When the economist Simon Kuznets introduced his new measure of gross domestic product to the U.S. Senate in 1934, the project of national income accounting fit the times: Americans lived in an economy of tangible production and goods you can count. Today, measuring "output" is a lot more difficult.

Kuznets's successors in the Bureau of Economic Analysis have recently been rewriting economic history, restating the size and composition of GDP all the way back to 1929. They're reclassifying two forms of intellectual property -- business spending on research and development as well as some forms of art -- as investments rather than expenses.

Investment is counted as part of GDP; expenses aren't, because they're seen as intermediate steps in production rather than final output. The industry most affected is pharmaceuticals, which puts 27 percent of its business investment into R&D.

The changes show how hard it is to track today's economy with a measure designed for an older one. (Official second-quarter figures will be released today.) The government likes to think of investment as a physical product: a forklift, an office building. It was slow to count business software as investment, only switching in 1999. That intellectual property -- ideas -- now counts as GDP recognizes the character of today's economy.

GDP has always had its problems. For a start, it's a narrower measure than is often advertised: It only counts production. Health, environmental quality, strength of community -- many of the things that affect the standard of living are left out. That's prompted a variety of wider alternative measures, as Jon Gertner wrote in the New York Times. None of them seem ready to displace the 80-year-old measure.

The government's tweaking addresses a different problem -- not that production is too narrow a concept, but that it has become more difficult to define since 1934. That's largely because growth in developed economies is changing from producing more to producing better -- that is, shifting to higher-quality versions of goods and services we already make. The trouble is that "better" is much harder to measure than "more."

Economists are trying, though, with a technique known as hedonic pricing. This uses a statistical model to identify how consumers value a product's different traits. Suppose consumers are willing to spend an extra $100 for a computer with a faster processor. As slow computers leave the market, a measure of GDP that didn't use hedonic pricing would count that difference as inflation, not economic growth -- whereas the consumer sees a better computer, not just a higher price.

About a fifth of GDP gets this adjustment. The rationale is clear enough, but the statistical wizardry adds a whiff of subjectivity. It has spawned all kinds of conspiracy theories about the federal government's efforts to hide inflation.

Improving quality isn't the only area where GDP is showing its age. The shift to a services-based economy is another, as is the shift to online commerce. But GDP's biggest oversight is investment in human capital.

Here's an example. If a business spends $10,000 on new computers and software, that $10,000 is counted in GDP as an investment. If the business spent the $10,000 on worker training instead, it would be counted as a business expense and is ignored in GDP. This makes no sense: Training is as much an investment as a new computer.

Economists Paul Samuelson and William Nordhaus once called GDP one of "the great inventions of the 20th century." They were right, but it will need some upgrades to stay relevant to the 21st.

(Evan Soltas is a contributor to the Ticker. Follow him on Twitter.)