This was an amazing piece of testimony.

We are seeing this all around us every day.

I can tell you that in engineering it is a problem where fundamentals have been lost in favor of transitioning to technology.

I had a conversation not too long ago where someone was very surprised I was a heat treater under 40.  Almost all of materials science has gone to semiconductors and silicon technology and meta-materials.  There is very little “turn iron ore into pig iron, pig iron into steel, steel into parts” taught in materials science anymore.  The attitude has been “that’s all bing done in China and India cheaper so why bother knowing it, we should focus on the advanced stuff.”

Then the shit hits the fan, a boat get stuck in the Suez canal, the Chinese lose control of a virus that restricts international trade, and all of a sudden we need domestic steel production and the only guys who know how to operate an oxygen blast furnace all have one foot in retirement.

The skills gap is real and it is hurting us more than we realize.

And just because this is a personal bugaboo of mine on a very similar topic of skils being replaced by technology: iron sights.

Optics are great and I’m glad they’ve become as cheap and plentiful as they have, but that’s no excuse not to be able to shoot irons.  Optics are glass, glass is more fragile than steel.  Batteries die.  Learn to shoot with iron sights.  If you can do that, you can always shoot optics, but being able to shoot optics doesn’t mean you can shoot irons.

Learn to shoot irons.

Spread the love

By J. Kb

19 thoughts on “The skills gap”
  1. This has hit the DOD as well. If you have the skills to reverse engineer or refurbish aviation hardware, there’s a lucrative opportunity.

    https://www.thedrive.com/the-war-zone/39537/the-air-force-needs-to-reverse-engineer-parts-of-its-own-stealth-bomber

    We had an almost identical discussion in my office since we do contract work for a bunch of different offices/companies. We all posited that this original work was done by some little sub-100 person operation that was a subcontractor of a subcontractor of a subcontractor to the prime contractor and one of two things happened:
    1) The DOD never got the full TDP, the full TDP was incomplete in a transition from paper drawings to digital drawings
    2) The people that worked at and ran the little shop that did this highly specialized work are all retired or dead now an no one at the program office thought about replacement parts or maintaining legacy fleets.

    We’re so fucked . . .

    1. Back around 2005, I did some work (at a couple of removes) for a small company that was reverse-engineering several electronic modules for an armored vehicle. Seems the supply of spares had run out, and either the Army didn’t have the original designs or some key components were no longer available (possibly both). So, back to the module-level functional specs, and a new design with current components.
      A few years later, I designed a replacement module for a big expensive piece of industrial equipment that was sidelined by the failure of a little board built around a no-longer-available analog math chip. My replacement, then, was built around a more recent analog math chip. Were I doing it today, I’d convert to digital, do the math in an MCU, and convert back to analog: much cheaper, and less temperature-sensitive.

  2. I can shoot with iron sights, the problem is my eyes can’t focus anymore, and trifocals don’t really do the job.

    1. I tried to go back and find where I commented on this, but I couldn’t.

      In my day job, I see a lot of manual processes get automated, and it follows this pattern:

      1. Identify a process that is repetitive, manual, and large in scale. If one person does this once a month its not a good candidate, but if 100 people do it all day every day folks start seeing dollar signs.
      2. Identify a way to replace the human with a program
      3. Pilot the replacement on a small scale while running the manual process in parallel
      4. Tweak the algorithm as you find issues.
      5. Roll it out on a larger scale, but don’t worry about what happens if the algorithm fails or needs to be changed due to external factors, because you can always fall back to the old manual process.
      6. Stop hiring people to replace the folks that do the job manually.
      7. Full rollout. Assign 90% of the remaining people to other duties (or lay them off, depending on skillset). The remaining 10% are retained to handle issues that may come up.
      8. Six months later, reassign (or lay off) the remaining 10% of people who understand the manual process because by this point almost all exceptions and outliers have been identified, and the algorithm can handle it without human intervention.
      9. Six months after that, total system meltdown because a variable changed that no one left at the company knew was going to change, even though it changes every year on a regular schedule. There’s no one left who knows the manual process and there’s no one who knows how to fix it. It takes 2-3 days to identify the problem, and you had to pull people from other parts of the company to do that investigation. If you’re lucky, it’s people who used to do that job manually, but usually you just have to take your really smart troubleshooters to play “one of these things is not like the others” and figure out what changed. While they are doing that, their normal (also important) work is on hold.

  3. I’m reminded of the assertion, some decades ago, that the USA could not go to the moon again even with 10 years of sustained effort.
    At the time I called “baloney” on that, given that it took 7 years the first time, so clearly with much of the knowledge already discovered and more advanced technology it can be done that fast or faster. Whether NASA could do it in that time, or any time, is an entirely different question.
    But you do raise a concern I missed. In the 1960s the engineers needed to do the work existed in the USA. Do they today?
    Even areas you might think are no problem might be. What about software? Sure, there are millions of programmers, but how few of them are capable of writing aerospace quality software? I’ve been a professional programmer for 47 years but I would not claim to be able to do that, and my code is probably more solid than that written by many of my “hack it till it stops crashing” counterparts. For example, the 737 Max episode shows clear that even companies that should have such people, like Boeing, in fact do not.

    1. Fifty years ago, car owner’s manuals explained how to adjust the engine valves. Now the manual cautions owners not to drink the fluid in the battery. Fifty years ago I could buy the parts to overhaul the fuel pump, water pump, oil pump, any part of a vehicle. Good luck actually repairing any system nowadays.

    2. Not only that, but the compliance paperwork (several flavors thereof) has gotten immensely more burdensome than it already was. Simply, on any project of scale, engineers spend more time doing paperwork than they do engineering. And the larger the project, the larger the burden becomes.

  4. Pkoning and I are of the same age. I’ve been programming professionally since I was 13. When I left University I went to babysit super computers on a research facility run by the US Army.

    I met my mentor. He went through University just a few years before I did. Our knowledge gap was huge! He had an understanding of electronics that I just didn’t (and still don’t but I have the skills to learn it if I needed). He could build a computer from transistors if he had to. I could not.

    I taught at University in a TA role. This meant that I had to translate the lectures into stuff the students could actual use. This meant that I taught bits and bytes and words, data structures, how we combine bits to make bytes and how we interpret bytes as characters or integers or unsigned integers or floating point numbers and all of the rest of it. What was memory and how do you address it? Yeah, lots and lots.

    10 years ago I was director of development for a “digital agency” fancy term for somebody that puts WordPress and Drupal websites together. In talking to the best of the employees we had, not a single one had ever taken a data structures course.

    Without google and stackoverflow they could not “develop”

    By the time I left University I had written three compilers, one assembler, and one operating system. The OS was almost a toy but it would have been able to handle any “real world” load given to it, but it only had to deal with a single input stream and a single output stream and had other weaknesses.

    My point in all of this is that even in the technical end of things, we just are not producing the people that actually KNOW how things work.

    I upgraded my OS yesterday. Part of my system isn’t currently working because the package developers haven’t bothered to make a package for the new OS release. I’m making the package. I’ve got a hundred pages of google results that all say “It’s to hard, I’m just going to downgrade python/OS until they come out with a package.”

  5. While DeGaulle’s quip “The cemeteries are full of indispensable men,” loss of corporate knowledge is real, serious, and increasingly prevalent in many areas of industry and life.

    1. Well, yes, but De Gaulle was a politician, and those are nearly all interchangeable and not missed when they disappear.

      A Dutch parliamentarian years ago came up with a nice observation which got him into serious trouble (because it was so true). He pointed out that politicians pretty much always vote the party line because they probably won’t get reelected if they show independent judgment. (That’s particularly true in parliamentary “party list” systems like Holland has.) So he referred to his fellow M.P.s as “voting cattle”. 🙂

  6. J.Kb – my latest build doesn’t even have iron sights.

    I’ve gone to the Dark Side

    New one: Red dot. Solar charging with battery (Holosun). And yes, a “pistol” with one of those eeeeeeevil arm braces. Probably soon to be a federal offense.

    And no, I will not comply.

  7. Meh. You learn what you need to learn and use. The problem with all these old-timey skills is that they are (almost) all perishable, particularly manual skills. Yeah, I’m old enough to use a stick shift — but I can’t buy a car with one. I used to be able to run a grader and forklift and backhoe and combine and bailer. Put me behind one now, and there’s a reasonable chance you’ll end up with a ten thousand dollar repair bill and I’ll end up underneath the machine in a ditch.

    Personally, I think it’s better to be very good at something that has immediate practical value than crappy at fifty things that I’ll never do. If there comes a time when these skills are *needed,* then someone will learn how to do it.

    1. I’ve lived my life believing that I should know how to do as many different things as possible to learn. Will I ever need all those skills? Probably not, but if I do, I don’t have to learn them, merely get better with practice.

      1. Yup. I learned Lathe for Dummies as a boy scout, then went on to Lathe 101 40 years later when I got a lathe of my own. And I still learn a new programming language every year or two (just picked up one more a month or two ago).
        Exercise is good, whether of the muscles or of the gray matter.

Comments are closed.