Please reflect on the following generalization: there are too may PowerPoint-type presentations at conferences and trade shows that are made by vendors with a purely commercial objective. The implication is that the world would be a better place if there were more presentations by oil and gas operators—they, after all, are the ones that hold the keys to the ‘truth.’
Well that may be so in some circumstances. But what about the following case, an oil company uses Schlumberger’s Eclipse reservoir simulator in a novel way that merits a paper. What is the relative contribution to the work from the operator and from Eclipse’s developers? While that is a reasonable subject for debate, I can see no circumstances where the contribution from Eclipse is zero. And I have a hard time understanding why, even if the contribution from the software is minimal, this entitles the author to regard Eclipse as a ‘commodity’ that is unworthy of a mention. In general, the likelihood is that the contribution from the software’s developers exceeds that of the authors.
Why have things got to this state? On the one hand a desire to avoid commercialism—some conference organizers have fairly arbitrary rules along the lines of ‘no product names.’ But another reason that such rules are respected is, I am afraid, a degree of vanity of the company man whose gut feeling is that he (or she) does the science or the engineering and that the commercial world simply provides ‘commodity’ software. I say this advisedly because, when I was a company man myself, I presented a paper at the SEG that was hugely dependent on the efforts of our seismic acquisition contractor. Even though I did give credit, I was a little surprised when the contractor failed to express great enthusiasm for my rather derivative oeuvre.
In recent years when I attend such talks, I usually try innocently to elicit some information in the Q&A as to what tools were used in the project. This mojo does not seem to be working so good. Recently I have been fobbed off with a response along the lines of, ‘What software used is not really germane to the debate (read ‘it is commodity’) and we do not want to be seen endorsing one product over another.’ Endorse? Who said anything about endorsement? I would be happy with ‘we used XYZ—but actually it sucks.’
There are many reasons for fessing-up as to what tools are used for the job in that this may be a significant factor in the research. There may be a bug in the software that is revealed at a later date. Its use may be inappropriate for the task in hand. Or it may be the best of breed. Any of which may be good to know if you are thinking about trying to do similar projects yourself.
I stayed on after the Copenhagen EAGE this month (report in in next month’s edition) for the workshop on open source software in geosciences. In her keynote, on the central role of geophysics in the reproducible research movement, Columbia University’s Victoria Stodden referred to a ‘credibility crisis’ in computational science. Stodden paid tribute to John Clarebout who observed that, ‘An article about computational science is not the scholarship itself. It is just an advert for the scholarship.’
I pricked up my ears when I heard this because it seemed to capture the sentiments I expressed above. By not disclosing what tools were used for a project, a paper is effectively reduced to a commercial. Link that in with the large booths on the exhibition floor that hire ‘talent,’ and you understand how the scientific conference is in danger of being subverted. Not just to advertise vendors’ products but to promote oils’ research and operational ‘excellence.’
So what does the reproducibility movement have to say? On the positive side, Stodden suggests that computation may represent a new ‘third branch’ of the scientific method—alongside deductive reasoning and empirical techniques. On the other hand, the ‘third branch’ tends to produce less than perfect results—what has been called the ‘ubiquity of error’ in software.
A lot of loose thinking goes on in the computational world which tends to produce ‘breezy demos’ (we’ve seen a few of these!) rather than reliable knowledge. Some have even claimed that ‘most published research findings are false.’ For Stodden and other open source hotheads, including representatives from BP, ConocoPhillips, Saudi Aramco, BG and more, are pushing for ‘reproducible research’—papers that bundle data, code and text so that they can be properly evaluated and their results used to further the greater scientific good. Stodden herself is promoting the reproducible research standard (RRS) to achieve this.
Of course there are other issues such as protecting intellectual property, copyright and so on. Mostly these are somewhat orthogonal to the publication issue. If a work is patented, then the IP is in the public domain already! In any event, if you want to keep stuff secret, you should not be presenting conference papers.
But this is often what happens as a presentation becomes a PowerPoint version of the dance of the seven veils—teasing its way around the interesting stuff while revealing little flesh. My own default position for analysing these is that what is not disclosed is more likely to fall into the category of ‘ubiquitous error’ than to represent a methodology so clever that ‘if I told you I’d have to shoot you.’
Stodden cited mathematician and aphorist Richard Hamming who in his 1968 Turing Award lecture observed that while Newton said, ‘If I have seen a little farther than others, it is because I have stood on the shoulders of giants,’ Hamming was forced to say, ‘Today we stand on each other’s feet.’
Follow @neilmcn© Oil IT Journal - all rights reserved.