Geoscience Reference
In-Depth Information
the software is noting command-based equivalents that reproduce these actions. For example,
Minitab, Inc. (2011) has a facility where menu commands are echoed as standard Minitab com-
mand-line entries as they are carried out. SAS/ASSIST software has a similar facility.
Metadata, as considered here, refers to the incorporation of background information with items
that are cut and pasted. This information may not be directly visible in the target document, but
could be extracted at some point. The suggestion is that the journalled commands used to produce
a given figure, map or table could be attached as metadata when cutting and pasting occurs.
Thus, the commands used to produce items in the final document would be embedded in that
document. These could then be optionally viewed, extracted or modified by a third party recipient
of the document.
At the time of writing, this section is speculative. Whereas the previous sections point to meth-
ods and software that already exist, the suggestion given earlier would require modifications to
existing software that as yet have not been implemented. There may be a number of technical dif-
ficulties to address here - for example, although some packages already offer journalling facilities,
they do not necessarily identify which commands were used to create which cut and pasted object
and in which order the commands were applied. Secondly, a protocol for the transfer of this kind
of metadata needs to be devised. Thirdly, in order that the methods used to obtain the results can
be understood and subject to peer scrutiny - a fundamental justification of reproducible research -
these automatically collated sets of commands must also be understandable by humans.
17.5 ACHIEVING REPRODUCIBILITY IN GEOCOMPUTATION
There is currently an active lobby for reproducible research reflected in many recent articles in key
journals such as
Science
and
Nature
, for example, Morin et al. (2012) argue that
The publication and open exchange of knowledge and material form the backbone of scientific
progress…
and furthermore that such practice should be regarded as
… obligatory for publicly funded research.
Ince et al. (2012) make similar points, in particular arguing that
Keeping code closed ensures that potential uncertainties or errors in a paper's conclusions cannot be
traced to ambiguity, numerical implementation, or machine architecture issues and prevents testing of
indirect reproducibility.
These ideas are showing signs of becoming
mainstream
- and their logic is certainly persuasive -
linking also with current notions of
open government
and open access to information. Related to
this are statements made by research funding councils in the United Kingdom - such as this from
the Engineering and Physical Sciences Research Council (EPSRC):
EPSRC-funded research data is a public good produced in the public interest and should be made freely
and openly available with as few restrictions as possible in a timely and responsible manner.
Engineering and Physical Sciences Research Council (2012).
17.5.1 c
ounter
a
rguMentS
Although the previous considerations provide a strong case for adopting reproducible research, it
would be unrealistic to assume that there are no objections. A number of these relate to the need to
