3 ms·

I thought this piece was really interesting. His thesis seems plausible, but I’m far from convinced. It seems equally plausible to me that as software gets eas

I thought this piece was really interesting. His thesis seems plausible, but I’m far from convinced. It seems equally plausible to me that as software gets easier to write, we just write more software. Industrialization of clothing meant the amount of labor required to create clothing dropped drastically. However more people are employed in clothing and fashion related industries than ever, because we now just own and wear more clothes than ever before. Won’t we just write more software?

gonzo41 6y ago
Industrialization of clothing meant the amount of labor required to create clothing dropped drastically Those jobs just moved to Bangladesh. I think they are just as inefficient as always. But the west traded textile jobs for other forms of service industries and more complex manufacturing. No one makes clothing inside a western country as cheaply as they do in Asia, but the conditions in Asia are not heavily automated. They just have more people and an economies at a stage where that sort of work will provide. I suspect we'll just make more software. You still have to think and there's an upper limit to how much you can automate, and how costly that last jump from some automation to full automation will take.
- Apocryphon 6y ago
  And what happens when Bangladesh (then Pakistan, then Ethiopia, then etc.) becomes developed enough? Eventually you run out of countries and regions to offshore to.
  - IshKebab 6y ago
    Clothes just become more expensive.
- jaredklewis 6y ago
  To clarify, in my metaphor I am comparing the efficiency of labor before and after the industrial revolution. Bangladesh is an industrialized country, so I am not sure I understand your point. To weave enough cloth to make a single shirt by hand takes orders of magnitude more time than it takes for a loom to produce the same amount of cloth. This is true regardless of what country it happens in and how much the person operating the loom is paid. Yes, the cheap clothes one can buy at H&M, Uniqlo, or Gap are made in Southeast Asia, usually by workers being paid inhumanely low wages. But those clothes are universally made with fabric that is produced by looms and other modern machinery. My point is just that the volume of clothes produced today (in southeast Asia, for example) would require orders of magnitude more workers to produce in a pre-industrial world. Despite this massive efficiency jump, there are not less jobs in clothing and fashion related industries; there are more because we wear so much more clothes than we used to.
  - selimthegrim 6y ago
    It may be middle income these days but I would call it industrializing rather than industrialized
adrianN 6y ago
Eventually the market for software will be much more saturated than today. It has already happened for consumer desktop applications. The rate of new commercially successful desktop applications has dropped in the last thirty years. Sooner or later software for businesses will be "done" too.
- axaxs 6y ago
  I think that's just a shift. It's not that the market for the functionality dried up, it just all moved to the internet. And that market is everbooming.
- chii 6y ago
  > The rate of new commercially successful desktop applications has dropped that's because a lot of what used to be possible only on desktop has been moved into web-based SaaS or mobile. If you count mobile phones as "desktops" - there's millions more apps on mobile than there are on real desktops.
- harpratap 6y ago
  This is like saying since diesel cars are dying therefore entire transportation industry is coming to an end.
  - adrianN 6y ago
    No, it's saying that since most people in developed countries who want a car already have a car, the market for cars is markedly different from the first half of the 20th century.
- gitgud 6y ago
  > Sooner or later software for businesses will be "done" too. Software will likely dominate all other industries, as the absolute essence of automation is software.
egourlao 6y ago
My dad loves to tell this story about him going to engineering school in France in the 70's. He was suggested to stay away from CS or software by his guidance counselors - the reasoning? "We'll soon enough have programs able to write themselves - studying to be a software engineer is a dead end". I'm not sure that some common engineering problems now being solved doesn't just mean that we can now redistribute that workforce on other unsolved problems, or to build new products on-top of those solutions.
- rhizome31 6y ago
  I had a similar experience in France in the early 2000s. I went to the unemployment office (ANPE) to ask for a training and they told me that software was a dead end and that it wasn't possible to get training. So I did my own research, found a training organization and went back to the unemployment office giving them the exact reference of the training I wanted. The exact same person then reluctantly pulled a drawer of her desk and gave me an application form. That training kick-started my career in software. Another thing that was commonly said over the 2000s period is that all software jobs would be quickly outsourced to India. That didn't happen either. Some big companies obviously tried that (and some still do) but a lot of them eventually realized that software wasn't something you could just specify and outsource to a team far away, with a different time zone, culture and language and expect to magically get working software as a result. In the meantime, I think that demand for software in India has also grown. I laughed at loud when in the early 2010s I received an email from an Indian recruitment agent offering me to come work in India.
fabatka 6y ago
I agree with you, by the way the phenomenon you describe is known as the Jevons paradox: https://en.wikipedia.org/wiki/Jevons_paradox https://en.wikipedia.org/wiki/Jevons_paradox
st1ck 6y ago
It's kinda weird comparison, because (beyond necessary) buying clothing is consumption. But buying software is often an investment, so people and businesses can justify spending enormous amount on it, compared to clothing.

gonzo41 6y ago

Industrialization of clothing meant the amount of labor required to create clothing dropped drastically Those jobs just moved to Bangladesh. I think they are just as inefficient as always. But the west traded textile jobs for other forms of service industries and more complex manufacturing. No one makes clothing inside a western country as cheaply as they do in Asia, but the conditions in Asia are not heavily automated. They just have more people and an economies at a stage where that sort of work will provide. I suspect we'll just make more software. You still have to think and there's an upper limit to how much you can automate, and how costly that last jump from some automation to full automation will take.

Apocryphon 6y ago
And what happens when Bangladesh (then Pakistan, then Ethiopia, then etc.) becomes developed enough? Eventually you run out of countries and regions to offshore to.
- IshKebab 6y ago
  Clothes just become more expensive.
jaredklewis 6y ago
To clarify, in my metaphor I am comparing the efficiency of labor before and after the industrial revolution. Bangladesh is an industrialized country, so I am not sure I understand your point. To weave enough cloth to make a single shirt by hand takes orders of magnitude more time than it takes for a loom to produce the same amount of cloth. This is true regardless of what country it happens in and how much the person operating the loom is paid. Yes, the cheap clothes one can buy at H&M, Uniqlo, or Gap are made in Southeast Asia, usually by workers being paid inhumanely low wages. But those clothes are universally made with fabric that is produced by looms and other modern machinery. My point is just that the volume of clothes produced today (in southeast Asia, for example) would require orders of magnitude more workers to produce in a pre-industrial world. Despite this massive efficiency jump, there are not less jobs in clothing and fashion related industries; there are more because we wear so much more clothes than we used to.
- selimthegrim 6y ago
  It may be middle income these days but I would call it industrializing rather than industrialized

adrianN 6y ago

Eventually the market for software will be much more saturated than today. It has already happened for consumer desktop applications. The rate of new commercially successful desktop applications has dropped in the last thirty years. Sooner or later software for businesses will be "done" too.

axaxs 6y ago
I think that's just a shift. It's not that the market for the functionality dried up, it just all moved to the internet. And that market is everbooming.
chii 6y ago
> The rate of new commercially successful desktop applications has dropped that's because a lot of what used to be possible only on desktop has been moved into web-based SaaS or mobile. If you count mobile phones as "desktops" - there's millions more apps on mobile than there are on real desktops.
harpratap 6y ago
This is like saying since diesel cars are dying therefore entire transportation industry is coming to an end.
- adrianN 6y ago
  No, it's saying that since most people in developed countries who want a car already have a car, the market for cars is markedly different from the first half of the 20th century.
gitgud 6y ago
> Sooner or later software for businesses will be "done" too. Software will likely dominate all other industries, as the absolute essence of automation is software.

egourlao 6y ago

My dad loves to tell this story about him going to engineering school in France in the 70's. He was suggested to stay away from CS or software by his guidance counselors - the reasoning? "We'll soon enough have programs able to write themselves - studying to be a software engineer is a dead end". I'm not sure that some common engineering problems now being solved doesn't just mean that we can now redistribute that workforce on other unsolved problems, or to build new products on-top of those solutions.

rhizome31 6y ago
I had a similar experience in France in the early 2000s. I went to the unemployment office (ANPE) to ask for a training and they told me that software was a dead end and that it wasn't possible to get training. So I did my own research, found a training organization and went back to the unemployment office giving them the exact reference of the training I wanted. The exact same person then reluctantly pulled a drawer of her desk and gave me an application form. That training kick-started my career in software. Another thing that was commonly said over the 2000s period is that all software jobs would be quickly outsourced to India. That didn't happen either. Some big companies obviously tried that (and some still do) but a lot of them eventually realized that software wasn't something you could just specify and outsource to a team far away, with a different time zone, culture and language and expect to magically get working software as a result. In the meantime, I think that demand for software in India has also grown. I laughed at loud when in the early 2010s I received an email from an Indian recruitment agent offering me to come work in India.

fabatka 6y ago

I agree with you, by the way the phenomenon you describe is known as the Jevons paradox: https://en.wikipedia.org/wiki/Jevons_paradox https://en.wikipedia.org/wiki/Jevons_paradox

st1ck 6y ago

It's kinda weird comparison, because (beyond necessary) buying clothing is consumption. But buying software is often an investment, so people and businesses can justify spending enormous amount on it, compared to clothing.