Looks like software application development is listed as having a relatively high automation "risk", around 21%. Anyone know how this was calculated. Doesn't h

Looks like software application development is listed as having a relatively high automation "risk", around 21%. Anyone know how this was calculated. Doesn't have to be a link for this particular site, just some good reading material. My initial reaction was that this number is hard to nail down because on reflection, so much of what I do gets automated. That's the nature of software. When I started programming for money in the late 90s, I still managed memory. That's automated now through garbage collection (for me). I also used to do a lot of direct network and socket programming, also automated away. I managed my own database connection and pooling. JDBC came along and abstracted that away, but a decade later, I heard someone who used an ORM say he wouldn't want to write all that "low level" jdbc code ;) I guess you're getting old when the high level leaky abstraction of your youth becomes the low level code for a new generation of programmers. My whole career has been one long progression of getting automated away, by this perspective. I remember when I proposed getting a web server up and running for a project in college, and a professor remarked on what I was doing. There was a time, he recalled, when doing a "spreadsheet" was fairly advanced and took some serious technical chops. The same would happen to web servers - but for the few years then they required specialized skill and had a big impact on business, this knowledge would be very valuable. The moment would pass, but something new would come up, and that would create all kinds of new apps that require an app developer. To me, the moment when nothing new comes up - ie., when there is no new moment for human app developers, all software development is now automated - that is the moment when we replaced by automation. I wonder if that's what this article is talking about? If that's the case, 20% might seem high, though not out of the question. But by a more general definition, the one I discussed above, I'd ay the odds that my current skillset will be automated away is roughly 100%. I want to be clear I'm not (necessarily) talking about churn, I'm talking about new breakthroughs that fundamentally change the job. That kind of change I don't mind at all, I enjoy it, it's probably why I'm in this field. Churn, on the other hand, is what makes me want to quit. I'm rambling on here, but it occurs to me that software development probably isn't that unusual. It's the jobs that don't change as much that are unusual. Lawyers, for instance, have certainly changed over the years, but we expect people to still be lawyers in 100 years. How many horse and buggy drivers expect that job to still be around? How many lathe operators or truck drivers expect this? I'm thinking, maybe the best metric isn't the odds that a particular job will be automated, since for many (most?) jobs, that's pretty much 100%. The question is, what are the odds that a job will be automated with no path to the next thing. That's the fundamental transformation going on here. And yeah, it actually is happening. A horse and buggy driver can become a bus driver, but what happens when the concept of a drive goes away completely? Will anything emerge? Another aspect of automation I'm interested in is the complete bindside. Music, I think, was one of the interesting automation stories of the last 100+ years. There were various attempts to replicated the human mechanical process of playing an instrument, and they all appeared very unlikely to do much replacement. Yeah, piano rolls made an appearance, and you could try this I suppose by hooking up some contraption to make a flute play. But what happened instead was that it turned out you can actually store music physically in various materials and play it back. Music went from 100% live to less than 1% live in about 50 years. But we didn't really automate music, we replaced it with a similar product. My guess is that this will be the story with software development. It seems absurd to me that this could happen through automation, but that's because I'm locked into one way of thinking about it. Kind of like trying to imagine how a crankshaft could power a bunch of strings tied to a bow and some pegs to try to recreate the physical process of playing a violin. It won't happen like that. Something utterly mind blowing, like storing music in grooves of different depth in vinyl, of all things, utterly unanticipated. Yeah, that's generally how this massive shifts happen.