Discussion:
The workforce cliff: What happens when the last IBM i engineer retires from government?
(too old to reply)
Subcommandante XDelta
2026-06-18 23:44:26 UTC
Permalink
From the IBM Mainframe World:

The workforce cliff: What happens when the last IBM i engineer retires from government?

Jennifer Nelson, 2026.06.15

https://www.route-fifty.com/people/2026/06/workforce-cliff-what-happens-when-last-ibm-i-engineer-retires-government/414184/

Many agencies still rely on mainframes, which are built on programming languages kept alive by a shrinking group of late-career specialists. This creates operational and cybersecurity risks.

:
bill
2026-06-19 01:06:11 UTC
Permalink
Post by Subcommandante XDelta
The workforce cliff: What happens when the last IBM i engineer retires from government?
Jennifer Nelson, 2026.06.15
https://www.route-fifty.com/people/2026/06/workforce-cliff-what-happens-when-last-ibm-i-engineer-retires-government/414184/
Many agencies still rely on mainframes, which are built on programming languages kept alive by a shrinking group of late-career specialists. This creates operational and cybersecurity risks.
I don't know how to break it to you but there are thousands of
experienced IBM Programmers working for the government I know
of just one agency that has over a thousand. And there are
still schools teaching with IBM Mainframes (like Marist in
Poughkeepsie, an IBM stronghold).

Or are we talking about another government?

bill
Scott Dorsey
2026-06-19 15:06:36 UTC
Permalink
Post by bill
I don't know how to break it to you but there are thousands of
experienced IBM Programmers working for the government I know
of just one agency that has over a thousand. And there are
still schools teaching with IBM Mainframes (like Marist in
Poughkeepsie, an IBM stronghold).
Yes, and they are all old. Many already are past their retirement age.
There are thousands of experienced OS guys out there... but are there
any of them in their twenties?
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
bill
2026-06-19 18:01:18 UTC
Permalink
Post by Scott Dorsey
Post by bill
I don't know how to break it to you but there are thousands of
experienced IBM Programmers working for the government I know
of just one agency that has over a thousand. And there are
still schools teaching with IBM Mainframes (like Marist in
Poughkeepsie, an IBM stronghold).
Yes, and they are all old.
Not as old as you might want to believe. The agency I am most
familiar with is constantly hiring new people. I seriously
doubt they are hiring people with one foot in the grave. As a
matter of fact, I suspect if I applied I would not make the cut.
Post by Scott Dorsey
Many already are past their retirement age.
There are thousands of experienced OS guys out there... but are there
any of them in their twenties?
Yes, unless you think only old people can get the jobs. IBM
Mainframes are very much alive. Schools still teach using
them. IBM has a very active educational program (a lot better
than VMS). And new people enter the field all the time.

We are still talking about IBM Mainframes and not COBOL, right. :-)

bill
Simon Clubley
2026-06-19 18:43:25 UTC
Permalink
Post by bill
We are still talking about IBM Mainframes and not COBOL, right. :-)
Anyone still using RPG in modern mainframes ?

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
bill
2026-06-20 00:12:06 UTC
Permalink
Post by Simon Clubley
Post by bill
We are still talking about IBM Mainframes and not COBOL, right. :-)
Anyone still using RPG in modern mainframes ?
I have heard of some legacy stuff but I know of nothing new
being developed. I think that would likely be on i-series
rather than z-series.

bill
Lawrence D’Oliveiro
2026-06-19 03:37:07 UTC
Permalink
Many agencies still rely on mainframes ...
Maybe not so many. And this will force it more quickly to zero, one
way or the other.
Scott Dorsey
2026-06-19 15:05:14 UTC
Permalink
Post by Subcommandante XDelta
The workforce cliff: What happens when the last IBM i engineer retires from government?
What we are seeing is that we have a lot of people, and that includes
scientists and engineers but also technicians and people involved in
day to day operations, who have been retired without any chance for them
to train new people up as replacements.

The DOGE-induced retirements have accelerated this process but it began
to be an issue well over a decade ago.

It is a matter of "what will we do when the last JCL expert retires?" but
it is also a matter of "what will we do when the last person that understands
the physics of tire skidding retires?" and "what will we do when the world's
greatest expert on tomato genetics retires and there's nobody to replace her?"

I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Simon Clubley
2026-06-19 18:37:07 UTC
Permalink
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?

[*] Willing to give them a pass on RS-485, but RS-232 ???

Simon.

PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?

PPS: On a lighter note, if you want to have some fun, introduce them
to the concept of bit-banging using Centronics interfaces or the control
lines on those RS232 interfaces... :-)
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Dan Cross
2026-06-19 18:59:03 UTC
Permalink
Post by Simon Clubley
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?
[*] Willing to give them a pass on RS-485, but RS-232 ???
Simon.
PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?
Well, was it actually RS-232 as in the +/-15V signaling stuff,
or was it UARTs and serial communications generally that they
needed ramping up on? People use UARTs all the time on modern
systems; using RS-232 specially is probably rarer.

- Dan C.
Simon Clubley
2026-06-19 19:20:24 UTC
Permalink
Post by Dan Cross
Post by Simon Clubley
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?
[*] Willing to give them a pass on RS-485, but RS-232 ???
Simon.
PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?
Well, was it actually RS-232 as in the +/-15V signaling stuff,
or was it UARTs and serial communications generally that they
needed ramping up on? People use UARTs all the time on modern
systems; using RS-232 specially is probably rarer.
On the modern systems I am aware of it's still proper RS-232 signalling
levels - that's why you need the MAX232 or MAX3232 to interface
with embedded stuff. I only tend to see the 3V3 or 5V stuff on embedded
boards themselves.

One interesting question however. I wonder if Scott is using RTS/CTS
for modem control or host-based flow control. There's a reason why CTS
was specified the way it is in the USB CDC ACM specifications even though
it's absolutely bloody useless for the normal case (at least these days)
of using those lines for host-based flow control.

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Waldek Hebisch
2026-06-20 01:59:58 UTC
Permalink
Post by Simon Clubley
Post by Dan Cross
Post by Simon Clubley
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?
[*] Willing to give them a pass on RS-485, but RS-232 ???
Simon.
PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?
Well, was it actually RS-232 as in the +/-15V signaling stuff,
or was it UARTs and serial communications generally that they
needed ramping up on? People use UARTs all the time on modern
systems; using RS-232 specially is probably rarer.
On the modern systems I am aware of it's still proper RS-232 signalling
levels - that's why you need the MAX232 or MAX3232 to interface
with embedded stuff. I only tend to see the 3V3 or 5V stuff on embedded
boards themselves.
One interesting question however. I wonder if Scott is using RTS/CTS
for modem control or host-based flow control. There's a reason why CTS
was specified the way it is in the USB CDC ACM specifications even though
it's absolutely bloody useless for the normal case (at least these days)
of using those lines for host-based flow control.
Hmm, I see no mention of CTS in USB CDC ACM specification. To me this
means that vendor of USB serial convertor can implement is in essentially
any way. One sensible way is to handle it in the convertor, without
involving the USB host.
--
Waldek Hebisch
Simon Clubley
2026-06-23 13:03:32 UTC
Permalink
Post by Waldek Hebisch
Post by Simon Clubley
One interesting question however. I wonder if Scott is using RTS/CTS
for modem control or host-based flow control. There's a reason why CTS
was specified the way it is in the USB CDC ACM specifications even though
it's absolutely bloody useless for the normal case (at least these days)
of using those lines for host-based flow control.
Hmm, I see no mention of CTS in USB CDC ACM specification. To me this
means that vendor of USB serial convertor can implement is in essentially
any way. One sensible way is to handle it in the convertor, without
involving the USB host.
That's the point - the host has no knowledge of the flow control
situation due to the specification so it MUST be done in the USB
serial converter which I find to be very annoying personally as
it makes the behaviour vendor specific instead of being defined
by a standard. You also have the problem of telling the USB serial
converter to ignore the flow control (if you need to).

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Dan Cross
2026-06-20 14:30:25 UTC
Permalink
Post by Simon Clubley
Post by Dan Cross
Post by Simon Clubley
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?
[*] Willing to give them a pass on RS-485, but RS-232 ???
Simon.
PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?
Well, was it actually RS-232 as in the +/-15V signaling stuff,
or was it UARTs and serial communications generally that they
needed ramping up on? People use UARTs all the time on modern
systems; using RS-232 specially is probably rarer.
On the modern systems I am aware of it's still proper RS-232 signalling
levels - that's why you need the MAX232 or MAX3232 to interface
with embedded stuff. I only tend to see the 3V3 or 5V stuff on embedded
boards themselves.
The UARTs are usually running at anywhere from 1V8 to 5V. The
level shifters to get to RS-232 signaling levels are a separate
part. UART<->USB adapters often avoid the RS-232 stuff and go
straight to discrete signal levels.
Post by Simon Clubley
One interesting question however. I wonder if Scott is using RTS/CTS
for modem control or host-based flow control. There's a reason why CTS
was specified the way it is in the USB CDC ACM specifications even though
it's absolutely bloody useless for the normal case (at least these days)
of using those lines for host-based flow control.
Many UARTs these days can be configured to do auto HW flow
control, so that you don't even have to take an interrupt for
it, which is kinda nice....

- Dan C.
Scott Dorsey
2026-06-19 19:17:05 UTC
Permalink
Post by Simon Clubley
Post by Scott Dorsey
I spent much of last week explaining RS-232 and RS-485 to interns, so there
may be some hope, but there are a lot of people to replace and not many
people being trained up to replace them.
Jesus!!! Seriously ? You now have to explain this stuff[*] ? How did you
get on ?
It was great fun. We got to watch bits going by on a scope and I got to
talk about how printers used to have a mechanical rotary switch (can't
call it a distributor anymore; kids have never seen a car with a distributor)
that was fired off by the start bit and mechanically disassembled the word
into individual bits.
Post by Simon Clubley
PS: For the RS-232 stuff, are they on physical motherboard ports, or on
USB based adapters ?
I did all the demonstrations with physical ports and I did warn them about
the pitfalls of USB-based adaptors often not having optimal voltages or
the ability to handle capacitive loads.

We didn't talk about handshaking. I did mention that there are various
handshaking options and that it's possible to send synchronous serial
data with an external clock but only in passing. At one point in the
early eighties it seemed like most of my life was spent dealing with
incompatible RS-232 handshaking standards.
Post by Simon Clubley
PPS: On a lighter note, if you want to have some fun, introduce them
to the concept of bit-banging using Centronics interfaces or the control
lines on those RS232 interfaces... :-)
I did mention that if you had a single GPIO port you could switch it
quickly enough to send serial data, but that timing was pretty critical
and it was best left to proper hardware.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
bill
2026-06-20 00:19:19 UTC
Permalink
Post by Scott Dorsey
I did mention that if you had a single GPIO port you could switch it
quickly enough to send serial data, but that timing was pretty critical
and it was best left to proper hardware.
Bit Banger on the Tandy Color Computer. Easily does 115K.

bill
Simon Clubley
2026-06-23 12:54:58 UTC
Permalink
Post by bill
Post by Scott Dorsey
I did mention that if you had a single GPIO port you could switch it
quickly enough to send serial data, but that timing was pretty critical
and it was best left to proper hardware.
Bit Banger on the Tandy Color Computer. Easily does 115K.
I've implemented Tx-only serial ports in software on microcontrollers
before to give me an extra serial port to output debugging information.
Worked very well as I remember.

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Simon Clubley
2026-06-19 18:39:24 UTC
Permalink
Post by Subcommandante XDelta
The workforce cliff: What happens when the last IBM i engineer retires from government?
Jennifer Nelson, 2026.06.15
https://www.route-fifty.com/people/2026/06/workforce-cliff-what-happens-when-last-ibm-i-engineer-retires-government/414184/
Many agencies still rely on mainframes, which are built on programming languages kept alive by a shrinking group of late-career specialists. This creates operational and cybersecurity risks.
If they are worried about that stuff for IBM, imagine how they probably
feel about VMS...

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
bill
2026-06-20 00:15:40 UTC
Permalink
Post by Simon Clubley
Post by Subcommandante XDelta
The workforce cliff: What happens when the last IBM i engineer retires from government?
Jennifer Nelson, 2026.06.15
https://www.route-fifty.com/people/2026/06/workforce-cliff-what-happens-when-last-ibm-i-engineer-retires-government/414184/
Many agencies still rely on mainframes, which are built on programming languages kept alive by a shrinking group of late-career specialists. This creates operational and cybersecurity risks.
If they are worried about that stuff for IBM, imagine how they probably
feel about VMS...
DiSA stopped approving VMS systems during my lat mobilization.
That was 2009. I wanted to update the STIGS but was told not
to bother.

bill
Loading...