Comment by crdoconnor | Hacker News Reader

Comment by crdoconnor | Hacker News Reader

Just once I'd like to see a balanced study on automation that:

* Properly takes into account jobs created by technology as well as destroyed.

* Properly takes into account jobs destroyed by austerity (e.g. government budget cuts).

* Doesn't conflate outsourcing with automation - e.g. statistically pretending that because a job no longer exists in the US, that it doesn't exist at all or (worse), as the ball state university study did, literally (and probably deliberately) mixing cost savings from automation and cost savings from outsourcing.

Instead all we get is the steady drumbeat of that old investor pipe dream - the idea that employees are an unnecessary expense that will soon be dispensed with as soon as we find a deep learning model thats sophisticated enough and studies like the ball state U one that argue that the investor pipedream is coming true at the same time as (weirdly) showing their political affiliations by arguing for lower corporation taxes.

At the same time the investor automation hysteria has been good for my wages, so... maybe I should just shut up.

What's wrong with the ball state university study? Are there conflicting studies which show the opposite?

It seems fairly obvious that any new automated technology removes more jobs than it creates. A team of 20 engineers designs and maintains self service checkouts, they're fabbed by maybe 100 more, and then thousands of grocery workers have their shifts cut.

Here's the part which makes me feel a little guilty:

The self service machines are priced according to what the customer is willing to pay, so as to be competitive with the labor they replace, so in effect the money that used to be going to those thousands of grocery employees end up being funneled to the handful who do the automating.

The economy becomes more and more winner-take-all, the gini coefficient rises a little bit more, and like you said we all enjoy our wage increases.

> It seems fairly obvious that any new automated technology removes more jobs than it creates

The clearest real world check for this theory should be the unemployment rate.

If you're right, it should be steadily rising as more and more jobs are automated away.

The unemployment rate doesn't include those who have given up, it's useful to look at the labor force participation rate. In particular for men ages 25-54 which are considered "prime aged": https://www.bls.gov/opub/mlr/2018/beyond-bls/mens-declining-...

Edit: here's a 2016 article showing an instance of unemployment rate going down (which looks good) because labor force participation rate dropped (which is not good): https://northcooknews.com/stories/511017085-illinois-workers...

> While the unemployment rate dropped to 5.5 percent, it is still higher than the national average of 4.9 percent. In addition, the decrease is mainly due to unemployed workers dropping out of the labor force. The number of people on and off state unemployment who are looking for work determines the unemployment rate. When workers stop looking, they are no longer counted, which leads to an apparent decline in unemployment.

It's a bit too early. Last year only around ~400K new robots we're bought.

This won't stay that way.

I have a different perspective.

Since the industrial revolution ~250 years ago, human workers have constantly been replaced by machines. All this time people have been sure it would result in unemployment, and all this time they've been wrong.

A worker today is about 30x as productive as one 250 years ago, because of this technological development.

Now, of course, it could be different this time. The past doesn't dictate the future etc. But at a minimum I'd want to hear why from someone who understands the history.

One historian that see AI as different is Yuval Noah harari. He has interesting books.

I'm reading Sapien right now.

>>It seems fairly obvious that any new automated technology removes more jobs than it creates.

It's obvious in the naive way that it's obvious that breaking windows creates jobs. It's a fallacy from over-simplification.

Automation results in labor savings that can then go into providing more of those services that cannot be automated.

Financially this manifests as demand for sought after human-provided services increasing when consumers see cost-savings on services that are automated.

Yea I see that in the long run, it's after all how we ended up with our current labor composition. But these jobs will be for other people in other places.

As an example, jobless auto workers didn't become car dealers in LA: by the numbers they seem to have dropped out of the labor force [1], hopped onto disability [2], and/or died from drug overdoses [3].

[1] https://www.brookings.edu/wp-content/uploads/2017/05/ccf_201... [2] https://apps.npr.org/unfit-for-work/ [3] https://www.prb.org/people-and-places-hardest-hit-by-the-dru...

Will our economy evolve to its next phase? Yes. Will many people suffer along the way? Yes.

Automation creating a net increase in opportunity means that generally, people who lost from the industry they work in being automated will see that more than made up for by what they gain when other industries are automated.

Amazon's automated deployment of web servers eliminated jobs in server hosting companies that set up dedicated servers for people but generated massively more opportunities by making it much cheaper, in time and money, for people at large to deploy their own server.

The proliferation of web based businesses in turn expands new classes of occupations that provide people with more niche services that enhance their quality of life.

> generally

Yes, generally this is the case. I'm just pointing out that historically, not everyone has won equally, and some have lost.

That goes with life in general. Equalizing gains with mandatory measures comes at the cost of a less responsive economy with less gains overall.

>It seems fairly obvious that any new automated technology removes more jobs than it creates.

This argument seems to suggest that if, hypothetically speaking, 65% of the population had jobs that were automated away such that only 1% of the population were needed to do that work, that 64% of the population subsequently would become unemployed.

Given that that example is actually not hypothetical and is very, very real (agriculture), why do you think we never struggled with 64% unemployment?

Well by some measures we did struggle a lot during these shifts. The last industrial revolution was a period of social unrest, featuring mass riots which killed people. As a response the government instituted universal high school and labor union friendly laws (and labor day became a holiday).

As for what people ended up doing, new industries opened up, but I wouldn't say they were created by the specific new technology per se... maybe this is just semantics but if automatic tractors frees people up to be hired for instagram video production, can one say that automatic tractors "created" the instagram video market?

>Well by some measures we did struggle a lot during these shifts. The last industrial revolution was a period of social unrest, featuring mass riots

Are you saying that automation doesn't necessarily cause joblessness but that it did cause riots?

The industrial revolution was a pretty nasty struggle for sure, but I'd argue that was more down to the enclosure movement - people being kicked off their land - than machines.

>As for what people ended up doing, new industries opened up

And your argument hinged upon the notion that this process has now somehow... stopped... why so?

>maybe this is just semantics but if automatic tractors frees people up to be hired for instagram video production, can one say that automatic tractors "created" the instagram video market?

A lot more than just instagram video market was created after the automation of farming. Frankly, I see the process of market creation being threatened more by wealth inequality than by automation.

> And your argument hinged upon the notion that this process has now somehow... stopped... why so?

Apologies if I was unclear, I did not mean to imply that opportunities would stop coming up, I only meant to draw attention to the fact that within the industry there would be fewer jobs.

(Ultimately this is a very good thing, we can't get much done if 9/10 people are dedicated to farming. In 30 years I'm sure we'll look back and say the same about manufacturing.)

But this is why it's important:

Economic theory says that workers are all fungible and will re-skill into these new jobs, while data show that the new jobs are taken by other people in other places: the old workers seem to drop out of the workforce [1], get on disability [2], and/or die from opiate overdoses [3].

[1] https://www.brookings.edu/wp-content/uploads/2017/05/ccf_201...

[2] https://apps.npr.org/unfit-for-work/

[3] https://www.prb.org/people-and-places-hardest-hit-by-the-dru...

I do agree that the inequality is a big issue and it's doubly edged: less consumer spending power means the market doesn't bother serving those without wealth, and automation itself tends to concentrate wealth into few large players.

>It seems fairly obvious that any new automated technology removes more jobs than it creates.

Widget A has $10 labor costs and $10 material costs. Widget B has $2 robot costs (=robot engineers labor) and $10 material costs.

My budget is $200. When I buy 10 Widget B I will have $80 remaining that I can spend on something else which will pay someone else's salary and therefore create at least as many jobs as were destroyed by the robot.

Unless of course your job was automated out of existence by a robot...Seems like a bit of an oversight in your calculations hmm?

> I will have $80 remaining that I can spend on something else which will pay someone else's salary

Will you pay someone else's salary or will you issue a stock buyback?

So far it seems like the later is happening, and as a result wealth accrues at the very top percentiles.

So in the past, out of that $200, $100 went to labor.

Now, only $80 are left, and capable robots also compete for that. So maybe you spend only $40 on labor.

So now you support only 40% percent of jobs. And that's without considering second order effects.

> The self service machines are priced according to what the customer is willing to pay so as to be competitive with the labor they replace

I rather doubt that. These are computers in big boxes; if your margin is too high, someone else will do it cheaper, possibly the customer themselves. (unless there's a patent?). Now, the sales are certainly more likely to come in when your price compares favorably with labor.

This is more or less the pitch I've seen when a tech alternative first comes out. For example "web advertising vs TV advertising".

But I guess that's because there's not competition yet.

Reply on news.ycombinator.com