Insights
Portfolio Spotlight: Interview with Dan Piette, Bluware CEO
As our portfolio spotlight on Bluware comes to a close, we caught up with Dan Piette, Bluware CEO, on the current state of the oil and gas industry, and how Bluware’s unique offering can benefit both geoscientists and the wider industry.
Let’s first address the current state of the market. Between COVID-19, low oil prices, and the economic downturn, the whole industry seems to be impacted, and especially in Houston, Texas. You have been through a few of these in your tenure. How do you foresee this crisis playing out for us locally and across global markets?
All commodities markets go through price cycles. When the economy is good, the oil market is great! When the economy is bad, the oil market is terrible. Because of this, the market and the companies in this market tend to overreact. The price of oil has dropped by at least 50% six different times in my career. Every time, whether during the boom or during the bust, there were people who thought it was never going to change.
This reminds me of a story. Back in early 2007 I was talking to a friend at the Society of Exploration Geophysicists (SEG) annual meeting. These were good times, if you remember, before the financial crisis of 2008 hit the United States, and oil prices had been rising steadily from 2000, going from about $20/bbl to about $140/bbl. We were enjoying our good fortune, and my friend said, “Dan, the best thing about this boom is that it is going to last forever!” I was about to chastise him for being so foolish as to believe such a thing, when he delivered the punch line, “Just like all the others.” Sure enough, he was right. By the end of 2009 oil dropped to under $50/bbl, and another boom was crushed. Just like all the others.
People think that busts last forever as well. When oil went into negative territory in April 2020, those of us who have been here before just shook our heads and sighed. Sure enough, West Texas Intermediate (WTI) oil is now trading at about $38/bbl.
I firmly believe that there will come a time when we do pass ‘peak oil demand’ as the realities of climate change, and our participation in it, becomes even more clear. By the time someone in their twenties has grandchildren, we may see a significant decrease in the production and consumption of hydrocarbons.
Now more than ever before, operators and the supply chain are pushing to find more efficient and cheaper ways of discovering and producing oil and gas. What are your thoughts on Bluware’s place in addressing this market need, and how and where are the solutions to shorten time to oil?
Just as I’ve lived through six different oil busts, I’ve lived through three separate technology waves in oil and gas. Each of them has introduced their own benefits, come with their own costs, and were resisted by the entrenched incumbents.
These three waves were the Unix workstations like the old Landmark Graphics and GeoQuest Systems, then PC’s like Technoguide’s Petrel, and now the current wave of artificial intelligence and deep learning.
Breakthroughs in artificial intelligence and machine learning are changing the way other industries work, but it feels like many of the products we see used in oil and gas exploration are a step backwards, not one forward. You see many companies offering a ‘black box’ approach to machine learning. You give them your data, they do some magic, and you get back a result. The challenge is that you can’t edit it, you can’t question it, and you don’t even know how they came up with it. You have to conduct quality control, and that often takes longer than it would have to manually interpret it in the first place.
In all fairness, this is not the result of deception, but the result that almost every bit of seismic data collected in the last 40 years is stored in a SEG-Y file. SEG-Y files, while great for nine track tapes, are not great for machine learning. You need to chop up the data, then screw it all back together, time and time again.
Bluware approaches the problem fundamentally differently, by taking a step back and doing a root cause analysis. It isn’t the machine learning that is not working, it is how the data is being prepared.
Paul Endresen, Bluware’s Chief Technology Officer, figured out years ago that if we change the data, we solve the problem. He came up with a system we call Volume Data Store (VDSTM), which gives every geoscientist the ability to approach machine learning, or more specifically deep learning, as an interactive interpretation tool. Geoscientist can now do what is most important, such as think about the origin of the oil, the history of the sediment and basin development, and the fundamental underpinnings of oil creation.
What technologies does Bluware offer that will specifically benefit geoscientists who are working remotely for the near term?
The machine learning and deep learning techniques I mention above really shine when they are used in the cloud. Having cloud-native technology gives energy companies the ability to interpret any amount of seismic data, anywhere, at any time. As long as the legal restrictions regarding the access and entitlements to data are met, any earth scientist can work from home on a PC, from the beach on a tablet, or from an airplane with a smartphone. By making sure that the computer power, and the control of that computer power exists in the cloud, a geoscientist can spin up as many nodes as needed to get any job done in a timely fashion.
This is different from what everyone else is doing. Bluware is not a ‘lift and shift’ operation, simply paving the cow path. The important thing about looking back at the three waves of computer technology in the energy industry is to understand and not repeat the mistakes we made in the past. Many of the early computer systems for seismic interpretation failed because all they were doing was a ‘lift and shift’ from paper to a computer. None of those early systems were fundamentally different from using paper to interpret seismic.
Why bother moving to the cloud if the only advantage you get is that the computer is remote? That is not moving forward. By changing the way you interpret the data, moving the applications to the data, and only keeping one copy of the data, you end up with a better, faster, and cheaper way to find oil.
I am bold enough to say we have reached that. For if your reach cannot exceed your grasp, what is a heaven for?