On 4/16/2015 8:32 AM, Peter A wrote:
Hi,
I would be interested in joining that Hangout. The situation
is a little bit of a mess right now. Besides authors not
getting acknowledgement there is the reproducibility issue.
Different implementations of the same algorithm can have very
different results. When I review a paper I try to remind
people to cite the software and the specific version. I know
of at least a couple of cases where performance claims were
invalid because they used software with a known "bug" in it.
I've been lazy and haven't bothered to publish papers for
most of my software. Used to get e-mails from people asking
how to cite it since there was no paper. I now think people
have learned how to cite webpages better.
For everyone's info: There is a 2014 book from CRC Press called
Implementing Reproducible Research [1]. There two chapters that
discuss the use of OS practices in the conduction of science
research, and they make points wrt reproducibility similar to those
Peter has made above. One of those chapters is by Millman and
Perez, and in keeping with many of their points they've made the
chapter available via github [2]. :)
My point with this reference is to support what Peter is saying WRT
reproducibility, but also to point out the other ways that open
source...both in the creation of scientific software and in research
practice/process...can/could/will/is benefiting the conduction of
scientific research.
My $0.03: I perceive that there is a great hunger in research
communities for the now clear benefits of using open source
practices: collaboration, transparency and the resulting
improvements in communication and education, technical innovation,
cross-org and cross-discipline information sharing, as well as the
rigor and scalability that comes from 'good practice/process' (e.g.
releng) and associated tooling...to mention but a few ;).
In other words: +1
Scott
[1] http://www.crcpress.com/product/isbn/9781466561595
[2] https://github.com/fperez/repro-chapter-oss
_______________________________________________
science-iwg mailing list
science-iwg@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/science-iwg
|