Government Operations Committee on April 3rd, 2014

[Ginny Dybenko]

I talked a little bit earlier about establishing roles and processes, and that's hopefully one of the underpinnings to the ODI, but once you have processes and roles established, then sharing with more international participants is facilitated.

[David Eaves]

I have about three pieces of advice today. I'm really glad you raised the G-8 because there are some things going on there that I think are quite interesting.

The first thing I would say is that at a very tactical level I worry about some of the ways we might be slipping around our G-8 commitments. In fact, I was very disturbed to realize two months ago that Industry Canada, which shares a database of corporate entities in Canada, used to share who the directors of those corporate entities were. Now the list of those directors goes beyond a $5 pay wall. Rather than being able to see who the corporate directors are, you now have to pay $5 per company. That actually runs contrary to the spirit of the G-8 agreement, which was to make corporate data more transparent to the public.

In fact, if you wanted to spot global corruption, tax evasion, or problems at a corporate level, having a corporate database that is downloadable and accessible is critical to doing that. The G-8 agreed to that, and yet we've gone in the opposite direction. From a tactical place, I would encourage this committee to be looking very closely at Industry Canada's move, to understand why they made this choice.

The second thing I would say is that there is an opportunity in looking at something like corporate data. The opportunity exists in how we harmonize this data across jurisdictions. The question would be on places where we think there's policy importance, corporate transparency, for example. How do we harmonize how we release the data with how the U.K. releases the data, with how the United States releases the data? This would make analysis across jurisdictions much easier, so spotting things like fraud, tax evasion, those types of things, would become significantly easier because the data has all been harmonized.

The third thing I'd say is that if we want to take a leadership role, one of the things that I don't think our G-8 partners are doing, and one of the things that makes everybody in the open data movement very, very nervous, is that there's no protection to our access to most of this data. Our only protection is to do an ATIP request.

If the country wanted to do something that was truly transformational, it would try to figure out whenever it was passing legislation what the core datasets are that make the legislation work. What are the core datasets that allow for the transparency so that the public can assess whether the legislation is working?

The NPRI, which is the data about pollution, is a wonderful example, and that dataset is protected by legislation. Government is required to collect it. It's required to share it, by law, and it is almost unique in that way. I would love to see how we are building datasets that we think are critical to infrastructure or to accountability in this country being protected by law. Users, whether they're corporate or just citizens would know that this data is going to be around—they can actually build infrastructure around it—and they would not be taking enormous risk because the government might get uncomfortable in the future and simply pull that data back the moment it does something that it doesn't like.

[Diane Ablonczy, Conservative, Calgary Nose Hill, AB]

You're so cynical for someone so young.

[The Chair, NDP, Pierre-Luc Dusseault]

Thank you, Ms. Ablonczy.

Ms. Ablonczy's time is up, but do you have something to add very quickly, Ms. Miller?

[Dr. Renée Miller]

I think we shouldn't underestimate the difficulty in integrating datasets, and David alluded to it.

The difficulty in taking two open datasets and aligning them, figuring out if two records actually refer to the same corporate entity—if the data that they represent is actually consistent with each other, if one is in metric and one is in imperial units—is an incredibly difficult problem. It is one that requires today, with current technology, many human years of intervention in order to really align two datasets. It's a very difficult problem. My advice would be to strategically pick areas where we can see real economic benefit in aligning the data with other G-8 countries.

[The Chair, NDP, Pierre-Luc Dusseault]

Thank you.

Mrs. Day, you have five minutes.

[Anne-Marie Day, NDP, Charlesbourg—Haute-Saint-Charles, QC]

Thank you, Mr. Chair.

I want to come back to the look of the website, which is very statistical. It looks like a table of contents. Its front page is still displaying information on the CODE event that was held over the weekend in Toronto.

Something Ms. Miller said, I think, caught my attention. She said that our students were leaving for Silicon Valley. We know that Silicon Valley was built from scratch. That's not a utopia. In the beginning, there was nothing there but desert. Yet an amazing computer technology hub was built in Silicon Valley.

Here, some things have been done for the film industry—be it in Ontario or in Quebec—that have contributed to its success. Similarly, a site dedicated to open data could be created. Our students could go to some wonderful locations, such as Gaspésie or Lake Louise, in Alberta. Time will tell whether this is a utopia or not. They could perhaps be gathered in one place and be provided with the necessary tools, similarly to what was done for the people who created Silicon Valley from scratch.

What assets could the University of Waterloo provide to the Open Data Institute?

Will the University of Toronto partner with the Open Data Institute?

Would it be possible to create winning conditions, either in terms of tax credits or something like that, to keep our researchers and students in that field in Canada—at the University of Toronto, among others?

[The Chair, NDP, Pierre-Luc Dusseault]

We will begin with Ms. Dybenko, if that's okay with you.

[Ginny Dybenko]

It's an excellent question, and an excellent observation as well. It's very early stages for the ODI, the Open Data Institute. There was very definitely an inclination to do this inter-institutionally. I think the reason that Waterloo was selected is just because of the amount of success that the Waterloo region has achieved over the past number of years in and around digital technology development and incubation of initiatives.

So as I said, it's early stages yet, but there's an inclination to not only share what has been developed but also to work together with other universities.

[Dr. Renée Miller]

Yes, and to follow up on that, I think it's a very needed initiative to create something like this in Canada. I have had students who have done start-ups in Toronto, and they eventually all go out to Silicon Valley because they eventually hit a wall and there's just not the culture and expertise and enough people here to draw on to sustain their endeavours. I think we do need something— an initiative as you're saying—that is cross-institutional and cross-provincial to bootstrap that process, to create the critical mass that you need to be able to sustain a culture like that.

[Anne-Marie Day, NDP, Charlesbourg—Haute-Saint-Charles, QC]

In your opinion, what would be the winning conditions? What should be implemented?

[Dr. Renée Miller]

It's the critical mass of expertise. So I think you need the significant investment of existing entrepreneurs. You need it on the business side, you need it on the technical side, you need it on the marketing side. You need a holistic approach to this; it's not just focused around students. We have incredibly bright students but we need the infrastructure that's holistic for creating that entrepreneurial culture.

[Ginny Dybenko]

Excuse me for coming in, but MaRS has done a terrific job in Toronto, as has Communitech in Waterloo. We're working together, but what typically blocks us is the lack of venture capital to really take the initiative to the second stage of development.

[The Chair, NDP, Pierre-Luc Dusseault]

Thank you, Mrs. Day.

Mr. O'Connor, you have five minutes.

[Gordon O'Connor, Carleton—Mississippi Mills, CPC]

Thank you.

The Government of Canada has identified 14 areas for data that they're supposed to produce, and I think when we talk about government, I believe we're talking about the bureaucracies because they're the big monsters out there that make the data. I'm a bit skeptical. For instance, one of the areas chosen is government accountability in democracy. I can't imagine any government of any stripe is going to pour data out on that, but maybe they will.

We're talking about the government because I don't think private industry provides much information in the sense that they're commercial. My problem with all this data is what compulsion can we give to a government to make them produce data? Because as I said, there are 14 areas here: education, justice, energy...it goes on and on. Governments are only going to provide the data they want to provide.

I'll ask each of you in turn to answer the question. I'll start with David.

[David Eaves]

I don't disagree. It's actually enormously difficult to get governments to provide data, especially data that might make them uncomfortable.

My hope is that longer term we could end up in a world where we have a more iterative review of government where we're less driven by the scandal that we can nail someone, particularly a public servant, to the wall against over a particular error. I'd rather end up in a place where we try to iterate around solutions. Actually spotting errors in government is seen as a good thing because it allows us to iterate and make it better rather than something that drives the scandal, particularly if it's the type of problem that's really non-strategic but can end up eating everybody's time.

This would certainly be the place I would love to go. The way I think you've got to go about it, as I mentioned earlier, is that you have to think about how you're going to draft this stuff into legislation because, when you draft it into legislation, it forces parliamentarians and it forces the government to think about what is actually high leverage data and what is the data that will cause us to behave in ways that we want to behave. They'll accept a longer-term plan.

The NPRI dataset, the one around pollution, I think is a wonderful example of a government's potentially embarrassing dataset, and yet, now encoded in legislation, it causes all the incentives both in the private sector and in government to be wonderfully aligned around how we minimize pollution. So I think finding those types of leverage points is going to be critical.

The other thing is, ultimately we do have access to information legislation that should allow us to have access to this data. So if I was going to be thinking about how we prod governments along, I'd just tweak the access to information legislation so that it says that when I make a request, I'm allowed to request a dataset, and you're not allowed to hand it to me in PDF or on printed sheets; you actually have to hand me a disc or send me a file that gives the database in a machine-readable way. At that point I can get the data either way. Aren't you better off making it accessible to me so that I don't eat up a whole bunch of time making requests over and over and over again? Would it not actually reduce the burden on government?

So we actually have access to the data either way. The questions are: how do we make it painless and how do we make it easier for government? So let's put it in the legislation where we have to and then let's improve the access to information legislation.

[Dr. Renée Miller]

I show my stripes as the bohemian academic here, but I don't think that the stick approach always works. As an academic, I'm required by my funding agencies to always provide any data that I come up with that's federally funded, and I have to make that available.

There's a tremendous number of scientists who feel that their data is their own, and they don't want another scientist to make a breakthrough based on the data that they spent time collecting, right? So we find ways around that. We don't publish the stuff that we think could be valuable. We publish enough to satisfy the law but we keep to ourselves the things that are going to give us that Nobel Prize.

So I don't think just mandating it is enough. Rather, I think what we want is creative government employees who understand that, if they need certain kinds of data and certain kinds of expertise, publishing a dataset may be a way for them to get the expertise that they don't have in-house and may be a way to get somebody else to solve problems that they have. I think that we should view open data as a way of providing solutions into government, which is what I was saying about the flow back into government. Can we provide data out there that, if somebody did something with it, I, as a government employee, could benefit from that and improve my processes? If we can get governments to start thinking that way, I think we'll get better data out there.

[The Chair, NDP, Pierre-Luc Dusseault]

Thank you.

Mr. Gayler and Ms. Dybenko, you have 30 seconds to answer.

Let's start with Ms. Dybenko.

[Ginny Dybenko]

I love the carrot and the stick idea. I think in general carrots work way better, so there have to be incentives. The thing that leapt to my mind is efficiencies. So if indeed the individual government department can actually see utilizing that data to drive efficiencies that are required because of reduced funding to their organizations, that would be ideal. Then ultimately they can look for fraud and theft and other kinds of things that are going awry within their departments.

[Mark Gayler]

I think I would build on earlier comments. I think this is about reducing friction. Many governments around the world build their open data policies on the freedom of information legislation or access to information legislation that they have in place. I think this issue is really about reducing friction and making access more available.

If you can publish the data and make it consumable and easily available, then you should do that and not hide it behind a lot of bureaucracy that may be unnecessary in a particular case. That's not necessarily always the case, of course, but I think you should reduce friction where you can, and that encourages people to publish. It encourages citizens to consume.

The other thing I would say to finish is that it's important to understand that open data is not WikiLeaks. These are separate things, so I think it's important to make that distinction as well.

[The Chair, NDP, Pierre-Luc Dusseault]

Thank you.

Mr. Blanchette, go ahead for five minutes.

[Denis Blanchette, NDP, Louis-Hébert, QC]

Thank you, Mr. Chair.

This time, I will address Mr. Eaves.

At the heart of your presentation is really what Mr. Gayler calls organizational culture. In other words, data is not simply being produced for the sake of data production, especially in the government. If that were the case, we would say that taxpayers' money was being wasted. This has more to do with a way to work, an approach.

Do you have an example of what should be changed in organizational culture? For instance, what trends and methods within the federal government would help gradually build a data sharing culture? Of course, I am not talking about confidential data, although this is a very important issue we could come back to.

As it has already been said, departments often have to operate in isolation. However, what is happening in the public administration is also happening in private companies. You would be surprised to see how isolated companies can also be in terms of their operations, if they are even remotely large.

I would like to hear your thoughts on this matter.

[David Eaves]

That's an excellent question.

I agree with the carrot and the stick. I'm a big fan of carrots and I've used them with this government, but I'm also a believer in sticks. So the question is how do you manage both.

And the question of culture is enormously important. It doesn't actually matter how many rules you have in place. What I'm most interested in is how you create incentives for public servants to want to share more. And I think there are a couple of things that we have at our disposal on how to make that happen.

First, most public servants have grown up in an era where ministerial orders or deputy ministerial orders have been to not share anything because there's only risk involved in sharing. And how do we begin to crack that? How do we begin to change that culture?

The first one is there are lots of examples around how transparency can advance a policy agenda. But my most favourite example, and the one I always give in talks, is around restaurant inspection data. It turns out in L.A. they decided they were going to publish restaurant inspection results on the front doors of restaurants. The moment they started doing that, you had more people going to restaurants that had better results and fewer people going to restaurants that had worse results.

And wouldn't you know it, but it turns out that as a result, you also had fewer people ending up in the emergency room with food-borne related illnesses, which is the most expensive point of contact in the health care system.

So if you want to drive a policy outcome of reducing health care costs, it turns out publishing restaurant inspection results in a useful manner is a great way of driving that. So we have a whole bunch of examples where transparency and sharing data actually advances policy agendas. So driving those stories through the public service and causing public servants to think about where transparency is actually strategically in your interests would, I think, cause people to begin to re-evaluate why they should be sharing.

And then if we could get them to be thinking about that, we might begin to crack the door open a little bit more where they see that actually the risks of sharing this other information doesn't quite feel as high as it did before. Now they see that there are actually benefits in these policy areas where sharing created these outcomes they liked.

So that's probably the direction I'd try to go.

[Denis Blanchette, NDP, Louis-Hébert, QC]

Of course, that is part of the issue, but we need to look at all the aspects. You used an example where the stick method is being used in restaurants.

Earlier, you talked about four points with regards to open data. In sum, points two to four illustrate how difficult it is for public administrations to truly use all the data they have. You are basically saying that people could help with that a bit.

We could perhaps take things a bit further by saying that data sharing could help the government fulfill its own mission. It's amazing to see how much diversity human curiosity can create. This applies in terms of the economy, but since all governments also have non-economic duties, the overall government mission could benefit from this.

[David Eaves]

And I would say that the product recall data.... The federal government has an app on product recall data, which strikes me as a great example. When you look at the data that underlies that app, only some of it is available in the open data portal. But you have to be the most OCD person in the world to fire up an app to check whether or not a product has been recalled before you buy it.

But if you made the data available, there's kind of a long tail of users. There are people who have dairy allergies, people who have wheat allergies, or people who have kids. They're not all going to go to this app. You've tried to aggregate all of those users into one app, and they're not going to use it.

But if you actually had the data available, organizations and associations that represent those people or that serve those people might grab that data and provide it to them in the places where they actually look and they actually read. You'd actually end up having a much higher policy impact around that dataset than having the government create an app.