On Nov 4, 10:31 am, David Kirkby <david.kir...@onetel.net> wrote:
> On 4 November 2010 17:14, Kwankyu Lee <ekwan...@gmail.com> wrote:
>
> > Hi Minh,
>
> > Thank you for the link to the article of J. Neubuser. I think the
> > article should be displayed in a holy place at the Sage website, as it
> > seems to me the original source of the philosophy on which all this
> > Sage project is founded.
>
> > Kwankyu
>
> Yes, it is a pretty important point that J. Neubuser raises.
>
> I must admit I was rather hoping there would be more similar
> references, ideally in peer reviewed journals, or books from decent
> publishers.
>
> Perhaps Richard Fatemen has written on this subject in a peer reviewed
> journal. I know he has similar concerns, but posts on
> sci.math.symoblic or Richard's own web site don't count for much.
>
> Dave
You misread my concerns. I basically view the paper by Demillo,
Lipton, and
Perlis on mathematical proofs (in Comm. of ACM many years ago) as
a fairly convincing explanation of the SOCIAL nature of proofs.
There are several questions here:
1. can you prove a program correct without looking at its source code?
answer: yes, sometimes.
2. can you prove a program correct by looking at its source code?
answer: practically speaking, almost never.
3. can you publish a paper with an incorrect proof?
answer: apparently fairly often.
4. is the possibility of having source code in an open repository a
promise of correctness?
answer: certainly not.
5. must all programs referred to in a journal paper be open?
answer: obviously not, otherwise papers relying on (say) Fortran
or written in Matlab would need to "open" matlab.
In conclusion, Stein and Joyner are just banging on a tin cup trying
to get people to drop a coin in it.
What is the obligation of a journal author? He/she should provide
documentation of some sort that advances the state of knowledge
in some way. The author(s) take primary responsibility for statements
in the paper.
What is the obligation of a referee?
1. To inform the author of problems with the paper, e.g.
a. known result
b. false, incomplete, erroneous somehow.
c. problems with exposition
2. To recommend publication or revision or not, to the
editor.
If, in the judgment of the referee, the paper has a problem with
exposition, perhaps on the grounds that the referee feels there
was an questionable use of a computer, then ...
If a paper says, here is a way of computing the 43rd moment of
the Whatasmatta-Yhu blenyfirth, and Mathematica gives the
result to 100 decimal places as xyxyxyxyx... , and confirmed
by Maple ... a referee who
rejects that paper on the grounds that neither is open source
is on thin ice indeed.
Even though I have exhibited identical bugs in both systems,
from time to time.
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
