Just because nobody understands it today, it does not become, "A secret text." For all we know, it was a shopping, laundry, or honey-do list. Don't get excited. For inspiration, reread "A Canticle for Liebowitz."

I experienced a mercury lamp UV burn of the sclera (white) of one eye, neck and hand in an electro-optics lab, late one night while in graduate school. My eye protection was open on the side, where the inadequately shielded light source was. My setup - my fault. I woke up in great pain and unable to open the eye. It hurt a lot for a few days. The sharp demarcation edge of the skin burn, showing where my right shirt cuff ended, was visible for many years.

Warning: Pay the same attention to avoid eye damage from laser pointers, especially green, blue, and violet ones. They often leak very strong infrared light that the eye does not notice, but that can burn the retina.

My recent experience in the domain of probability and statistics for capacity planning:

• * The correct answer was in two 10+ year old items in stats.stackexchange.com.
• * Asked three times with slightly different wording, each time ChatGPT summarized the problem correctly but then provided an incorrect derivation and result, with a different result each time.

It is very feasible to do an analog optical multiplication of all points on a 2D surface concurrently. How?

1. Project (or mask) a UV image onto one surface of a cold thin sheet of photochromic glass. It will darken according to the image.
2. Project (or mask) a different UV image onto the other surface of a cold thin sheet of photochromic glass. It will darken according to the image.
3. To multiply the two images, shine a collimated beam of visible light through the glass. Alternativey, for a 2D, Fourier transform place the sheet at the focal point of a laser beam.

There are not a lot of developers who are coders but not programmers, in the sense that they write:

• * Programs whose algorithms they don't develop, such as transcoding existing programs from one language to another, or, in the 1950's, writing FORTRAN versions of known mathematical and scientific algorithms. AFIK, the programmers referred to their products as codes, but did not refer to themselves as coders.
• * HTML, CSS, and SQL DDL, which are code, but not programs with control logic, etc.

There are vastly more coders in physicians offices, clinics, nursing homes, and hospitals who figure out the standardized diagnosis and treatment codes to use on insurance claim forms, based on the patient's diagnosis and what the health care provider did.

Let's leave the coder job title to them.

Even before email in the enterprise, tickler systems were common in the insurance industry. They would remind the department to check for things that should have happened or were to be done now. These included getting reports from health exams for underwriting and rating, and from adjusters for claims processing.

These might be automated, but more often were index cards in a box. They were inserted in order of future due date.

Double-slit time diffraction at optical frequencies
Archiv: https://arxiv.org/pdf/2206.043...
Non-Academic article:
https://physicsworld.com/a/phy...

The Lenovo Product Specifications Reference many desktop and All-in-One models and variations with linux or Freedos preinstalled.

re: "It's just something that we've never had to deal with," said Elizabeth Donley

We dealt with skipping time on a scale of up to two weeks, all at once, not a mere skip second. That was when countries transitioned from the Julian to the Gregorian calendar.

• In Europe, Thursday, October 4, 1582 was followed by Friday, October 15, 1582.
• In the British Empire, Wednesday 2 September 1752 was followed by Thursday 14 September 1752.
• Russia switched after the revolution, January 31, 1918 was followed by February 14, 1918.
• Other countries made the transition at other times in between.

This is not a new problem. Professional Engineer examinations test certain topics, and not others. If the person's engineering skills are in different domains, then the person will not be able to pass the test, and so will not be able to work independently in the test's domains and so should not certify designs in those domains. By constraining use of the Engineer title, the profession limits its own domain of applicability.

I have a BS and MS in Electrical Engineering. In my entire education, I had exactly one lecture on single and three phase AC generators and motors. There was no way I could ever pass a Professional Engineer exam. On the other hand, people who did pass the PE Exam could not design computer, communications, command, and control circuits and systems. The PE Exam does not require and test in these areas.

PE licensing was created both to protect the public from shoddy work and to create a guild for the engineering profession.

Professional Engineers will not sign off on a design or construction with a faulty foundation. Why focus on provably correct software when it will run on provably untrustworthy equipment? People who are not Professional Engineers design, manufacture, specify, and deploy computers, phones, tablets that have no parity checking or ECC memory to be used in hospitals and other life critical situations.

Some electronic equipment and software comes with disclaimers stating that they are not intended, sold or authorized for use in life-critical purposes. Most doesn't bother. These can be designed and manufactured to a level of quality control inappropriate for life-critical usage. However, these disclaimers are widely ignored.

What software is really not life critical, depending on how it is used?

Example: I was in the hospital. My meal tray arrived, accompanied by a print out of what was on it, and nutritional information. If hospital staff had given me a medication dose calculated based on that information, I would have been dead in under an hour. There was a bug in the dietician's software. Prior to this event, that software had not been considered to be in the life-risk category, except for its allergen-ingredient avoidance component.

Math helped my programming. Learning Boolean logic and DeMorgan's Law, helped my programming.

Programming helped my math. I learned FORTRAN in 10th grade. In 11th grade, I wrote programs to do my math homework: Binomial expansions, numeric integration and differentiation, Newton Raphson Method, Plotting Equations ( such as circles and ellipses and anything else of the form 0=f(x,y) ) etc.

The pricing for A100 chips in the quote seems high. In the USA, an A100 SXM4 card with 80GB RAM seems to cost about $15,000. That is based on pricing at 4 for $60,000 at https://bizon-tech.com/bizon-x...

I had an on campus interview as I was finishing graduate school. They were interviewing for two different divisions of the company. A few weeks later I received two letters in the mail. The letter from one division said that they liked me, and I should fill out the enclosed job application. However, there was no job application enclosed. The other division wrote that they were not going to hire me, "Because we are hiring candidates more qualified than you." I think that by giving a definitive reason, they hoped not to be accused of racial, ethnic, gender, or other discrimination

I had worked for that same company for the summer between college and graduate school and had done very well there. I called people I had worked with. They told me that the truth was that they were not hiring anybody. There was a hiring freeze. The company only interviewed on-campus so as not to lose their slot in the on-campus interview schedule. They told me to go work for someone else, and that they would call me when hiring resumed. I did that. They called about 9 months later. I had a great career there for 33 years.

If so, many hard problems are suddenly solvable. NOT!
OK. Imagine I inform you that it has been proven that P=NP. Based on that recital of wonderful news, please, suddenly solve many hard problems.
Can't? Why not?
Announcing that P=NP does NOT assure that anyone will suddenly solve a problem quickly, whether finding an algorithm or completely executing the algorithm with particular inputs within a lifetime.

letter to journal article author:
re: COBOLing Together UI Benefits: How Delays in Fiscal Stabilizers Impact Aggregate Consumption
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3928948

Thank you for your insightful article.

You may find that how long it took to get UI systems able to handle higher demands depended mostly on whether the system was based on a vendor UI package or was custom created for that state.

You may find that major difficulties happened in states that crossed a specific monthly-new-claimant threshold, namely one requiring an additional decimal digit in the applicant sequence number for that month.

Question: Did any of the slower handling states cross a monthly new-applicant threshold from 99,999 to 100,000 or from 999,999 to 1,000,000?
Why asked: That transition to a more-digits monthly maximum forces changes to the format of preprinted & on-screen forms, reports, and beneficiary cards, as well as files and databases, and all systems which they feed.  This is because beneficiary ID numbers are often generated from the person's application's yy-mm-sequence#. From the article's map, the COBOL and antiquated states appear to mostly be the larger states, which may have experienced this maximum-digit transition.

Question: Which states use a state-unique UI system, rather than a vendor package?
Why asked: Antiquated states may be likely to have custom UI applications, rather than UI applications based on a vendor package. The custom applications are maintained in-house, often by hired-per-project contractors, who have a ramp-up period to get on board with a maintenance effort, no matter what the programming language.
Vendor packages have the advantage of having a team of people already on-call and familiar with its code. Also, some enhancements can apply to all states using it.

Hypothesis testing:

    Which states crossed the monthly registration quantity from n digit numbers to n+1 digit numbers?
    Which states use completely custom vs. vendor package based UI systems?
    What was the ramp-up time in months for systems to be adequately updated?

Methodology questions:

    Are COBOL and antiquated categorized separately or lumped together?
    Are the categorization and language labeling based in some cases on ~10 year old data? Was the categorization validated?
    Where is the time-series analysis, correlating time-to-upgrade the systems with time-to-resume spending, so as to discover whether these were longer for COBOL or antiquated systems?

Note that the online consumer access systems for filling out forms, submitting requests or querying status are almost NEVER written in COBOL, even if the core backend system is in COBOL and even if the system used in "terminal" mode by state employees is in COBOL. Crashing of these public-access front end web applications and their as-of-last-night online replica databases is due to their inadequate design and capacity, rather than the COBOL back-end.