[okfn-discuss] why data and process in academic publishing matters

[Tony Bowden]

On 22 April 2013 11:52, Velichka Dimitrova <velichka.dimitrova at okfn.org> wrote:
> With regard to changing policy, it might be then "a coincidence" that
> Lagarde told Osborne in the same week that UK needs to re-think austerity. ...
> the issue they have researched is so close to peoples' hearts. Austerity and
> public sector cuts affect directly many people.

I in no way defend the UK austerity program[1]. But I don't believe
that it would have been different in any significant way had the R&R
paper never existed. The paper was conveniently latched onto by people
for whom it was useful (just as how some people have already latched
onto an excessively narrow reading of Herndon's paper to bolster
*their* view that actually the data shows that increased spending is
the way to go).

These things are complex and nuanced, and as much about execution
(both financially and psychologically) as about philosophy and
posturing. Fundamentally no-one knows what the right approach is,
largely because there is no guaranteed right answer. Turning *any*
economic position into a state religion is a very bad idea. Reality is
much too complex to fit any currently known economical model.

> I wouldn't necessarily conclude however as Marc has suggested that this
> would lead to less researchers sharing data. On a higher level journals and
> research institutions might also decide that it is even more important to
> reinforce data availability policies and subject authors to more scrutiny.
> It is a very embarrassing case, where less transparency is not a solution.

Presumably this case will also make it to the EuSpRIG "Horror Stories"
list soon, but read through the current list at
http://www.eusprig.org/horror-stories.htm should strike fear into
everyone — multi-million, or even billion dollar miscalculations in
Excel are remarkably commonplace.

I've tried using arguments around this in past FOI requests to get
actual spreadsheets out of government departments, rather than just
the figures — with mixed success rates. There are lots of tools
available to look for common types of errors (even the simplest of
which would have highlighted the R&R error[2]), but these aren't very
well known outside the spreadsheet community itself. It could be
interesting to run them against lots of publicly available
spreadsheets though...

Tony

[1] Which is an entirely different thing from being against austerity
more generally — after all I live in the country which has had the
most successful austerity policy of recent years. But just because it
was right _here_ doesn't make it right generally, or that other
countries would even have the ability to replicate the Estonian model.

[2] Well, to uncover the _spreadsheet_ errors that most people are
focussing on. The paper also had much deeper flaws than that, but
that's another issue.