The biggest major subtractive policy change I experienced was as an undergraduate almost 50 years ago. There had been a drop/add period at the beginning of each semester during which students could change what courses they were taking. After that, until midsemester, they could drop, but not add any more. There was so much shuffle and paperwork that the university stopped asking for registration at the very beginning of the semester. It subtracted the preregistration process. It said, as the end of drop/add approaches, tell us what you are really taking.

Nowadays, this is called the "Shopping Period." See
https://poorvucenter.yale.edu/ShoppingPeriod and
http://catalog.yale.edu/handbook-instructors-undergraduates-yale-college/courses/-selection-period/
I hope it was a pre-April Fool's day announcement that this system is being terminated. See
https://yaledailynews.com/blog/2021/03/10/after-years-of-planning-but-ahead-of-schedule-yale-college-scraps-traditional-shopping-period/

My estimates based on recall are very different:
           1985    1985    2020
           High    PC      PC
Memory MB  $15K    $1K     $0.01
Disk   MB  $40     $10     $0.00003
CPU    MIP $1M     $5K     $1
Note that:
"High" means high end mainframe or Tandem server.
The CPU MIP price includes the entire main frame of the HIGH mainframe or the PC chassis with display adapter, small disk, and I/O ports

The First Sale Doctrine does not apply when the parts have not been sold.
It is not uncommon to have a fancy product where some parts are present even though they have been neither purchased nor licensed. In IBM z, p, and x (now Lenovo) servers, these features are activated by purchasing "Upgrade on Demand" activation keys. In most cases, these are not just licensed software and microcode capabilities, but also extra processors and memory. Sometimes, the customer pays for the privilege of having them present and ready for activation without taking the machine offline.
These activations can be permanent, but some can be "rented" by the day or even hour, activated manually or automatically for unplanned peak periods, without requiring a machine, hypervisor or operating system restart.
Unless activated permanently, these parts remain IBM assets. The contracts probably address who is responsible in case of loss or damage to them by theft, malicious behavior, or natural disaster.

My OnePlus One (3GB/64GB) is five and a half years old. I charge it every night, more if it gets much use. I hope it will last until I can get three new features at a reasonable price:
* 5G
* WiFi 6
* High Accuracy GPS

"Refuses to say" is innuendo and a cheap journalistic attack when the answer is unknown by the one questioned. A valid truthful answer to "How many employees are currently ill with disease x?" requires information gathering which is prohibited by HIPAA, even if the information were available, which it is not. Nobody has test results from the employee population up to the current minute. As a result, a valid answer is impossible.
I am especially sensitive to this issue because I was once the one referred to in a trade press article as the person who, "Refused to say." At a presentation on a new release of a product to a user group, I said that one component had been totally rewritten. I said that any time some code is rewritten, new defects are introduced that hadn't been there before. I urged the audience when conducting regression testing of this release to "beat the hell out of" that component. (Regression testing is the part of testing to make sure that what used to work still works just as it used to, producing the same results.)
The reporter wrote in his article that the IBM representative said that there were new defects in the rewritten File Control component of CICS 1.7, but that he "refused to say what they were."
(By the way, I didn't get into any trouble within the company. The people in the product's support center and development lab agreed with me.)

Oracle brought its first SQL based product to market in 1979, before IBM's SQL/DS in 1981. Before the product, IBM and its employees had published research papers on the relational model and the SQL syntax. Copyright prohibited others from plagiarizing the text of the articles and presentations, Nothing prohibited others from implementing in code what the articles described in print.

Was the curriculum up to date? The curriculum was timeless. You could get credit for Computer Science, Systems Analysis, English, and French, but not FORTRAN, APL, LISP, or ALGOL.
Were you taught the right things? YES. Analytical methods, black box testing, analog and digital signal processing, statistics...
Was academia open to new ideas and new ways of doing things? Yes.
Did your education prepare you well for real life tech work in a non-academic environment? Undergraduate: No. Graduate School: Yes.
The point is that there is a difference between an academic education and trade school training. With an adequate academic education, the student can acquire skills that allow for applying and growing fundamental principles to the specific technologies as they come and go.

Shutting down WWV* in the near term is bad.
Recommending shutting down WWV* in the near term is brilliant.
Shutting down WWV* in the long term is a reasonable possibility to consider.

Shutting down WWV* in the near term is bad.
Many things depend on them, without having time, plan, or anything leading to reducing dependency on them, by switching to the available alternatives.

Recommending shutting down WWV* in the near term is brilliant.
NIST and WWV* have very little in the way of name recognition let alone appreciation. Does the packaging on products that rely on WWV* mention WWV or NIST? Usually they call themselves "Atomic clocks" or say that they rely on radio signals, without saying from who or where.
The controversy raised by the recommendation will serve to raise awareness of the value provided by NIST to the USA and the world.

Shutting down WWV* in the long term is a reasonable possibility to consider.
There is an alternative radio time signal. Using signals from several satellites, GPS receivers figure out the time to within 100 ns. I believe this is more accurate than WWV* receivers can accomplish, due to propagation variations and not knowing where they are. Less expensive and lower power receivers could get just time at accuracy a little better WWV* receivers.

Who pays for replacing devices? In the consumer space, countries handle this for analog TV, and in some countries, AM and FM radio.
In both consumer and commercial space, vendors often drop support for older technologies: New applications that don't run on older hardware. Phones that don't get OS upgrades. Software that doesn't get security updates, etc.

What about technology that many users do not pay for, or at least not directly? This starts with non-toll roads funded by taxes & registration fees and extends through DNS to search engines funded by advertisements.

What obliges the provider of that technology to continue to do so at its own expense or when its funding runs out?

Is it reasonable for the transmitter of a streaming broadcast to encrypt it and charge a fee for the decryption key? Satellite and cable TV operators operate that way. Non-military GPS could become a fee service.

I did this in 1986 to the predecessor to the M, the keyboard for the PC/AT, with the 10 function keys on the left:

The keycaps were two parts. The upper parts were interchangeable with keycaps for the newer 3270 terminals. Every 3179 terminal came with a box containing an alternative set of keycaps with both normal and APL symbols. I salvaged a set from the trash and installed them on my PC/AT's keyboard.

Fortunately, with only two or three exceptions, the placement of the APL symbols for APL\PC matched what was on the keycaps. I took care of those inconsistencies with a fine point permanent marker.
Viola, no stickers needed.

How could 40 year old code be written for a processor that was announced 32 years ago? https://en.wikipedia.org/wiki/Intel_80386

The subject line was edited to fit length constraints.
The intended subject line: IMHO: IBM Management has been very dubious of the ethics, legality, and safety of unpaid labor, whether called internship or apprenticeship. From IBM's perspective, TANSTAAFL.

DISCLOSURE: I am no longer an IBMer. This post is opinion, not based in current knowledge.

IBM operates in many countries, with varying labor laws, including minimum wage laws. Managers, until they get a go-ahead from legal, are not going to pay less than minimum wage. Where terminating an employee is restricted and subject to messy reporting and regulations, the company has to weigh effort vs. value. Also, IBM values what comes along with a documented employer-employee relationship, namely the accompanying assignment of intellectual capital rights and work product to the employer, non-disclosure obligations, etc.

Don't demean what it means for IBM to embrace internships and apprenticeships. Assuredly there was a lot of work on the part of HR and legal at the global level and in each country to do it in ways that are legal and safe for the company, and then to communicate to departmental managers and HR staff exactly how to do it.

Disclosure: Over 40 years ago, I was a "summer student" at IBM. We were paid better than minimum wage. The expectation was that the company did not expect a lot of value from the students' work, other than the ability to observe a potential future employee, how the student learned, worked, and fit in with the culture. There was a lot of evangelizing by the employees to the students about how terrific IBM was as a company and an employer. It worked for me. After grad school, I came back for another 33 years.

One full page newspaper ad from the week the IBM PC was announced, called it "The IBM of Personal Computers." The "B" in IBM is for "Business." The IBM PC was a business machine, not a toy. What distinguished it from Apple and Radio Shack computers? It was the first PC to provide parity memory.

The IBM PC had parity memory, not ECC. On a parity error, BIOS displayed an error message on the top line of the display and stopped the computer (HALT or disabled interrupts then loop, such that the error message remained visible.) If I recall correctly, restart required the BRS (Big Red Switch) to power down and then up, rather than <alt><ctrl><del>. When the computer was powered up again, as always back then, BIOS did a Power On Self Test that among other things, tested memory.

The parity memory meant that the IBM PC produced results as programmed, or not at all.

Some customers did not really care. They preferred results, even if incorrect ones. Clone companies started providing a BIOS setup option to disable the parity checking. Next they started saving 11% on memory costs by not including parity memory bits at all. Customers did not care. The IT trade press and personal computing trade press did not raise a ruckus, and almost never included the presence or absence of parity memory in product reviews.

Then IBM had started to become "Market Driven." The market did not value it, so eventually, IBM dropped parity memory from end-user PC's, but retained ECC memory for servers. At an internal IBM technical conference, I asked IBM Executives whether anything had changed that made parity memory unnecessary in PC's. The heads of both the PC Division and IBM Microelectronics (IBM made its own memory back then) agreed that dropping parity memory was the wrong thing to do, from both technical and validity-of-results perspectives, but was something the market did not value.

The follow-on was for the industry to create premium "workstation" class end-user computers with ECC. Unfortunately, as far as I know, no company offers laptops, notebooks, tablets, smartphones, or IoT devices with ECC.

It's time for laptops, notebooks, phones, desktops, and anything else that needs to be trusted or taken seriously to switch to ECC RAM. As memories get cheaper, larger in total and smaller in geometry, and especially when people use devices at 35,000 feet, there is no excuse for everything (executable code and the content) to be untrustworthy.