I’ve written a bit about the fourth turning on this site. But aside from the political ramifications, there’s a human one as well. After a few generations, we forget not only what our forefathers learned through pain, but a whole lot of institutional knowledge is lost. It’s not passed down, or each new generation thinks they don’t need it or it’s stupid.
I read a piece sometime ago about an archeological dig either in Rome, or in some other part of their empire. Romans were pretty sophisticated for their time. They had a form of running water, as well as a way to heat the whole house. Of course, this was for the well-to-do, but a lot of their old ‘tech’ can be seen today, such as aqueducts. Anyhow, what they found was that in the main hall of some place they uncovered, they found evidence of campfires. Someone forgot how the heat worked.
You see this in the third world. For instance, in the film Empire of Dust, the main Chinese character is busting the interpreters balls about not maintaining what was left for them. The Belgians built and maintained the infrastructure. Surely they had the locals help with the work. No matter. No one learned, no one bothered. At one point he asked the dude about the railroad – “Is that railroad kaput?” it was. On an Anthony Bourdain show in the Congo, he showed the locals working like fiends to get a man cart running on the rails. It was about the only thing there to run. A later scene pictured a dude cutting a chunk of pig-iron with a cold chisel. Bourdain chirped about him being resourceful, hard working. I thought it stupid. I’m sure the local market had cheap hacksaws.
In that case, they didn’t invent and maintain the gear, but they had people that did. They couldn’t be bothered.
Closer to home, in IT, institutional knowledge is a thing. Everywhere I’ve been, there’s been a greybeard that knows all those obscure bits of hardware and software. Gearing up for Y2K, I had to do an inventory of gear to see what, if anything, would be affected. Most of my gear (installed before my time) didn’t have the clock set whatsoever. Made no difference to operations, It only made the logs useless. But the medical lab where I worked had an ancient PC running MS-DOS that had an app on it that tracked the time for the couriers (that picked up samples). No one was around any more that knew how that homegrown app worked. Hell, there were only a handful of us that could reinstall the OS, and configure the devices if we had to. They had to buy a whole new setup.
Even with the company I work now, I had an employee that I called a “space monkey”. He had a series of scripts to startup virtual machines used for training classes on AWS. Then stop them at the end if the class. He had no clue how they worked. He had no clue how each set of three per student were interconnected. They all needed updating and patching, which he did each time he lit them up. He had no idea how to capture the image, then modify the scripts. All he knew how to do was press the shortcuts (buttons), like a space monkey. The guy that created them was long gone, and didn’t care to help one bit.
After flubbing around for months, He leaned on one of the cloud engineers to help. Dude sized it up and told him to rebuild the machines and do it the right way. His quote was epic; “No one’s gonna break their skulls trying to figure that crap out”.
I can go on and on with examples. Where am I going with this?
Denniger had a great piece on AI again, like he does. In addition, pointed out:
The reason has nothing to do with whether AI is any good or not; it is just that you have no traceback on the decision process(es) nor did anyone in the firm make those decisions, so the technical debt in that application is 100% each and every time.
Having the AI code it up appears to be faster and cheaper. It only is up front; as soon as something goes wrong its 10x as expensive and slower than having a human write in the first place because you can’t stick your head into the programmer’s cube or office and have him say “oh yeah, here, that’s easy” and have him fix it within minutes.
Take this forum code. You get an error. I look in the log generated by it and know exactly what function generated the error. In all probability I also know why it happened and how to fix it. Five minutes later the issue has been diagnosed, repaired, the code recompiled and installed. If an AI generated the forum code in the first place but for whatever reason a human then has to fix it later you’re talking about hours, days, weeks or worse before said human can figure out what happened because no human has any clue as to how the code is laid out or why it was done that way in the first place.
This is where so-called “AI” fails and it always will because there is no transference to human consciousness. It is exactly the same thing as letting a high school kid use AI to answer his Algebra math test. He doesn’t know how to factor, reduce an equation or find roots.
We’ll get to a place like my spacemonkey. All they’ll know is how to press the buttons. It doesn’t work any more. Now what?
The younger guys will figure this out? I don’t think so. Like his last line above, these guys never learned the concepts. That was in a book by Clifford Stoll named “High Tech Heretic”. One of his points was that it was a waste to teach kids how to use a spreadsheet program. Teach them the math first, then they’ll figure out Excel pretty fast. That’s been a mantra of mine in the networking business – Learn the fundamentals and tech. The rest (configuring the machines) is data entry.
Going get weird.
Speaking of a tech detonation:
The abrupt departure follows days of melodrama over Huang’s admitted affairs — two have been acknowledged by Huang, so you know there are more out there — and the fact that his most recent paramour has all the appearances of being a ChiCom intelligence officer in addition to running a front company designed to bust U.S. restrictions on China’s access to Nvidia chips. This has cost Nvidia about 30% of its stock value, or about $1 trillion- yes, that was with a “t”- since the affair broke on October 29.
…Ms. Huang was an executive at Singapore-based Megaspeed, a major Nvidia customer that U.S. authorities have investigated for suspected diversion of advanced chips toward China. A recent New York Times investigation found that Megaspeed had no offices, facilities, or apparent employees despite accounting for between 10 and 15 percent of Nvidia’s revenue. It ran from leased space inside an accounting firm that specialized in managing shell corporations. – RedState
10-15% of revenue to the chi-coms.
Hoofa! That’s gonna leave a mark.
