Book Club: Code Economy, Ch. 10: In which I am Confused

Welcome back to EvX’s Book Club. Today we start the third (and final) part of Auerswald’s The Code Economy: The Human Advantage.

Chapter 10: Complementarity discuses bifurcation, a concept Auerswald mentions frequently throughout the book. He has a graph of the process of bifurcation, whereby the development of new code (ie, technology), leads to the creation of a new “platform” on the one hand, and new human work on the other. With each bifurcation, we move away from the corner of the graph marked “simplicity” and “autonomy,” and toward the corner marked “complexity” and “interdependence.” It looks remarkably like a graph I made about energy inputs vs outputs at different complexity levels, based on a memory of a graph I saw in a textbook some years ago.

There are some crucial differences between our two graphs, but I think they nonetheless related–and possibly trying to express the same thing.

Auerswald argues that as code becomes platform, it doesn’t steal jobs, but becomes the new base upon which people work. The Industrial Revolution eliminated the majority of farm laborers via automation, but simultaneously provided new jobs for them, in factories. Today, the internet is the “platform” where jobs are being created, not in building the internet, but via businesses like Uber that couldn’t exist without the internet.

Auerswald’s graph (not mine) is one of the few places in the book where he comes close to examining the problem of intelligence. It is difficult to see what unintelligent people are going to do in a world that is rapidly becoming more complicated.

On the other hand people who didn’t have access to all sorts of resources now do, due to internet-based platforms–people in the third world, for example, who never bought land-line telephones because their country couldn’t afford to build the infrastructure to support them, are snapping up mobile and smartphones at an extraordinary rate:

And overwhelming majorities in almost every nation surveyed report owning some form of mobile device, even if they are not considered “smartphones.”

And just like Auerswald’s learning curves from the last chapter, technological spread is speeding up. It took the landline telephone 64 years to go from 0% to 40% of the US market. Mobile phones took only 20 years to accomplish the same feat, and smartphones did it in about 10. (source.)

There are now more mobile phones in the developing world than the first world, and people aren’t just buying just buying these phones to chat. People who can’t afford to open bank accounts now use their smarphones as “mobile wallets”:

According to the GSMA, an industry group for the mobile communications business, there are now 79 mobile money systems globally, mostly in Africa and Asia. Two-thirds of them have been launched since 2009.

To date, the most successful example is M-Pesa, which Vodafone launched in Kenya in 2007. A little over three years later, the service has 13.5 million users, who are expected to send 20 percent of the country’s GDP through the system this year. “We proved at Vodafone that if you get the proposition right, the scale-up is massive,” says Nick Hughes, M-Pesa’s inventor.

But let’s get back to Auerswald. Chapter 10 contains a very interesting description of the development of the development of the Swiss Watch industry. Of course, today, most people don’t go out of their way to buy watches, since their smartphones have clocks built into them. Have smartphones put the Swiss out of business? Not quite, says Auerswald:

Switzerland… today produces fewer than 5 percent of the timepieces manufactured for export globally. In 2014, Switzerland exported 29 million watches, as compaed to China’ 669 million… But what of value? … Swiss watch exports were worth 24.3 billion in 2014, nearly five times as much as all Chinese watches combined.

Aside from the previously mentioned bifurcation of human and machine labor, Auerswald suggests that automation bifurcates products into cheap and expensive ones. He claims that movies, visual art services (ie, copying and digitization of art vs. fine art,) and music have also undergone bifurcation, not extinction, due to new technology.

In each instance, disruptive advances in code followed a consistent and predictable pattern: the creation of a new high-volume, low-price option creates a new market for the low-volume, high-price option. Every time this happens, the new value created through improved code forces a bifurcation of markets, and of work.

Detroit

He then discusses a watch-making startup located in Detroit, which I feel completely and totally misses the point of whatever economic lessons we can draw from Detroit.

Detroit is, at least currently, a lesson in how people fail to deal with increasing complexity, much less bifurcation.

Even that word–bifurcation–contains a problem: what happens to the middle? A huge mass of people at the bottom, making and consuming cheap products, and a small class at the top, making and consuming expensive products–well I will honor the demonstrated preferences of everyone involved for stuff, of whatever price, but what about the middle?

Is this how the middle class dies?

But if the poor become rich enough… does it matter?

Because work is fundamentally algorithmic, it is capable of almost limitless diversification though both combinatorial and incremental change. The algorithms of work become, fairly literally, the DNA of the economy. …

As Geoff Moore puts it, “Digital innovation is reengineering our manufacturing-based product-centric economy to improve quality, reduce cost, expand markets, … It is doing so, however, largely at the expense of traditional middle class jobs. This class of work is bifurcating into elite professions that are highly compensated but outside the skillset of the target population and commoditizing workloads for which the wages fall well below the target level.”

It is easy to take the long view and say, “Hey, the agricultural revolution didn’t result in massive unemployment among hunter-gatherers; the bronze and iron ages didn’t result in unemployed flint-knappers starving in the streets, so we’ll probably survive the singularity, too,” and equally easy to take the short view and say, “screw the singularity, I need a job that pays the bills now.”

Auerswald then discusses the possibilities for using big data and mobile/wearable computers to bring down healthcare costs. I am also in the middle of a Big Data reading binge, and my general impression of health care is that there is a ton of data out there (and more being collected every day,) but it is unwieldy and disorganized and doctors are too busy to use most of it and patients don’t have access to it. and if someone can amass, organize, and sort that data in useful ways, some very useful discoveries could be made.

Then we get to the graph that I didn’t understand,”Trends in Nonroutine Task Input, 1960 to 1998,” which is a bad sign for my future employment options in this new economy.

My main question is what is meant by “nonroutine manual” tasks, and since these were the occupations with the biggest effect shown on the graph, why aren’t they mentioned in the abstract?:

We contend that computer capital (1) substitutes for a limited and well-defined set of human activities, those involving routine (repetitive) cognitive and manual tasks; and (2) complements activities involving non-routine problem solving and interactive tasks. …Computerization is associated with declining relative industry demand for routine manual and cognitive tasks and increased relative demand for non-routine cognitive tasks.

Yes, but what about the non-routine manual? What is that, and why did it disappear first? And does this graph account for increased offshoring of manufacturing jobs to China?

If you ask me, it looks like there are three different events recorded in the graph, not just one. First, from 1960 onward, “non-routine manual” jobs plummet. Second, from 1960 through 1970, “routine cognitive” and “routine manual” jobs increase faster than “non-routine analytic” and “non-routine interactive.” Third, from 1980 onward, the routine jobs head downward while the analytic and interactive jobs become more common.

*Downloads the PDF and begins to read* Here’s the explanation of non-routine manual:

Both optical recognition of objects in a visual field and bipedal locomotion across an uneven surface appear to require enormously sophisticated algorithms, the one in optics and the other in mechanics, which are currently poorly understood by cognitive science (Pinker, 1997). These same problems explain the earlier mentioned inability of computers to perform the tasks of long haul truckers.

In this paper we refer to such tasks requiring visual and manual skills as ‘non-routine manual activities.’

This does not resolve the question.

Discussion from the paper:

Trends in routine task input, both cognitive and manual, also follow a striking pattern. During the  1960s, both forms of input increased due to a combination of between- and within-industry shifts. In the 1970s, however, within-industry input of both tasks declined, with the rate of decline accelerating.

As distinct from the other four task measures, we observe steady within- and between-industry shifts against non-routine manual tasks for the entire four decades of our sample. Since our conceptual framework indicates that non-routine manual tasks are largely orthogonal to computerization, we view
this pattern as neither supportive nor at odds with our model.

Now, it’s 4 am and the world is swimming a bit, but I think “we aren’t predicting any particular effect on non-routine manual tasks” should have been stated up front in the thesis portion. Sticking it in here feels like ad-hoc explaining away of a discrepancy. “Well, all of the other non-routine tasks went up, but this one didn’t, so, well, it doesn’t count because they’re hard to computerize.”

Anyway, the paper is 62 pages long, including the tables and charts, and I’m not reading it all or second-guessing their math at this hour, but I feel like there is something circular in all of this–“We already know that jobs involving routine labor like manufacturing are down, so we made a models saying they decreased as a percent of jobs because of computers and automation, looked through jobs data, and low and behold, found that they had decreased. Confusingly, though, we also found that non-routine manual jobs decreased during this time period, even though they don’t lend themselves to automation and computerization.”

I also searched in the document and could find no instance of the words “offshor-” “China” “export” or “outsource.”

Also, the graph Auerswald uses and the corresponding graph in the paper have some significant differences, especially the “routine cognitive” line. Maybe the authors updated their graph with more data, or Auerswald was trying to make the graph clearer. I don’t know.

Whatever is up with this paper, I think we may provisionally accept its data–fewer factory workers, more lawyers–without necessarily accepting its model.

The day after I wrote this, I happened to be reading Davidowitz’s Everybody Lies: Big Data, New Data, and What the Internet Can Tell us about who we Really Are, which has a discussion of the best places to raise children.

Talking about Chetty’s data, Davidowitz writes:

The question asked: what is the chance that a person with parents in the bottom 20 percent of the income distribution reaches the top 20 percent of the income distribution? …

So what is it about part of the United States where there is high income mobility? What makes some places better at equaling the playing field, of allowing a poor kid to have a pretty good life? Areas that spend more on education provide a better chance to poor kids. Places with more religious people and lower crime do better. Places with more black people do worse. Interestingly, this has an effect on not just the black kids but on the white kids living there as well.

Here is Chetty’s map of upward mobility (or the lack thereof) by county. Given how closely it matches a map of “African Americans” + “Native Americans” I have my reservations about the value of Chetty’s research on the bottom end (is anyone really shocked to discover that black kids enjoy little upward mobility?) but it still has some comparative value.

Davidowitz then discusses Chetty’s analysis of where people live the longest:

Interestingly, for the wealthiest Americans, life expectancy is hardly affected by where you live. …

For the poorest Americans, life expectancy varies tremendously…. living in the right place can add five years to a poor person’s life expectancy. …

religion, environment, and health insurance–do not correlate with longer life spans for the poor. The variable that does matter, according to Chetty and the others who worked on this study? How many rich people live in a city. More rich people in a city means the poor there live longer. Poor people in New York City, for example, live longer than poor people in Detroit.

Davidowitz suggests that maybe this happens because the poor learn better habits from the rich. I suspect the answer is simpler–here are a few possibilities:

1. The rich are effectively stopping the poor from doing self-destructive things, whether positively, eg, funding cultural that poor people go to rather than turn to drugs or crime out of boredom, or negatively, eg, funding police forces that discourage life-shortening crime.

2. The rich fund/support projects that improve general health, like cleaner water systems or better hospitals.

3. The effect is basically just a measurement error that doesn’t account for rich people driving up land prices. The “poor” of New York would be wealthier if they had Detroit rents.

(In general, I think Davidowitz is stronger when looking for correlations in the data than when suggesting explanations for it.)

Now contrast this with Davidowitz’s own study on where top achievers grow up:

I was curious where the most successful Americans come from, so one day I decided to download Wikipedia. …

[After some narrowing for practical reasons] Roughly 2,058 American-born baby boomers were deemed notable enough to warrant a Wikipedia entry. About 30 percent made it through achievements in art or entertainment, 29 percent through sports, 9 percent via politics, and 3 percent in academia or science.

And this is why we are doomed.

The first striking fact I noticed in the data was the enormous geographic variation in the likelihood of becoming a big success …

Roughly one in 1,209 baby boomers born in California reached Wikipedia. Only one in 4,496 baby boomers born in West Virginia did. … Roughly one in 748 baby boomers born in Suffolk County, MA, here Boston is located, made it to Wikipedia. In some counties, the success rate was twenty times lower. …

I closely examined the top counties. It turns out that nearly all of them fit into one of two categories.

First, and this surprised me, many of these counties contained a sizable college town. …

I don’t know why that would surprise anyone. But this was interesting:

Of fewer than 13,000 boomers born in Macon County, Alabama, fifteen made it to Wikipedia–or one in 852. Every single one of them is black. Fourteen of them were from the town of Tuskegee, home of Tuskegee University, a historically black college founded by Booker . Washington. The list included judges, writers, and scientists. In fact, a black child born in Tuskegee had the same probability of becoming a notable in a field outside of ports as a white child born in some of the highest-scoring, majority-white college towns.

The other factor that correlates with the production of notables?

A big city.

Being born in born in San Francisco County, Los Angeles County, or New York City all offered among the highest probabilities of making it to Wikipedia. …

Suburban counties, unless they contained major college towns, performed far worse than their urban counterparts.

A third factor that correlates with success is the proportion of immigrants in a county, though I am skeptical of this finding because I’ve never gotten the impression that the southern border of Texas produces a lot of famous people.

Migrant farm laborers aside, though, America’s immigrant population tends to be pretty well selected overall and thus produces lots of high-achievers. (Steve Jobs, for example, was the son of a Syrian immigrant; Thomas Edison was the son of a Canadian refugee.)

The variable that didn’t predict notability:

One I found more than a little surprising was how much money a state spends on education. In states with similar percentages of its residents living in urban areas, education spending did not correlate with rates of producing notable writers, artists, or business leaders.

Of course, this is probably because 1. districts increase spending when students do poorly in school, and 2. because rich people in urban send their kids to private schools.

BUT:

It is interesting to compare my Wikipedia study to one of Chetty’s team’s studies discussed earlier. Recall that Chetty’s team was trying to figure out what areas are good at allowing people to reach the upper middle class. My study was trying to figure out what areas are good at allowing people to reach fame. The results are strikingly different.

Spending a lot on education help kids reach the upper middle class. It does little to help them become a notable writer, artist, or business leader. Many of these huge successes hated school. Some dropped out.

Some, like Mark Zuckerberg, went to private school.

New York City, Chetty’s team found, is not a particularly good place to raise a child if you want to ensure he reaches the upper middle class. it is a great place, my study found, if you want to give him a chance at fame.

A couple of methodological notes:

Note that Chetty’s data not only looked at where people were born, but also at mobility–poor people who moved from the Deep South to the Midwest were also more likely to become upper middle class, and poor people who moved from the Midwest to NYC were also more likely to stay poor.

Davidowitz’s data only looks at where people were born; he does not answer whether moving to NYC makes you more likely to become famous. He also doesn’t discuss who is becoming notable–are cities engines to make the children of already successful people becoming even more successful, or are they places where even the poor have a shot at being famous?

I reject Davidowitz’s conclusions (which impute causation where there is only correlation) and substitute my own:

Cities are acceleration platforms for code. Code creates bifurcation. Bifurcation creates winners and losers while obliterating the middle.

This is not necessarily a problem if your alternatives are worse–if your choice is between poverty in NYC or poverty in Detroit, you may be better off in NYC. If your choice is between poverty in Mexico and poverty in California, you may choose California.

But if your choice is between a good chance of being middle class in Salt Lake City verses a high chance of being poor and an extremely small chance of being rich in NYC, you are probably a lot better off packing your bags and heading to Utah.

But if cities are important drivers of innovation (especially in science, to which we owe thanks for things like electricity and refrigerated food shipping,) then Auerswald has already provided us with a potential solution to their runaway effects on the poor: Henry George’s land value tax. As George accounts, one day, while overlooking San Francisco:

I asked a passing teamster, for want of something better to say, what land was worth there. He pointed to some cows grazing so far off that they looked like mice, and said, “I don’t know exactly, but there is a man over there who will sell some land for a thousand dollars an acre.” Like a flash it came over me that there was the reason of advancing poverty with advancing wealth. With the growth of population, land grows in value, and the men who work it must pay more for the privilege.[28]

Alternatively, higher taxes on fortunes like Zuckerberg’s and Bezos’s might accomplish the same thing.