Sometimes it’s Tuttle, sometimes it’s Buttle– and sometimes there’s an Interpol mauve alert for ZUTTLE, no_photo.jpg, DOB ____-__-__, nationality Pick_One, and the default “X” for all the checkboxes:
armed at all times
immediate danger
disguises
drugs (all, everything)
poisons
immuntity to poisons
violent fugue states
martial arts
…with focus on maiming and mutilation
improvised weapons
pursuit with parkour
pursuit with carjacking
Romani
computer viruses
psychopath
Anonymous
dead aim
occult
organ trafficking
cannibal

Programming is increasingly a young man’s job. At 41, my options are narrowing, not expanding even with my skillset. Maybe it is time to get the band back together and go into rock. Bowie’s last album is awesome.

Another thing that there’s no room for in that model: learning new languages/frameworks/platforms/etc. If you’re an employee at one of those frappunicoed up, nose-to-the-grindstone companies, you don’t get to decide to dick around for an hour or two a day, every day for a month, writing a hello world and then a toy app using some new technology, so you’ll eventually get the hang of it, and it may or may not be useful in the future– and you never can tell, until the future suddenly happens.

Or: maybe one guy goes and dicks around with some new framework, mulls it for a while, and then can tell other people it’s useful to learn, and I can help out, etc. And if that technology explodes, he’s the difference between a shop where NOBODY knows the new system, to where SOMEBODY knows the new system.

Or: somebody plays around with some smartphone from six years ago that nobody bought, but discovers that it has some great feature that maybe you could copy.

So you’re completely correct: “You hire some young, energetic people. Make sure that they can pass technology skills tests”. In other words, they’re walking in the door already having invested serious time (and not on your nickel) into learning whatever you think your company wants to do right now.

…But say the technology evaporates– if you’re running the company, you basically have to scream “everyone learn this new thing!! so you can quickly write impressive and stable pieces of software by a month from now!!– all while finishing our existing contracts!”. That’s very un-agile, so you just reboot the company (maybe or maybe not a new name) with a new crop of already-know-it workers, but the programmers don’t know eachother, or you, from Adam, and productivity will be extremely unimpressive, compared to if you had let the old crowd dick around a few hours a week and pick up a few new platforms/languages/etc.

Here’s an opinion that’s sort of tangential: Rands once said: if you have some guy there whose job has turned out to sit on one piece of COBOL that happens to be a bottleneck for the whole infrastructure and that’s all he knows, and has a DMV-like attitude toward, well, everything, still pulling a paycheck and still necessary, well, it’s the fault of you, Rands-advised manager, for not having done what you should now do: have him go take classes in new stuff, especially since the system in which the COBOL-O-TRON is crucial might get totally replaced at any time and with little warning…
But, with some quite motivated training, he might get to be guy who understands the new system better than anyone else.
(Conversely, losing him might be a bad idea, because, as you find out only once he’s gone, it so happens that he, without thinking much about it, is the only employee who knows the crucial order that the Big Servers have to be booted in, after a power outage, even though it theoretically shouldn’t make any difference. Or: he’s the only person in the company who could write at above a 5th-grade level!)

I don’t think Rands had advice about what to send this guy off to go learn. Maybe Rands just ment: wing it.

I think we also have blurry lines between linux system administrators and IT. In IT there is one desktop-support guy, and the rest are dealing with infrastructure. Mostly it’s stuff like ensuring the LANs are correctly partitioned (we send around a lot of video), dealing with internet and satellite providers, dealing with building management over cooling and power circuits, and so on. It’s sort of a holistic-everything-adminstrator. Not so much “Linux” system administration beyond DHCP, NTP, and a giant RAID with minimal permission enforcement visible over SMB. A lot of what was Linux-specific admin stuff is now outsourced — Google for email, a web guy for the corporate website (which doesn’t need lots of low-level Linux fine-tuning; it’s just an off-the-shelf CMS with a custom skin on a virtual private server).

Based on talking to engineers and support people at Google and Amazon, I’m guessing that a lot of the Linux admin work is migrating toward companies providing servers as a service: cloud stuff like Amazon, Linode, and Rackspace; everything Google does; even Microsoft’s Azure has Linux servers now. Those companies worry about a lot of the messy details so you don’t have to — just write your Ruby app and deploy.

Things always seem down after “getting dumped.” You’re a smart guy with lots of experience. I expect this will be behind you soon and, with the benefit of time and hindsight, will seem like only a bump in the road.