Hacker News Reader

Comment by tuatoru | Hacker News Reader

An independent line of evidence looks at the formation of new occupations over time.

If automation were taking over jobs, or jobs were being refactored because some tasks are automated, then we'd expect to see new job titles appearing in employment advertising and classifications such as the Bureau of Labor Statistics's.

This does happen, as it has happened since the start of the Industrial Revolution. "Social Media Manager" is a new job that has arisen in the last several years, for instance.

But, and this is the point, this "job churn" been happening more slowly over the last 20 years than at any time since WW II.

That is also incompatible with the claim that an automation revolution is under way.

View on HN

> "Social Media Manager" is a new job that has arisen in the last several years, for instance.

How many people worked on assembly lines in, say, 1910 vs. 2010? How about as a percentage of an average manufacturer's staff?

How many people worked as social media managers in 2010?

I'd bet that (adjusted for population growth) for every 100 assembly line workers in 1910 you now have 2-3 stressed out engineers plus a bunch of machinery. And maybe one social media manager every few companies.

Sure, but somehow the total number of jobs has still managed to increase from 1910 to 2010. Automation shifts the work humans do. So far, it hasn't replaced the need for humans to do work, or we'd have a shrinking job market, which we didn't have prior to Covid (a temporary situation).

Wages for the bottom 3 quintiles have been stagnating or growing well below those of the top quintile for decades. It is because they have lost their ability to negotiate for higher wages, due to an unfavorable shift in demand and supply curves.

They are contending with additional supply of labor from other countries where people are willing to work for less. In addition, software and the ability to scale at zero cost has made people more productive, so one person can do much more than before. And it also leads to big companies getting even bigger meaning people only have a few employers to choose from.

This change also happens at a faster rate than before in history, so while in decades past, technology obviate people slowly enough that they were able to retrain and do other things, now it could be moving too fast for people to retrain (or perhaps they’re simply not capable of learning the new task).

If you track the rise of administrative staff from 1990 to 2020, you see a massive increase. For example, HBR posted that healthcare jobs grew by a total 75% from 1990-2012, and that 95% of that growth was non-medical staff.

This growth doesn't really have anything to do with necessary work. It has to do with how incentives function for management.

We do not and have never had an economy where everyone is needed for production. In economics, as the marginal increase in utility for a worker goes down you would think that there would be less workers hired, or at least they would be more evaluated prior to hiring. That doesn't take into account the competing incentives of middle and upper management who have no real incentive to increase production efficiency or where that incentive isn't properly balanced against the social and internal gain from having a larger budget and staff within an organization.

If you have the ability to hire and gain from hiring people more than you lose from their work being totally unnecessary, you hire. It's not your money you're wasting. Afterwards, you or your new hires will create systems of increasing complexity to justify the continued employment and hiring of new workers.

The human population and therefore the size of the market has also increased rather a lot in that time. The overall size of the job market isn't shrinking, but what's the median share per worker?

Also we shouldn't discount the various sophistries used to manipulate the unemployment rate. Many, many people aren't counted as 'unemployed' when they probably should be, because either they aren't applying for jobs, or they get occasional casual or gig work.

> That is also incompatible with the claim that an automation revolution is under way.

    “How did you go bankrupt?” Bill asked.

    “Two ways,” Mike said. “Gradually and then suddenly.”

Indeed, productivity growth is actually lower than in the past, so we're actually in a phase of slower automation. I think one thing that doesn't get considered enough is the viewpoint that we're currently in a post-automation revolution world. We generated a lot of excess capacity and weren't able to funnel it to productive uses, hence the rise of many things we see discussed here like "bullshit jobs," excessive administration, zero-sum industries, parasitic industries getting people to buy things they don't need, etc. My guess is that increases in automation won't eliminate jobs but will just further exacerbate these issues, unless there are some sort of societal changes.

It's funny that productivity growth is a sign of slowing automation, when it's not actually about growth in productivity as much as it is about the consolidation of capital in fewer hands and lower continued expenses.

The real messed up part is that expenses don't actually have to have gone down - you could have a smaller number of people charging the company a lot more and siphoning off money to do the same work because they don't actually run a company based on efficient production and engineering standards, they run a piggy bank. A lot of those went bankrupt in 2008 and were bailed out by the government. More are on the brink and being bailed out now.

The assumption that job titles reflect critical features of the job, or aspects of automation impacting the workforce, is...laughably easily falsifiable.

What about the replacement of W-2 with 1099 task-oriented roles? Where does that show up in the job title data set?

The 2M Uber drivers, just in the US?

The tens of millions of jobs offshored that no longer show up in US surveys?

That requirements for the job of a CEO now generally include technology...?

A piece contemporaneously on the HN front page? The "job" this person could have had 10 years ago still exists under the same title, but they are no longer qualified for it.

https://forklog.media/after-10-years-in-tech-isolation-im-no...

With all respect to Robin Hanson, his distilled claim is, to use his term, "bullshit."

It would be good for him to be subjected to automation of his job, then he might develop an understanding of the dimensions of the situation beyond the limits of the data he's myopically studying.

Having read the paper, it is based on expert classifications and summaries of job titles...it is like a game of telephone, where

real world -> noisy data collection -> noisy data classification -> noisy expert analysis -> specific study -> paper

leads to something where the paper is just too disconnected from the real world.

Also let's not forget that automation still means someone needs to write some code, which needs to be written, maintained, updated, synced with other projects etc. The act itself creates new jobs already.

For instance: In IT we also have this. Dev is the typical product development team, Ops is the team that takes care of keeping the lights on. Now with all the automation we have a new job called Devops which is somewhat in the middle betwen both Dev and Ops. We thought both groups would move to the middle and form this new Devops group. But what most companies did is completely hire from zero for their Devops teams.

Now what we see is that Devops is growing, while Dev and Ops are shrinking. But the sum of both shrinking groups is smaller than the sum of new Devops jobs.

It creates new job opportunities for those that can train for them. The open questions are always: will automation create enough, higher-skilled and better paying jobs than the ones it will substitute? How does a society re-train the newly jobless to perform more productive tasks that require higher/different skills?

Of course that given enough time the labour market will adapt and correct itself, the issue is what happens in the meanwhile with real people, with real lives. The numbers eventually balance out, real lives suffer.

Automation is generally done due to cost savings. So if you replace a thousand jobs, then you pay for automation less than you paid for these thousand jobs, so obviously there's less than thousand people involved in creating and maintaining and selling that automation. And if the automated jobs are low-paying jobs and the "automators" are mostly highly-paid engineers, then the gap is even larger.

If there's an engineer paid $200k + overhead for working on automation, then that only can happen if his/her job alone results in automating away ten $30k jobs.

> then we'd expect to see new job titles appearing

Not at all! If we replaced all accountants with software, no new jobs title would appear. If anything, they would Dissapear!

No, you'd see jobs like Business Coach, Growth Strategist, and variations on such becoming more prominent and numerous - jobs the purpose of which is to take what the software does and explain it to people, as many times and in as many ways as it takes.

You'd need to automate Early Childhood teaching through to post-grad university student supervision to eliminate those jobs. Humans trust (make decisions based on) stories.

And in fact the most important part of what an accountant does is to convey stories. Software does all the rule-based stuff.

>That is also incompatible with the claim that an automation revolution is under way.

Unless someone was claiming that the current automation revolution is different than previous ones because of software eating the world.

Automation eliminates jobs. Sometimes lower costs in one area allow more activity in another, but that does not change the fact that the automation eliminates jobs and should not be expected.

Reply on news.ycombinator.com