brain down to the neural level.

Context: There's a race heating up between the U.S. and China for who has the most powerful supercomputer.

• While it will be the most powerful system in the U.S. when it goes online in 2021, an Argonne National Lab spokesperson said it's not clear whether it will be the fastest computer in the world at that time.
• Raj Hazra, vice president of Intel's enterprise and government group, told Axios that leading in computing power isn't nearly as important as what that nation does with its capabilities.

"From the perspective of winning the race, it’s not just getting to exascale, but what does exascale get you to that is important. The race matters in terms of stoking innovation. To compete you have to be able to compute."

Go deeper:

• How the U.S. plans to stay ahead in the quantum race
• Creating the first quantum internet
• The West’s China blind spot

From 2011 to 2018, the number of millionaires in the world jumped to 42.2 million, from 29.7 million, a 42% leap. But the amount of money spent on art barely increased.

Erica writes: Whether due to geopolitical instability or because the new crop of millionaires just doesn't care about art, sales are lagging, reports NYT. The dollars spent on art was $64.6 billion in 2011. Last year, it was $67.4 billion, a mere 4% increase.

The United States remains the largest market in the world for art, with 44% of the sales. Behind the U.S. is Britain with 21%, then China with 19%.

How Sears lost the American shopper (Suzanne Kapner — WSJ)

Self-driving cars and the fear of the unknown (Joann Muller — Axios)

WeWork is watching you (Ellen Huet — Bloomberg)

The math that tells cells what they are (Jordana Cepelewicz — Quanta)

How Big Tobacco hooked kids on sugary drinks (Andrew Jacobs — NYT)

As we've reported, the world is drinking less booze, and to stay relevant, big distilleries and brewers are pouring money into cannabis drinks, which appear to be the next trend.

Erica writes: But there's one big problem with weed beverages, reports WSJ. They just don't taste that good.

• The taste is "grassy" and "funky," Ron Silver, a restaurant owner in New York, told the Journal.
• The problem: The oily cannabis extract that is used to make edibles doesn't mix well with water, and the drinks have to be shaken before every sip.
• One solution: Silver told the Journal he's been experimenting with cannabis-infused sweeteners.