On the other hand, as a nurse my wife can relay stories going back to her earliest working days where more experienced medical personnel often had to step in to prevent a newly minted doctor from harming or potentially even killing a patient. I remember that giving oxygen was often the issue here, where the new doctors seemed to think that giving more oxygen was always better, but the nurses knew that giving too much oxygen can put a patient into crisis. (I forget why.) She hasn't mentioned this particular issue in ages now, though, so I guess these days medical schools are doing a better job here.

According to the Stack Overflow's 2016 Developer Survey [1], "56% of developers in fact do not have a college degree in computer science or related fields"

Admittedly, the data is a few years old (and I didn't want to pull the same data from the 2018 survey), but such a requirement would put many people out of a job. I also haven't seen any concrete data that would support any correlation between the efficacy of a software engineer and whether they had obtained a college degree, so this seems like a poor metric.

I share your incredulity regarding the lack of any sort of "occupational license" for the industry, though I remain unconvinced that instituting one would have a positive effect.

[1] : https://stackoverflow.blog/2016/10/07/do-developers-need-col...

Except that the degree itself may not be worth much in practical terms. For example, as a working student (paid computer nerd by day, student by night) I quickly came to the conclusion that much of what I was being taught in the CS program just wasn't going to be useful out in the real world. For example, at one point I was taught the detailed in and outs of two's complement arithmetic, but that's something that I've never used in practice, and I've only even seen it referenced maybe once every decade or so, if that. (See the following link for a very recent drive-by encounter with this stuff: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p090...) Meanwhile, pretty much anything that's been of any real value to me on the job I've taught myself on a just-in-time basis.

As more recent examples, a few years back I took all three of the original free MOOCs from Stanford. In the ML class, Andrew Ng at one point said something like "If you've ever taken calculus then you will recognize this particular equation as being a derivative; otherwise don't worry about it because that's not really important." Later he said something similar about linear algebra and matrix multiplication. And indeed I had taken both calculus and linear algebra back in the day so I recognized what I was looking at, but it just wasn't important within the context of that advanced class.

I also took the Database MOOC, and during that the instructor taught quite a bit of formal database theory. (I forget the proper name for it.) But despite having worked with databases for decades now I've only seen such theory referenced maybe twice, and that just in passing, so it's hardly the "critical, need-to-know" stuff that the instructor insisted that it was.

Meanwhile, in the AI MOOC almost everything taught there was new to me, despite my having taken several AI classes back in the day; so basically zero overlap between then and now. (I did recognize some of the statistical stuff, though, from my statistics class way back when.) But will any of the new stuff ever be of any practical use to me? Probably not. Interesting certainly, but probably not at all practical.