The End of Capitalism: The Academic Delusion

For blue collar workers, the root of all evils was the 1980s. While the world was listening to Madonna and Michael Jackson, wearing giant shoulder pads and holding giant boomboxes over your head, a series of developments crushed into the American Dream like Jaws into a fishing boat. The result was a deep and lasting depression into the heart of America, one that made life outside of large cities precarious and poverty and disillusionment widespread.

It was a cocktail of changes, each of them largely independent of the others, and yet the combination of them sucked wealth into a funnel and moved it to a few different places, like a vacuum that sucks pay dirt into a bag and cleans out your living room. Only that the dirt was gold dust and you depended on it for survival.

We have been following the consequences of the demise of blue collar work for forty years now. Even a modest revival of some manufacturing is not going to bring back the industry of decades past, and automation, globalization, and communication are here to stay. There will always be someone in some country far away that is willing to work for a tenth of what you absolutely need to survive; if not, then there will be a machine that will do that for a small yearly maintenance fee.

But similar changes to what happened in the 1980 are approaching rapidly now, and the target is a series of jobs we considered absolutely safe and where the human touch was irreplaceable. Unlike with the restructuring of blue collar occupation in the 1980, though, this latest change will not take decades and white collar work is threatened in a matter of weeks.

What Happened to Blue Collar Jobs?

Automation is a giant gift from humanity to itself. Take something boring, tedious, dull, or dangerous and find a machine that can do it. The result is a potentially happier humanity that is more productive and machines that don’t mind. (Until the Final Uprising of the Robot Race, that is.)

The first jobs that disappeared thanks to automation came with the Industrial Revolution, just around the time America was founded. Back in the day, there were still plenty of bona fide slaves around, so the fact jobs were automated was not as big a deal as now. It’s still not a coincidence that the American Civil War happened, though: the industrial centers of the North were perfectly fine without slavery, an institution that was still largely vital to the economy of the South.

The invention of the assembly line allowed an enormous increase in production quantity, quality, and productivity that made paying higher wages possible. At the beginning of the 20th century, that translated to a huge increase in wage earning potential.

Many years later, the desire to increase profits drove a new round of automation, a third industrial revolution. The key changes that made this one possible were globalization of production markets, advances in industrial machining, and the advent of the computer age. These three were largely unrelated, but worked together to swiftly destroy the industrial base of wage earning. Production could either be automated or shipped off abroad, and the newly connected world was able to track everything happening, so that replacing local handiwork with global automation didn’t result in loss of accountability.

How Are White Collar Jobs Threatened?

Computers of the 1980s were dumb. They could only be programmed to do dumb things that were predictable and linear, like screwing in the same screw on a car’s body over and over all day. Sometimes that dumb looked smart, like with Eliza, a software program that pretended to be a psychologist but that you’d unmask in a minute as the dumb, predictable thing it was.

Computers of the 2020s are less dumb. They are capable of extracting a significant portion of the knowledge of humanity and applying them to problems at hand. We call it artificial intelligence, but it’s still largely dumb. It looks very impressive on first glance, but closer inspection reveals that it’s all smoke and mirrors.

While we are not dealing with intelligence, that doesn’t mean there isn’t a good use case for it. After all, most blue collar jobs were taken away by machines dumb enough to only ever be able to screw the same screw into a car’s body!

What computers of the 2020s can do that those of the 1980s couldn’t is information processing. That has a technological component: computers of today are much better at computing – they are faster, have larger memory, more storage capability, and better programming languages. But it also has a connectedness component: more and more information is available digitally, from automated temperature sensors to searchable public records.

The result is that computers are increasingly getting a clear picture of the world as it appears to humans. They can process this picture in a way that humans don’t (or maybe even can’t). So they can take on work from us that was until now the exclusive domain of humans.

It’s not the first time this happened. In the ancient days, a calculator was a person that performed calculations, like for an accountant. It wasn’t a very high status jobs (as Charles Dickens teaches us), but it was enough to feed a family. As soon as computers showed up, calculators became so obsolete that nobody thought it confusing to reuse the word for a handheld gadget that performed arithmetic calculations.

Just as the first kind of calculator became obsolete, the second kind followed in its footstep. Not a lot of people (outside of high school) still use calculators, because those tellingly have been replaced by smartphones. This is emblematic, because the second type of calculator is the type of machine used in the 1980s to destroy blue collar jobs – single purpose, dedicated, dumb. The third type of calculator is the type of machine that will be used in the 2020s: universal, powerful, connected.

But Surely AI Can’t Replace Humans!

You hear it often that white collar professionals (we don’t call them workers in the same way that we say expats and not immigrants) think their jobs safe, because there is a component of it that cannot be (easily) automated. “I want to see a computer perform surgery,” or “Nobody is going to trust a computer with their billions,” or “No computer is ever going to convince a jury.”

There is some merit to those claims: there will probably always be components of white collar jobs that cannot be automated or will not. But here is the clinch: that’s also true for blue collar jobs. What is dramatically more important is what component of white collar jobs can be automated, and it turns out that is the vast majority of it.

Blue collar workers are already familiar with the cruel logic of automation: first, it is used to improve the lives of existing workers; second, it is used to replace the most menial and dangerous of jobs; third it starts cutting into job availability. This creates pressure on workers (sorry, professionals): work is harder to find, so more money needs to be extracted from each job. Finally, the pressure creates a tipping point: professionals must demand more and more money for the same quantity of work, which generates more pressure to automate them away. It’s not a coincidence that the labor movement was crushed in the 1980 because it had overextended itself trying to save jobs.

Take a law firm as an example. The partners will first start automating low level functions like those of paralegals. When you can just search in a law database for relevant cases, you don’t need someone highly skilled to do that. You might find a low-skilled intern, or you might leave it to the lawyers to do the same job. You can pocket the resulting difference.

Eventually, clients will realize they can do the same. They are still paying for the work of the paralegals at full law firm hourly rates, so they will start performing it to save. The technology, after all, is available to everyone. That translates to lower revenue for law firms and a significant amount of work is not performed by them any longer.

Finally, the law firm will have to jack up prices to keep revenue constant. The result is that more clients are going to get rid of law work because they can automate it more easily, while the law firm is costing more. Notice that this is all gradual, just as in the case of blue collar jobs being automated. It will affect the lowest hanging fruits first, but picking those creates a pressure that dooms the entire system.

Unlike with blue collar jobs, though, this transition does not require significant capital expense. You can automate a factory, but it’s going to cost a lot of money to get the infrastructure and machinery. Even offshoring that factory comes with the costs of shipping: the actual cost, the delay in arrival, the variability of shipping risks. Replacing a law firm only requires access (not even purchase) of software. And every company on the planet more or less has access.

Why White Collar Work?

The human race has as defining trait information processing. We are intelligent, which means we can take information and reach conclusions. For millennia, that was enough to set us apart from all other earthly animals.

The problem for humans has always been that while we are really good at information processing, we have no built-in mechanism for information retrieval. If we have a hypothesis/rule, then we can reach conclusions. But we don’t have a way to get to that rule that is built into us.

We have invented great ways to pass on information. We invented speech to communicate it, we invented writing to store it. We invented science and the art to advance it. But, for millennia, all of that was reserved to a select few.

Two millennia ago, Jesus was talking about scribes – the high-ranking individuals whose claim to fame was the ability to write. In the Middle Ages, universities were founded on the premise of teaching the law and medicine and finance, science, arts, and theology.

What stands out about those universities is that they were simply information transfer centers. Information was moved in a systematic and comprehensive way from one set of individuals (professors) to another (students). What the students would do is use that information to process more information. Medical students would become doctors, who were asked to figure out what was wrong with someone (diagnosis) and to cure them. Both functions are largely of the information processing variety. Lawyers were asked to figure out what laws and cases applied to a specific situation. Financial guys were asked to figure out what investments were worthy. Insurance guys what risks presented themselves.

All of that is about information. One component of it is information retrieval, which for a long time was the domain of humans. The other part is information processing, which we have taught computers extensively.

In the 2020s, information retrieval is something computers can do on their own. Which means the last domain of human exclusive is gone, and the vast majority of information processing work, which means of white collar professions, can be automated.

Why Do White Collar Workers (Professionals) Feel Safe?

A lot of the discussion since the 1980s has focused on the difference between “good” jobs and “bad” jobs. The dividing line is generally income more than benefits. A “good” job is one that has high pay for a regular amount of stable work with predictable hours. A “bad” job is one that pays little, is precarious, and has variable hours.

We have come to think of white collar jobs as “good” with a few exceptions. Teachers are of low pay, usually; sales jobs are typically not very stable; medical professionals in emergency settings have typically variable hours. But all in all, white collar professionals have “better” jobs.

It is natural to think that this quality is inherent to jobs that require education, and in fact education became very expensive as a result of the pressure to use it to improve one’s (professional) life. Because education was used as an investment in job quality, and because education takes time in addition to money, white collar workers feel safe.

Indeed, it is common for white collar workers to mention the amount of time they studied their field to impress others. Who hasn’t heard of the lawyer or physician that are frustrated because they are dealing with clients/patients that think they can form a professional opinion after googling for five minutes when they studied the field for years?

This gives this type of professional a false sense of security. Once the relevant information from their fields is digitized, it takes computers minutes to process it, not years. The barrier to being replaced is then not the human speed of knowledge acquisition, but the rate of digitization. Because white collar professionals are used to thinking of the cost of processing (learning) as the barrier to entry, they do not realize the speed of replacing their function is determined by the rate of digitization.

What Are We Going To Do About It?

We could of course learn from the catastrophe that was the Demise of Blue Collar Work and set in motion changes that would prevent a repeat. We could see the approach of the Castle Defense of the beleaguered professions extracting more money out of a dwindling customer base, we could see the issues faced by the weakest professionals – those freshly out of college. But we haven’t, so far.

The problem is, as mentioned, that this new revolution is coming much faster than the older one. While he have a template from which we could draw information and guide our future, we don’t have the time to do so. Unless we rush to a solution, we will be the problem.

The background to this, though, is a series of professions and professionals that have been pampering themselves with the idea that job loss is other people’s problem. That of the “uneducated.” The irony of it is that it’s precisely their role as information processing, educated individuals that makes the jobs of white collar professionals obsolete, very quickly.