[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Quote by Knuth

From: Christopher Dimech
Subject: Quote by Knuth
Date: Sun, 18 Jul 2021 15:17:43 +0200

> Sent: Sunday, July 18, 2021 at 11:50 PM
> From: "Jean Louis" <>
> To: "Christopher Dimech" <>
> Cc:, "Emanuel Berg" <>
> Subject: Re: Quote by Knuth
> * Christopher Dimech <> [2021-07-18 10:43]:
> > > Well, don't we do the very same mathematics as ancient Greeks did, only
> > > expressing it in a different language?  (And of course, we now know
> > > more, since our knowledge grows.  OTOH, many things in contemporary
> > > mathematics are not very trustworthy due to the complexity and high
> > > probability of errors.)
> >
> > I understand that.  About twenty years ago, I tried it.  It gets things
> > even more complicated than they are.  I rather have a number of small
> > well contained implementations.  I disagree with Knuth that a practitioner
> > of literate programming becomes an essayist, whose main concern is with
> > exposition and excellence of style, rather than how to perform the actual
> > computation.
> If you would be the inventor of such an excellent typesetting system
> such as TeX you would have that good idea that other people should
> learn and apply it as well.

Knuth is over-rated.  He did not invent anything.  Mathematical type-setting
had existed long before Knuth.  Today people have became obsessed with
typesetting, even though the most important thing is the information within
rather on how nice an alpha can be printed.

> By all means I do agree with Knuth, though in limited manner and
> specifically to context of the work or to specific domains or specific
> branches of the work.

Yes, in a very very limited manner.

> For majority of Emacs packages that is really not necessary as there
> are documentation strings or docstrings. That is literate
> enough. Print the Emacs package and read the docstrings. Or one can
> generate the list of functions in the package and format it it nicer
> and get somewhat nicer printout. But Emacs packages do not really
> control crucial or very responsible domains of human activities.

Correct.  But Knuth idea goes more than that because he was obsessed with
presentation and wanted science to be like a work of literature.  Without
understanding that most work is not work of literature.

> Who cares if some frame or window makes a problem, even if Emacs
> crashes there are remedies, if some highlighting is not correct,
> somebody will need to correct it but it does not impact large number
> of people.

Well said.

> Literate programming is highly necessary in crucial and high
> responsibility related applications. Let us say applications
> concerning handling of nuclear power plants, any other energy related
> applications, medical applications, satellite control, communication
> control. One simply SHOULD NOT program without good description on
> what that program does and how. Literate programming would also help
> the programmer easier to solve the problems before the program come to
> shape.

Nuclear power plants are simply engineering.  Things are never done as you think
even for nuclear and medical applications.  As long as things work relatively
well, the corporations and their engineers are happy.     There are enormous
risks which nobody cares about.  Otherwise nothing will get accomplished.

> That I personally just start the function because I write what I think
> is far from being understood in some future. It cannot be
> demonstratably easily understood on this mailing list even by
> experienced and more skilled people. Our thoughts do not align to
> other people thoughts and so our programs may look quite different
> from one to each other. Then future programmer may need to rewrite
> functions or improve upon it. In fact there should be a program that
> rewrites it in other new or more powerful programming language.

Yes, people got to understand well what you have said.

> Project has been programmed, it is finished and years pass, now come
> the new generation and that generation of people should be able to
> understand all details of the program in general. That understanding
> would come from literate programming.

Mostly they trash it than trying to understand.  That happened to me.  Simply
gave up and done things again as I understand them.

> In general, when there are simpler programs why complicate and make it
> too much literate? We can see that there is no reason for that in
> practice. One README or INSTALLATION file and docstrings and
> commentaries are obviously our practical way of literate
> programming.

Yes, many times being too literate means many will skip most of what is written.

> The demand for literate programming depends on the importance of the
> program.
> > I know a few professors myself claiming to work on the dynamic
> > properties of everything and the bullshit they say they have
> > developed.  They say they develop the theories, they develop the
> > computational algorithms needed, they do everything.  Until you do
> > some work with them and realise there's not much to their work.
> > Welcome to the world of academia in the western world.
> That is how it is, among those useless there will be number of useful
> inventions. I see that as a ratio and ratio has to be upheld. Maybe
> the ratio of useless invention is 80 to 20 of useful inventions, but
> if you do not uphold the ratio then you will get less useful
> inventions in future.

You just have to see the number of citations in academic journals. Almost 44% 
of all
published manuscripts are never cited. Today there are so may articles that I 
read anything.

In a time-frame of ten years, the top 1% researchers in physics and mathematics
had about 2073 citations.   If you have even 1 citation for a manuscript you are
already in the top 55.8%.  With 10 or more citations, your work is now in the 
24% of the most cited work worldwide; this increases to the top 1.8% as you 
100 or more citations.

Main take home message: the average citation per manuscript is clearly below 10!

I found an article by Scott Weingart, who gave very similar results, with 50% of
all published papers in the journal Scientometrics having fewer than 4 
70% fewer than 7.

Would I care about literate programming.  Hell no!

> --
> Jean
> Take action in Free Software Foundation campaigns:
> In support of Richard M. Stallman

reply via email to

[Prev in Thread] Current Thread [Next in Thread]