Open up
Matt HallAfter a short trip to Houston, today I am heading to London, Ontario, for a visit with Professor Burns Cheadle at the University of Western Ontario. I’m stoked about the trip. On Saturday I’m running my still-developing course on writing for geoscientists, and tomorrow I’m giving the latest iteration of my talk on openness in geoscience. I’ll post a version of it here once I get some notes into the slides. What follows is based on the abstract I gave Burns.
A recent survey by APEGBC's Innovation magazine revealed that geoscience is not among the most highly respected professions. Only 20% of people surveyed had a ‘great deal of respect’ for geologists and geophysicists, compared to 30% for engineers, and 40% for teachers. This is far from a crisis, but as our profession struggles to meet energy demands, predict natural disasters, and understand environmental change, we must ask, How can we earn more trust? Perhaps more openness can help. I’m pretty sure it can’t hurt.
Many people first hear about ‘open’ in connection with software, but open software is just one point on the open compass. And even though open software is free, and can spread very easily in principle, awareness is a problem—open source marketing budgets are usually small. Open source widgets are great, but far more powerful are platforms and frameworks, because these allow geoscientists to focus on science, not software, and collaborate. Emerging open frameworks include OpendTect and GeoCraft for seismic interpretation, and SeaSeis and BotoSeis for seismic processing.
If open software is important for real science, then open data are equally vital because they promote reproducibility. Compared to the life sciences, where datasets like the Human Genome Project and Visible Human abound, the geosciences lag. In some cases, the pieces exist already in components like government well data, the Open Seismic Repository, and SEG’s list of open datasets, but they are not integrated or easy to find. In other cases, the data exist but are obscure and lack a simple portal. Some important plays, of global political and social as well as scientific interest, have little or no representation: industry should release integrated datasets from the Athabasca oil sands and a major shale gas play as soon as possible.
Open workflows are another point, because they allow us to accelerate learning, iteration, and failure, and thus advance more quickly. We can share easily but slowly and inefficiently by publishing, or attending meetings, but we can also write blogs, contribute to wikis, tweet, and exploit the power of the internet as a dynamic, multi-dimensional network, not just another publishing and consumption medium. Online readers respond, get engaged, and become creators, completing the feedback loop. The irony is that, in most organizations, it’s easier to share with the general public, and thus competitors, than it is to share with colleagues.
The fourth point of the compass is in our attitude. An open mindset recognizes our true competitive strengths, which typically are not our software, our data, or our workflows. Inevitably there are things we cannot share, but there’s far more that we can. Industry has already started with low-risk topics for which sharing may be to our common advantage—for example safety, or the environment. The question is, can we broaden the scope, especially to the subsurface, and make openness the default, always asking, is there any reason why I shouldn’t share this?
In learning to embrace openness, it’s important to avoid some common misconceptions. For example, open does not necessarily mean free-as-in-beer. It does not require relinquishing ownership or rights, and it is certainly not the same as public domain. We must also educate ourselves so that we understand the consequences of subtle and innocuous-seeming clauses in licences, for example those pertaining to non-commerciality. If we can be as adept in this new language as many of us are today in intellectual property law, say, then I believe we can accelerate innovation in energy and build trust among our public stakeholders.
So what are you waiting for? Open up!
Reader Comments (4)
Matt: I like the idea of open data, and I think it would be interesting to map the "depth volume" of the earth (to make up an expression) similar to how we've pretty effectively mapped the surface volume (both natural and man-made). We can visit any city in the world now, view buildings in 3D, view street views, and even the interiors in some cases, so why not down below?
Imagine a version of Google Earth where you could cut sections and get accurate geo-data at any point. There is a huge amount of data to be sure (check my math but it's about 5 times as much), and it's much more difficult to take an accurate picture of it.
Although, and I'm stepping way out of my league here so correct me if I'm wrong, you don't need to take a picture of it in the same sense as you would with surface data. Subsurface data can be stored as data points which are more efficient than pictures (compare the filesizes between bitmap and vector file formats).
The biggest hurdles I see are: 1, collecting accurate data that the community accepts as accurate; 2, storing that data in a central data centre; and 3, accessing that data is a way that is usable.
I can imagine there is a feeling of proprietary-ness regarding sub-surface data that companies collect, but can they really own it? That's like saying "I saw this piece of land first, so it's mine".
Oh wait, we've been doing that for thousands of years.
At least other nations have historically been respectful of each other's discoveries, and nations have openly shared their spaces and the resources therein.
Oh wait...
Still, it's worth a shot. Good luck!
@Reid,
A company called RockWare has built some simple corner-point geo-referencing tools so that any map or any seismic line can be plotted in ArcGIS or in Google Earth (I think). I watched a demo on their website, and all the seismic lines were hanging above the earth surface! So although they were positioning tiff flies of data in the correct geographic location, it was a fail. I have done some extremely rudimentary overlaying of geologic cross sections on the z-planes of objects in Google's sketch up - which was a faff, but fun nonetheless.
Your three biggest hurdles are all correct, but there is at least one more; Subsurface data is never regularly sampled. High density sampling in places of economic or scientific interest, low sampling in areas of no interest or difficult access. I suppose the same can be said of Google Earth. High sampling in the streets of Manhattan, but low sampling the mountains of Mongolia. A so-called global open repository would also be lacking, because the correctness of a seismic image, for instance, is in the eye of the beholder. Subjective because of the interpreter's intentions. Simply putting images out there for 'public' use might be a useless endeavor. Data can can take hours or weeks to interpret, and even then it can still look like noise. Photographs don't typically have that quality. They just are what they are. The optical analogy is the Lytro light-field camera. One environment, many different image possibilities. It is difficult to say which one is "correct" or "accurate".
Can companies own the data? You bet. "I saw this piece of land first ...", this is why the Alberta government cracked down on oil and gas trust companies about six years ago. Same goes with the best and the most recent satellite photos on Google Earth, to some extent. Data are regarded as assets, and assets are valuable. But as Matt has described in the past, that is not what companies should be relying on for a competitive advantage.
@Evan:
Companies can own the data, but they don't own what the data represents. I may take a picture of a mountain, and the picture is mine, but the mountain is not. Maybe my picture is interesting or useful or somehow revealing of a detail or perspective that someone hasn't seen before, but the mountain was always there to begin with, and it is the same it always was, regardless of how I looked at it.
Subsurface is different though because there is no real way to plainly see the "truth"; all we (well, you) can do is interpret the data, and the aggregation of these interpretions becomes the de facto "truth". It's like understanding the mountain by looking at it's shadow.
There is a fantastic resource of openfile petroleum data avaible for Western Australia. Its called WAPIMS and you can access it by signing up on http://www.dmp.wa.gov.au/
See review: http://myresources.com.au/publications/oil-a-gas-bulletin/news/3980-wa-information-systems-win-award
It includes well data, biostrat, seismic, biostrat, core photos, core descriptions, image logs... everything!. Its a great resouce and its freely available.