Well I would somewhat agree with your sentiments. But in my experience its the professor who may be either lacking the appropriate set of skills
and/or motivations.

From the point of view of the professor,

1) The point of the work is to create another notch on the publication bedpost. As one professor once told me, “Publish in haste, regret errors at leisure”.One rarely gets paid to clarify code, making it easier to read, use, or understand.

2) Having unreadable code is advantageous, because it makes it harder for others to check your work, which again is good for the publication rate. Indeed, beyond the evolution of the code itself, once enough time passes and the code has a
reputation and is well known in the community, NOW it can be used for situations completely unintended, because no one remembers the dirty details.

Example, in my own experience, a certain well known group at a very prestigious American
university would use a code over an over again to compute dielectric response functions for
solids with non-cubic structure, even though the math in the code was only right for simple cubic.
There are many other examples.

Sometimes the person being hired may want to bring in coding styles or procedures that are verly: sophisticated and really only obscure the underlying science and math. That happens too, although rarely.

I seem too remember a Physics Today article, which analogized scientific code development throughout the years to the development of suspension bridges. According to the article, we have come through the Brooklyn Bridge phase (hard, tight, fast FORTRAN
very well written by labs in the 70s for integrating diff eqs or high accuracy calculations of
special functions). Now we are in the Tacoma Narrows Bridge phase. Lightweight, loose, and everything hanging by a thread, whether it be FORTRAN2 or a mixture of m4/bash scripts/perl/python/siloon/C++/MPI.

Finally may just as well ask, is the professor ready to take a qualified candidate
out on a retreat for 6 months, the way Feynman did, and rigorously train them in diagrammatic
methods, running the hand calculations over and over and over again, as Schwinger suggested, or is he
too busy creating another non-rigorous derivation-by-hand-waving correlation function named
by an acronym containing the first letter of his last name?

To sum up, I would say BOTH the professor and the student should be “required to be masters of the basic tools, and proficient in a broad set of specialized and modern tools. It is inexcusable not to be.”

Thanks,
John

BTW: There is a slight error your example. It should be TOhtml. Your example has a capital-H. Must be a “normal”-h.

In humans the lethal dose of Arsenic is 1mg/kg/day. So one one-thousandth of a gram per kilogram of body weight in one day is enough to kill a person. But you don’t think that an organism which can say â??screw that, Iâ??m using Arsenic, itâ??s good enough for meâ? is revolutionary?

It is revolutionary because it just expanded the notions we have of where we can possibly find life on other planets.
Sorry it wasn’t intelligent life on another planet for but it is still absolutely astounding.