|
|
Posted by blmblm@myrealbox.com on 03/24/07 10:02
In article <rem-2007mar24-003@yahoo.com>,
Robert Maas, see http://tinyurl.com/uh3t <rem642b@yahoo.com> wrote:
> > From: Richard Heathfield <r...@see.sig.invalid>
> > > If you want to validate a textbook, you have a group of students
> > > actually try learning from it, you don't have a bunch of experts
> > > nitpick it to death.
> > That depends on what you're validating. If you're checking that
> > the book is comprehensible to students, then by all means use
> > students in your test suite.
>
> What I'm trying to check is whether it conveys information which
> when put into practice turns out to be correct, i.e. the reader's
> software using what my document suggests actually does what the
> student hoped it'd do which my document said it should do.
>
> > But if you're trying to ascertain whether the information you are
> > giving out is *right*, then you had better break out your six-pak
> > of independent experts after all.
>
> I don't have a six-pak of experts available on my call. I don't
> have billions of dollars to throw around hiring experts to test
> what I write.
[ snip ]
> > Students, by definition, do not have enough information to judge
> > whether a book is correct or not.
>
> If the book is merely stating facts about nature, you are correct.
> But if the book is saying how to achieve various data-processing
> tasks via various corresponding source code, any student with
> access to a decent C compiler has entirely enough resources to test
> anything they read in my CookBook/Matrix document.
Well .... A hypothetical student does have enough resources to
determine whether code you supply, or he/she writes using your tutorial
(1) compiles with a particular C compiler (whichever one he/she is
using) and (2) seems to give correct results for whatever test data
is supplied, by you or the student. Those are useful things to know,
and they are evidence that what's in your tutorial has merit.
But .... Compiling code with a single compiler and testing it
with a limited amount of data doesn't necessarily tell the whole
story, right? Does it imply that the code will compile with any
standard-conforming compiler? Does it imply that the code gives
correct results for other input data? Does it imply anything about
whether there are better solutions/approaches than what you provide?
These are the kinds of questions about which an expert's opinion
might be helpful.
> In addition a
> student using my CookBook/Matrix as a resource to tell them how to
> program various tasks is a lot more motivated to actually try what
> they read, because after all they had a desire to learn how to
> program a task, and then to actually program it, before they went
> to my CookBook/Matrix to read how to do it.
--
B. L. Massingill
ObDisclaimer: I don't speak for my employers; they return the favor.
Navigation:
[Reply to this message]
|