If the South African government asked me to recommend whether to enforce a standard document format across all its departments, and if so, which standard to choose, I’d unhesitatingly respond, “yes,” and “Open Document Format“.

However, this decision would be based on functional requirements, such as whether the data would be accessible independent of particular software choices, whether the format is human-readable, whether it could integrate with other systems, and whether, in future, other software from other vendors might be able to read it.

The government reached the same decision, but some of its reasons are horribly misguided. The standards document the government produced (PDF) declares that standards for interoperability are to be preferred according to their “degree of openness”. The quality of being an “open standard” is then defined:

• It should be maintained by a non-commercial organization
• Participation in the ongoing development work is based on decisionmaking processes that are open to all interested parties.
• Open access: all may access committee documents, drafts and completed standards free of cost or for a negligible fee.
• It must be possible for everyone to copy, distribute and use the standard free of cost.
• The intellectual rights required to implement the standard (e.g. essential patent claims) are irrevocably available, without any royalties attached.
• There are no reservations regarding reuse of the standard.
• There are multiple implementations of the standard

This strikes me as a very badly misguided definition. Not even Richard Stallman, founder of the Free Software Foundation, believes software must be “free” as in “free beer” (though if it is, all the better, of course).

Yet the government’s specification is full of references to “non-commercial”, “free of cost” and “without any royalties”.

It is, of course, perfectly within its rights to specify a functional requirement for an open standard. But demanding that it is free, that it is maintained by a non-commercial organisation, that all intellectual property is given away for free, is going way beyond any reasonable, functional definition of “open standard”.

It excludes any supplier that provides software conforming to perfectly open, accessible, and functionally satisfactory standards, but does so for a fee. It might even end up excluding all those open source developers who spent their twenties toiling away for free in the vain hope that one day, they’d get to pay for their sports cars and luxury homes from support revenue.

Someone is going to get very, very badly burnt if this is how government thinks about software. Last time I checked, you evaluated benefits and weighed them up against costs, to arrive at a purchasing decision. This makes zero cost — at least for a significant subset of software — part of the required benefit. Granted, it is a neat way to ensure that all government software deals will, in future, have an infinitely high benefit to cost ratio by definition.

Microsoft has already lost the battle to have its own horrible (and horribly named) document format accepted as a standard. South Africa’s own standards bureau were among those that advised the International Standards Organisation not to ratify it.

But if Microsoft — or any other commercial software supplier — takes legal action for being unjustly excluded from government business because of this deluded document, they’d be fully justified, in my humble, faux-lawyerly opinion.

If you want to know how “scientific consensus” is cooked up, look no further than the speaker lists at climate change conferences. Actually, do look further: to who doesn’t get invited to such shindigs.

Last week, the American Statistical Association sponsored a workshop on climate change. The goal, according to David Marker, one of the organisers and facilitators, was to “delineate a statistical science perspective on understanding climate change and to develop a consensus statement on the areas of clear knowledge, as well as those areas in which great uncertainty remains”.

Of course, developing a “consensus statement” is easy if you neglect to invite people like Steve McIntyre and Ross McKitrick, who famously broke Michael Mann’s infamous hockey stick temperature chart (right). Showing Mann’s methodology to be so badly broken that it turns even random data into a hockey stick, is one of the more significant statistical contributions to climate research in the last decade. Considering that the question of how to use proxy data to reconstruct a historic temperature record, and whether this record does or doesn’t show warm and cool periods corresponding to solar activity, remains a hot topic of discussion within the ASA (PDF newsletter), this on its own would appear to qualify them for invitations to such a workshop.
McIntyre has also been swimming against the tide of obstructionism and even secrecy by government scientists to audit US temperature measurement stations. Though the focus of the research is the siting of these stations and how “bad data” gets “corrected”, a surprise discovery forced James Hansen, who heads the NASA Goddard Institute of Space Studies, to correct the benchmark temperature record which — like Mann’s Hockey Stick — the politicians on the UN’s Intergovernmental Panel on Climate Change use to scare us into give them more powers to tax and regulate. Again, the significance of McIntyre’s work on the statistical treatment of climate change data appears to be substantial. His comments on some of the topics raised at the ASA workshop show insight and familiarity with the issues.

This isn’t the first time that McIntyre and McKitrick have been omitted from speaker lists. Shortly after McIntyre’s discovery that NASA GISS had been cooking the books, a conference on integrity in scientific research likewise overlooked the Canadian, even though he’d raised red flags over the secrecy with which NASA GISS treats its data collection sites and its statistical methods.

Understanding climate change requires the combined skills of atmospheric scientists and statisticians, said Marker. The former understand the physical relationships being investigated, while the latter know how to determine which hypotheses are strongly supported and which are still subject to uncertainty.

And here I thought the consensus of the people that get invited to consensus-development workshops is that the science is settled.