Natural Resources Committee on May 8th, 2018

[The Chair, Liberal, James Maloney]

Thank you, Mr. Lourie.

Mr. Mustard.

[Donald Mustard, Researcher, As an Individual]

Thank you for inviting me. My apologies: there were some pictures to go with this, which you will eventually get.

The driving of the last spike in 1885 was the culmination of a nearly two-decade effort to bring certainty to a nation. It forged a national identity in steel and steam, in iron and timber, and 150 years later, Canada, a prosperous nation with strong linkages to the south and opening markets in Asia and Europe, seeks certainty in the development and delivery of energy data.

Why is data important? I guess you've had a lot of presentations on why it's important. I'm sure that in your daily life you do know.

For the next 10 minutes, I'll outline why an agency that can provide timely, reliable, and transparent energy data is necessary. I'll discuss the necessary elements of data management, acquisition, and sharing, define leadership gaps in transitioning to data-driven decision-making, and the steps to greater energy certainty, not only as a national policy but as a national imperative.

In preparing for this presentation, I came across a May 2017 Economist briefing that drew a striking analogy between data and oil. “Data”, the authors propose, “are to this century what oil was to the last...[the] driver of growth and change.” They continued:

The new economy is...about analysing rapid real-time flows of often unstructured data...photos and videos generated by users of social networks, the reams of information produced by commuters on their way to work, the flood of data from hundreds of sensors in a jet engine.

The statement resonates with me as someone with a lifetime of experiencing work in and around the mining and energy industries, first as a researcher-scientist and oil production worker, later as a communications professional and atomic radiation worker, and, for the last 16 years, as an investigator of railway and pipeline accidents.

Why is data important? Or, more importantly, why is a national data agency necessary?

As the article hints at and a colleague of mine recently told me, the fundamental problem is that we're not getting snapshots of information about energy fast enough to make informed decisions about things such as energy planning and environmental impacts and such. It might be easier to ask, what are the costs of not having a national energy data agency in Canada? I think we're all living that.

If you had my pictures right now, you'd see a picture of the Mackenzie Valley, where the pipeline was first proposed as a joint venture partnership. A new effort came forward 27 years after that, which finally I think overcame the barriers related to aboriginal and industrial co-operation. As most of us probably also know, in December 2017, after a six-year-long process to reach approvals, the partners walked away—again.

We're losing opportunities, whether these are the right opportunities or not. I think the persons responsible for that.... I think there was a comment to this effect: “I don't know what the problems are, but a process that should take two years in a business cycle can't take six”. The fundamental economics change so quickly that they can't do what they'd like to do.

The story also goes towards September 15 of 2008—in my story—where there was a meeting convened in New York of the leaders in the financial world to discuss the bankruptcy of Lehman Brothers. When they asked these leaders of business what their exposure to Lehman Brothers was, nobody knew. Once again, it was a great crisis in data.

What you need to know is that to manage, one needs to measure. To measure, you need to audit. To audit, you need data. In today's day and age, to know what's happening and what is going to happen, you need to know what is happening now. You need real-time data. The EDM Council, the Enterprise Data Management Council, which grew out of efforts to teach the finance industry how to do this, described a holy trinity of data management.

You need unique and precise identification of things. You need unified views of meaning across organizations, locations, linkages, and interconnections, and the procedure is actually the reverse of what you might think. You've got to start with what your business practices are, what you are trying to do, and work back, reverse-engineer, to what you need to do that, what the critical data elements are that are necessary to do these processes. Then once you've identified those critical data elements, you need to clearly and uniquely identify them, the taxonomy and the ontology, so you actually can work with them and everybody's working with the same understanding.

You need to establish a unified view across organizations.

I'd like to just note that many of us are dealing with something called Phoenix. Most people see Phoenix as a data processing problem, and it is in fact not a data processing problem. It's a data meaning problem. They started with the process, and then tried to make.... I guess the analogy might be that they started with the person and tried to make the pants fit the person.

We are now in a situation where we don't have the data we need, but we also don't have a common understanding. Further with the focus on modern data storage, large storage models, we all know the concept of the data lake. Really a lot of the discussions I heard when I was first sitting here, half an hour ago, were about structured data, and that's the least of our problems. In fact, our biggest problem may be that we're waiting until this data is structured before we actually take it and use it and apply it. It's too late then. It has already got the data tax applied, and the data tax is one of the things that you guys are discussing. You can't get the data fast enough to make meaningful decisions. You lose the opportunity whilst you're waiting for something to happen.

There's another thing that we're lacking. I spent the last two years in a program at Columbia University, a masters in applied analytics, and I went there because this was a program that was focusing on the leadership skills you need to do data processes. I was one of the people who worked on Lac-Mégantic, one of the investigators, and when we came through that process, there were 18 causal factors to the accident. As a person standing back, an individual, as I sit before you, what I saw was 18 opportunities to intervene that were missed, and when you looked at it, there was more than enough data. There was tons of data, but there wasn't data being prepared and provided in a timely manner and analyzed in such a way that you could take action to prevent the catastrophe.

Interestingly enough, the data that is currently out there in the public venue is the data mostly from media, and media gets involved when it's newsworthy. So the focus in the resource sector is typically on the low-frequency catastrophic events, which have horrific results. We're failing on two levels, one because we're not getting the data to prevent them or mitigate them, and on the other side because we're not sharing the information of what is happening.

One of the things that I've been trying to push forward is that we can't just say.... This institute is important, but you can't just think of it in isolation. Data cannot function in isolation. We have lots of silos in government where we've collected wonderful information, and I sometimes call it hoarding. I come from a family of great hoarders, so I understand a little bit about it. The reality is that we collect an enormous amount of data—and this is not unique to government. Data is dirty, it's messy, it's hard to work with, it's frustrating, it's inconsistent, and we don't want to give it out until we know we've got it right, and that's not how it works in today's day and age.

We need to get that data out of its silos. We need to get it into a process where we can actually access and use it, and the data lake and the modern data principles don't care what format it's in, as long as we've identified what it is and we know where to get it, and we'll process it when we use it, right? I call that schema on read.

What the program at Columbia was designed to do was to create a—

[The Chair, Liberal, James Maloney]

Mr. Mustard, I'm going to have to ask you to wrap up.

[Donald Mustard]

Yes, you're going to have to cut me. I'll close up. I'll give you the four points, the things that I'd like to see here and I think we need to do.

Firstly, augment, don't duplicate. The EIA has been doing this in the U.S. for 40 years. Let's not try to repeat what they're doing. Let's learn from them and let's get it going, right?

Employee enterprise data management: let's make sure we have a common understanding and meaning. Everybody knows we're playing in the same field and from the same rule book.

I'd like to see trained subject matter experts, a distributed cross-functional network of chief analytical officers with subject matter expertise to interact with leadership in all departments and agencies, and also to work back with our natural engine for this, which is Statistics Canada.

We need shared open transparent data in a fast-fail iterative lean process. We need to be able to work in a way where we're testing, and we're not worried. The modern world is open source. We share our mistakes and our successes, but by working together we get a better result.

Thank you for your time.

[The Chair, Liberal, James Maloney]

Thank you for your presentations.

Mr. Harvey, you're going to start us again, I understand.

[TJ Harvey, Liberal, Tobique—Mactaquac, NB]

Thank you, Mr. Chair.

I want to start by touching on something Mr. Lourie said. At one point in his presentation he said we need to figure out how to have an intelligent debate in this country on a range of issues. I want to encourage him that if he figures out how to do that he should let the rest of us know. We're all looking for the same goal.

You talked about the unravelling of our energy data management overall in the late 1990s and early 2000s. What do you feel precipitated that?

[Bruce Lourie]

I think the technical answer was the need to reduce government expenditures. A major government review took place in the 1994-95 period of the Liberal government of the day. That led to significant cuts across the board. My experience is that often things that look easy to cut like science budgets, technical support, supporting modelling get cut easily because there aren't the vociferous lobby groups out there to support those things.

Then I think those kinds of cuts and that kind of thinking just proceeded. Over the past 10 years there was a more deliberate effort, almost ideologically driven, to reduce the science capacity of the country. It's really a combination of things.

[TJ Harvey, Liberal, Tobique—Mactaquac, NB]

Do you feel that if an organization like the U.S. Energy Information Administration had been in place at that time and was an independent body that was not related to NRCan and had no relationship with Stats Canada but was operating under a federal funding mandate completely removed from a long-term work plan, this type of organization could have survived such a time frame and would have most likely changed the outcome of what our data situation looks like today?

[Bruce Lourie]

I believe that 100%. I think it's a combination of building, as they did in the U.S. and in Europe and the U.K., a credible entity first of all that has value to people. It has to be credible. It has to have value. It has to deliver products to people. It has to be well managed. All those things need to be done well.

Once people start relying on something like that, as not only Americans but we do here on the EIA, then it becomes difficult to take something away from people that has value. It has business value, policy value, social value. I agree very much with that statement.

[TJ Harvey, Liberal, Tobique—Mactaquac, NB]

Coming from private business—and my honourable colleague Mr. Falk will probably agree with me on this—there's an old saying in business that if you're green, you're growing. If you're ripe, you're rotten. One thing that I think is important to note is that Canada is an established economy just like the U.S. is an established economy. The way you govern an established economy is far different from how you would establish a new, emerging economy.

If you look at it from a business perspective, if you took over ownership of a company and that company was presenting you with 18- or 20-day month-end reports on the fiscal situation of the cash flow of that company you would say that was unacceptable and that you needed to get to five days.

Isn't that what you would say?

Then you would make adjustments to allow that to happen. It wouldn't take six years, and it wouldn't take thousands of hours of consultation to get there. You would take decisive actions that would allow for a course change correction that would happen in a more timely manner. I think one thing I'm always frustrated about with government is that a lot of times we focus on results. It's very difficult to effect change if you focus on results. If you focus on change, you always effect results.

If you could see this organization unfolding, what do you think the best first step would be that the government could take in the creation of an independent organization that would serve this purpose?

[Bruce Lourie]

I'll jump in quickly.

[TJ Harvey, Liberal, Tobique—Mactaquac, NB]

Not over a five-year period either.

[Bruce Lourie]

I would say the first step would be defining what it means to create something like this. I think that's defined. I think the work that Allan has done at CERI has pretty much defined what this thing needs to look like. This is a classic example that perfect can't be the enemy of the good. It's getting things started. It's sitting down with the provinces, the utilities, and the provincial regulators that want to sit down with the federal government and asking what this thing going to look like and who we need around the table.

Let's figure out, as well, what we are already doing, because the idea of augmenting what's being done is important. A lot of data has been gathered all over the place by all kinds of organizations, and we don't need to duplicate that, but we need to get people into a room to figure out a better way of coordinating.

[TJ Harvey, Liberal, Tobique—Mactaquac, NB]

It's important for the federal government to come with an overarching structure to consult with each of these jurisdictions with a plan of how they see this independent organization functioning and then start with organizations, both public and private, that are on board with that, and try to build from there.

Do you think that it's something that has to be done with achieved consensus from the get-go?

[Bruce Lourie]

The first thing is that the federal government has to get their act together in terms of what NEB, NRCan, Transport, and Environment are doing. There has to be some internal consolidation of what data is being gathered right now in the federal government. Then I would say there needs to be a more iterative process. I'm not a big fan of consensus, but I'm a big fan of getting the right people in the room.

[Donald Mustard]

I would agree with the iterative word. I think you need to treat this like an incubator process. I think you need to go in there lean and mean. I used to have a small business. I did it for 10 years. What I realized was that, when you have no money, you learn to do things that you didn't think were possible. So I think you don't need a lot of resources, but you need to get out there and talk to the people who are doing it.

Take a page out of the Chinese business book: “We're not going to invent it; we're just going to take it once you've got it working, and we are going to avoid all those overhead development processes.” Train in every department and agency in the government the skills to lead. You need subject matter experts who understand the data-driven process, who can work with technical teams, who have learned how to do change management, and can work and flex quickly and build that network to support through an engine. I would do it through Statistics Canada, because they have the machine. Those entities are going to need to be tapped into your C-suite. They need to be sitting at a table with the decision-makers and helping translate what they want to do into something of business value.

[The Chair, Liberal, James Maloney]

Thank you.

Mr. Schmale, I believe you're splitting your time with Mr. Falk.

[Jamie Schmale, Conservative, Haliburton—Kawartha Lakes—Brock, ON]

I might go the distance, Chair.

[The Chair, Liberal, James Maloney]

Okay.

[Jamie Schmale, Conservative, Haliburton—Kawartha Lakes—Brock, ON]

Thank you, witnesses, for your time. We greatly appreciate it.

Mr. Mustard, can you continue that thought about Stats Canada being the lead organization? As we kind of weigh that back and forth here, creating a new agency, department, or whatever you want to call it, is kind of....

[Donald Mustard]

You want to know how StatsCan could be a lead organization?

[Jamie Schmale, Conservative, Haliburton—Kawartha Lakes—Brock, ON]

Correct.

[Donald Mustard]

It's an old and wooden organization.

[Jamie Schmale, Conservative, Haliburton—Kawartha Lakes—Brock, ON]

Do you think we can reform Stats Canada as well?

[Donald Mustard]

I think what I'm talking about here is that any analytical process is a triumvirate. You have the subject matter expertise, you have the statistical skills, and you have the technical science skills. Those data science people need to be pointed in the right direction. That's what your subject matter expert is. He's not somebody who is going to be doing the work, but he's going to know enough about it to be able to tell people what to do and to be able to work with those technical people who, frankly, often aren't extroverted in nature. They are very good at doing work but not necessarily....

You know, one of the big problems with statistics is, one, you do statistical studies on things that everybody knows the answer to anyways, so why did you bother? Or two, you find a really neat answer to a really cool problem that has no value whatsoever.

That's how you train it, but the engines, tools, and machines are there, the super computers that you need to process. The knowledge to build a data centre and to manage enterprise processes is there. Why reinvent that? Help them, inspire them, motivate them, and give them a purpose. You would be amazed at what they can do.