What do you say when the boss asks for your opinion?

Neil McNaughton recalls his own wrong answer along with a historical boardroom-driven technology fail. Pondering BP’s billion-dollar contract with Palantir and the rebranding of the SPE ‘information’ special section as ‘analytics’ he postulates that AI group think has taken-over the industry.

Back in the mid-1980s when I was a geophysicist, the boss sauntered into my office and asked, ‘Neil, what would you think of a digital geophone?’ At the time, seismic recording had already been digital for twenty years or so, but the analog-to-digital conversion took place in the recording truck. I was scratching my head trying to see the point in having all that fancy electronics distributed across the length of the spread in every single jug. What about the bandwidth, the conversion range, how many bits etc.? My answer was non-committal. Which of course was the wrong answer! The right answer, when the boss comes in proposing to make a major acquisition, whether it is for the manufacturer of a new geophone, or a purveyor of the latest digital technology for machine learning/artificial intelligence, is ‘Yes boss, that’s a great idea!’

Anyhow, the boss took no notice of my agonizing and went on to acquire the geophysical company (Input-Output). The digital geophone took another decade or two to see the light of day, so my skepticism was not completely misplaced. Today, you can even build your own digital geophone with a kit from Raspberry (Pi) Shake. But I digress.

Back in the day, deals like this were made on the golf course, struck with a handshake of colorful individuals who either possessed considerable financial means or who were capable of smooth-talking others into ponying-up their cash. But not all such ‘top-down’ deals, i.e. ones that suddenly appear in the boardroom, rather than from in-house specialists, are done on the golf course. A spectacular instance of a top-down deal was done in France in the 1970s when the then Elf Aquitaine’s top brass got into a huddle with leading French politicos and a scurrilous Italian ‘inventor’. The great ‘avions renifleurs*’ scandal began when Aldo Bonassoli managed to get the ear of France’s SDECE intelligence agency. This led to top secret tests of his ‘gravity wave’ device that was claimed to detect oil directly**. The tests were performed in the presence of Elf’s top brass, France’s president, Valerie Giscard d’Estaing was even involved, but without any of Elf’s geophysicists in attendance. In the end nothing came of the ‘technology’ but Elf and, one imagines, the French taxpayer were parted from a considerable sum.

I’m not party to how these deals are done today, but I am pretty sure that the ‘top-down’ approach to major decision-making is at least as prevalent today. In the field of computing it has been honed to a fine art as company bosses not only make deals buying startups and making bets on the next big thing, but they also boast about them.

A report in the Financial Times has it that the new CEO, Bernard Looney, wants BP to be ‘the leading digital upstream business’. As a part of this drive, the FT reveals that BP has ‘expanded its relationship with Palantir Technologies, the ’secretive’ Silicon Valley data-analysis company, with which it has a $1.2 billion 10-year contract for its data integration platform’. Palantir’s technology is at the heart of a ‘digital twin’ of BP’s global infrastructure, performing simulations to determine optimum routing for BP’s production and optimizing maintenance. Looney is quoted as saying that ‘immediate access to data is vital to attract the next generation of employees, it is the older hands that are having difficulties adjusting. […] The problem that we have is that people have been working a certain way and believe that answer that they’ve got is right. […] Anywhere in our system where there are tons of data. There is value that is being unrealized.’

Now I am not saying that Palantir is the next ‘avion renifleur’, but there are some uncomfortable parallels here. A secretive company with black-box technology***. A boss who disses in-house expertise. The promise of great benefits. What is new is the way that this staggeringly large investment in software is broadcast to all and sundry. In the early days of the digital oilfield, BP excitedly spoke of a ‘billion dollar’ bottom line addition. Now the boast is of a billion dollar software spend!

The paradigm of massive data lying around with unrealized value comes straight from the marketing material of the consultants and IT vendors. As a geophysicist, and an oldie, I feel almost personally responsible for this terrible state of affairs. But I am scratching my head to think of how the massive amount of seismic data ‘at rest’ could be usefully put to better use and exactly what value is unrealized. Most professionals have an approach that involves doing just the right amount of processing or other work. The new paradigm has it that you have never done enough. That black box is always ready to re-run some big data algorithm and come up with the next insight that was ‘invisible to the human eye’..

The AI revolution in its latest manifestation is old enough now for us to expect to have seen some great results. Here at Oil IT Journal we have been tracking progress in the AI field over the last few years. What we see is continued enthusiasm for trying stuff, especially the stuff that is more or less freely available from Google and Microsoft. The results? Well, we have seen proofs of concept that show image recognition of scanned logs and NLP on document retrieval that are interesting. But accuracy is usually quite low, often in the 60-70% range. This means that AI is applied in special circumstances like a data room where there is a need for quick, rough and ready results.

I’m just back from the Calgary SPE ATCE and am in the process of unpacking my notes and looking through some presentations. I will report back in a future issue on the state of play in upstream AI. But so far, nothing staggering revolutionary to report. What did come over loud and clear from the ATCE is the group think that has it that the AI revolution is already mainstream. To reflect this apparently obvious ‘fact’, the SPE Special Section for ‘Management and Information’ has been split into two disciplines. While splitting out ‘Management’ makes sense (it was always puzzling to group this with ‘information’ and not very many managers seemed to be onboard), the rebranding of ‘information’ as ‘data science and engineering analytics’ seems to be an acknowledgement that all that old-world scientific computing stuff is out of the window.

* Wikipedia: The Great Oil Sniffer Hoax.

** Bonassoli’s device is one of many scams/inventions that your average geophysicist is confronted with during their career.

*** Some put it even less politely.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.