[jandrewrogers]

To rephrase:

Surpluses and shortages of skilled labor are very unevenly distributed. Differences in ease of automation will magnify this. If there is a demand for 100 neurosurgeons and then cut everyone's hours by 20%, you effectively created a shortage of 25 neurosurgeons. Decreasing hours doesn't increase supply and supply of highly skilled labor is not fungible i.e. you can't trivially retrain a PhD in electrical engineering or truck driver to become a neurosurgeon.

This leads to the following conundrum:

If we forcibly cut hours for everyone across the board then it will create severe supply shortages for the most highly skilled labor that is most difficult to automate, some of which already have severe shortages because it is so difficult to create supply. If we cut hours such that labor supply is proportional to demand then the most highly skilled labor that is most difficult to automate will be required to work by far the most hours, which isn't fair to highly skilled labor and creates a disincentive for required labor.

Systematically reducing working hours may benefit the majority but it creates perverse social and economic dynamics for the highly skilled minority whose labor society can't easily replace.

[Retric]

We already forcibly cut doctors hours by giving them inefficient education and excessive paperwork. A doctor working 3 days a week in an efficient system could spend more time with patients in a lifetime than the average US doctor does.

The US labor market only really has shortages by design.

[jandrewrogers]

I agree that some labor shortages are the product of artificial restriction. But I strongly disagree that all labor shortages in the US have this property. Some labor shortages are unambiguously intrinsic and very difficult to eliminate. Sometimes there simply aren't enough people with deep expertise necessarily acquired over several years to meet demand.

Take my specialty for example. I mostly work on operational multi-modal spatial and sensor analytics at extremely large scales and high velocities (as is typical for these data models). Right now, half the Fortune 500 are trying to hire people that know how to design these systems and throwing silly money at anyone that seems like they can. There is no open source software that can do it and half the required computer science is not in literature, it is an extremely deep technical specialty that takes years of experience to learn. There are, maybe, a half-dozen people in the world right now that know how to design these systems end-to-end from first principles and likely a demand for several hundred. There is no way to manufacture that supply on a time horizon that matters to anyone that wants to hire them.

Even one level lower, high-end systems engineering talent demand is at least an order of magnitude higher than the actual supply. This requires very deep experience to be competent that you can't learn in bootcamp or six months of on-the-job training. Yet despite being paid extremely well even by software engineering standards, as an industry we don't come remotely close to producing enough of them. In fairness, it takes serious devotion to craft and no small amount of talent to become high-end systems engineer -- but few people with the raw talent have that ambition or interest, even though it pays extremely well. You can't force people to do what they have no interest in doing.

[Retric]

6 months may not be enough to train up someone without a technical background, but when a skill shortage extends beyond the time horizon of training a pool of 100’s of thousands of people up it’s very much self inflicted as the company simply does not want to pay market rate + training.

PS: People with related skills can always pick up these deep specialty skills with extreme speeds. I have seen someone paid contractor rates to learn a extreme specialty. Including that training, he actually finished the project in less time than the original team had wasted.

[jandrewrogers]

There are many deep specialties that no one picks up with "extreme speed" no matter how technical they are, certainly not to the level required by companies that want to hire these skills. Think database kernel engineering or non-trivial parallel systems design. Acquiring these skills happens almost exclusively by apprenticing for years with real experts. In six months you could go from no skills to mediocre skills with a lot of training, but no one wants "mediocre" working on their database kernel for good reason.

It would be like me assuming that I, a broadly competent technical expert, could quickly and easily develop a deep expertise in e.g. high-performance graphics engines. A diagnosis of Dunning-Kruger would not be incorrect were I to make such an assertion.

[Retric]

This is not a new problem.

https://en.m.wikipedia.org/wiki/California_Institute_of_the_...

Look into it’s history and Walt Disney has been training animators for decades. It’s a deep skill that is takes significant time to master, so you need an actual pipeline.

Continuing the idea, NASA trains astronauts. They don’t need very many world wide, but they need a few and the only way to get them is to train these people.

I could go on, but outside of a months to few years for absolutely new fields shortages are by design.

[petra]

The assumption that highly skilled jobs are harder to automate may turn out as wrong.

Also, many given the right augmentation tools, the training time for highly skilled jobs may be much shorter.

Doctors(but maybe not neurosuregeons) is such case, because of regulation. Nurses , with some training and tools could replace many.

But this could be true for other professions.

[SubiculumCode]

Forcing hour cuts would seem to me to have more unintended consequences than sat UBI policies.