Hacker News Reader

Comment by lkrubner | Hacker News Reader

This is a good comment. It is interesting to consider how much rhetoric there is about robots and AI and Machine Learning and how these technologies can revolutionize work and lead to big productivity gains. And yet, in reality, the economy shows no sign of this. Productivity growth was very strong in the mid-20th century, but since 1973 productivity growth has been stagnant.

View on HN

I don't know much about the subject. How do we measure productivity? What units do we use?

I ask because it _feels_ like everything has changed since 1973 at least. So I'm curious where my intuition differs from reality.

> How do we measure productivity?

Productivity as typically reported is a statistic called "Total Factor Productivity" (TFP).

Here's how it's calculated: https://en.wikipedia.org/wiki/Total_factor_productivity#Calc...

> What units do we use?

TFP is unitless.

AI and Machine Learning are newer, and probably have plenty of productivity gains to point to, but the thing that comes to mind for me is the amount of CAD and computer analysis that goes into our products these days. You can make a car that lasts twice as long and is more aerodynamic. Musk's rockets use computers for design and algorithms to land. The down stream effect is new satellites with more bandwidth (e.g. Iridium).

In the end, a designer can make a product people get more use out of.

Productivity has been stagnant, or productivity growth has been stagnant?

Constant productivity growth is a terrible argument that computers haven't contributed to productivity. No single innovation, however broadly defined, will produce constant growth. As one source of growth peters out, another one picks up.

Call centers, accounting, and help desk and other things that used to take a small army are now a subscription service for far less money. The “efficiency” rating is probably wrong if it doesn’t think that’s increased efficiency.

> Productivity growth was very strong in the mid-20th century, but since 1973 productivity growth has been stagnant.

That's not true; the first decade of the 2000s rivaled the post-war era in terms of TFP growth:

https://www.bls.gov/lpc/nfbbar.gif

Even Brynjolfsson's ultimate thesis was not that "computers don't induce growth in TFP", but rather that "computers and business practices need to converge in order for computers to have an effect on TFP".

Reminds me of this article:

Google and Facebook contribute zero economic value. That's a big problem for trade.

https://www.washingtonpost.com/news/innovations/wp/2016/10/2...

Really, 0 economic value. Nobody gets anything of value from using those services, its just wasted time.

The linked article is about how despite those services obviously having value to their users, they aren't accounted for in economic statistics, e.g. in measuring productivity, because they are "free" (i.e. paid for by subjecting oneself to advertising/tracking).

I am suspect of this being true because GDP is measured in several ways, and you can measure the contribution of tech companies through their salaries and profits or through their expenditures and sales to ads. (i.e. their contribution to GDPcan be measured in their revenue)

But this is also a red-herring, its at most an accounting issue, not a real economy issue.

Changes in GDP are caused either by debt or by productivity increases. Productivity growth since 1973 has _not_ been stagnant.

Labor productivity since 1973 has been much lower than in the decades before 1973:

https://qz.com/633080/the-rise-and-fall-of-american-producti...

Because all the competitors are doing it, too.

This is nonsense. The inflation-adjusted cost of producing one can of beer would be a pretty standard measure of productivity (in the beer industry). It can rise and fall completely independently of whether beer companies all use the same strategies or all use unique strategies.

Reply on news.ycombinator.com