Monday, 10 Dec 2018

An innovative way to gauge the invention

Sewing machines dated but in good working order lie under cards from Italy at Philip's Shoe Repair in Washington, DC in 2017. Sewing machines were among the most influential inventions of the nineteenth century. The left machine is more than 50 years old and the right machine is more than 100 years old. (Jahi Chikwendiu / The Washington Post)

The US Patent Office has stored the text in more than 10 million patents. But that's often all they have: a huge amount of text. Many old patents do not contain any form of citation or industrial specification that researchers could use to understand the history of American invention.

Now a team of economists has created an intelligent algorithm that treats this text – often the only consistent data we have for many of the country's most famous inventions – to create a measure of the influential inventors and industries of the past 180 years years.

To measure the influence of a patent, they first look for patents whose text, once cleaned and cleaned of common words and aberrant words, has little in common with those that preceded it. As the electromagnetic motor patented in 1888 by Nikola Tesla, such inventions are revolutions, not iterations.

But they do not blink in the pan either. Like the nylon fiber patented by the DuPont chemist, Wallace Carothers, in 1938, they shape the ideas and vocabulary of the inventors who follow them. Thus, in addition to searching for inventions with little precedent in history, the algorithm also looks for inventions that have been followed by a wave of patents using similar terminology. This suggests that, like Tesla's motor or Carothers's nylon, they have had a lasting impact on American innovation.

Innovation defies any measure. It's intangible. But economists continue to try, because innovation helps determine productivity, and productivity helps determine economic growth.

Previous attempts to compare innovations between epochs have included patents or patent citations. Such accounts are easily skewed by corporate practices or the quirks of a patent clerk. Patents rarely cited other patents until 1946. A first telegraph patent had hardly been cited, but it is not unusual for modern patents to accumulate hundreds or even thousands of citations.

The new analysis was published as National Bureau of Economic Research working paper by Bryan Kelly (Yale School of Management), Dimitris Papanikolaou (Kellogg School of Business, Northwestern University), Amit Seru (Stanford College of Commerce) and Matt Taddy. Taddy now works at Amazon, an online retailer whose founder and general manager, Jeffrey P. Bezos, also owns the Washington Post.

The team's database began in 1836. The previous records were destroyed in December of the same year when the hotel housed the US Patent Office, one of the only buildings spared by the British during the war. War of 1812, went up in smoke. Victims include fans such as Eli Whitney (1794) gin, Cyrus McCormick (1834) and Samuel Colt (1836) revolver, which could not be included in the list. study.

The algorithm requires five years of data both before and after the grant of a patent. As a result, researchers evaluated only patents issued between 1840 and 2010.

Their text similarity measure compares favorably with current methods. It correlates with citations-based rankings when they are available and tends to give higher ratings to important patents on the Internet. lists such as those established by historians of patent offices. Unlike quotes, the algorithm can compare patents from one era to the next. Unlike historians, the algorithm allows researchers to revise the text of millions of patents at a time.

A simple patent count per person seems flat for most of US history, but such a measure is skewed by countless incremental advances. Sharp peaks and valleys emerge once researchers have considered only the most influential patents.

The number of important inventions grew in the 1920s and peaked in 1932, near the nadir of the Great Depression. This seems counterintuitive. But the economics historian Alexander Field of Santa Clara University has pleaded for his existence in a 2003 article of the American Economic Review. At the time, he said, it was a "radical hypothesis".

"Most people would not think of a decade in which the unemployment rate would be mostly double-digit as a likely candidate for many technological breakthroughs," Field said.

Field argues that the innovations of this era crossing a broad border of the American economy laid the foundation for the rapid industrial and economic expansion of the country during and after the Second World War.

Their algorithm was designed to detect industry-level trends, but we can also use it to select influential individual patents among all those issued since 1836. There will be noise. The same goes for qualitative and quote-based measures, said Papanikolaou.

Modern inventions often require dozens or even hundreds of patents and it is not always possible to determine which of these patents best represents a drug, a smartphone or a jumbo jet. To eliminate the noise of a list that might place pharmaceutical formulations and cryptography methods next to the skateboard components, we have combined the researchers' algorithm with a list of about 250 patents. Historical significance compiled by their team.

The items listed below were in the top 0.5% of all patents rated by the team and were rated as outstanding by human experts.

The researchers found a close relationship between a company's most influential patent portfolio and its market value. The most influential company in their data set was International Business Machines. IBM has issued about three times more revolutionary patents than its closest competitors, AT & T and Motorola.

Petra Moser, an economist at the Stern School of Business at New York University, has been studying patents and innovation for a long time, analysis of two world fairs of the nineteenth century. Patents do not capture all the human advances, she said, but "they are our most complete list of innovations."

Mr. Moser said that the measure created by Papanikolaou and his colleagues would be "extremely useful" in his work and that it reinforces the understanding of when industries such as chemical manufacturing and crop production are to their most innovative traditional patent-based measures. Innovation, she said, allows humanity to improve its quality of life without consuming more scarce resources.

"We need to understand what drives innovation to be able to feed people and not destroy our planet," said Moser. "Innovation is the engine of growth."

In the present era, gene splicing patents have laid the foundation for companies such as Genentech. Herbert Boyer, pioneer of canonized methods at the top of the list of most influential patents alongside academic Stanley Cohen, founded the biotechnology company in 1976.

Boyer and Cohen's work on recombinant DNA is typical of the modern wave of American innovation, which relies heavily on software and bioscience. This wave seems to have crested. Fewer influential patents have been issued almost every year since 1998.

There may be a good side for those who are worried about the country. long period of low productivity. Economists have long quantified the productivity of workers and machines, but they have struggled to measure the gains from new inventions. By measuring innovation, researchers hope to better understand one of the most difficult components of economic growth. In their research, Papanikolaou and his colleagues discovered that it took years for America's inventiveness to be reflected in the country's economic output.

"The years when there has been a lot of innovation tend to be followed by periods of strong productivity growth," Papanikolaou said.

Even on its downward slope, the third wave of US innovation could continue to bear fruit.


Post Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: