 is in price of 40% to 50% per year. So that's quite substantial. OK, so let me go to the, let me check my time. So the question, one of the questions I had you ask is, what should the role be of big data for future ICT management? And I think that's a very important question. How should we deal with big data? And let me just come up with the conclusion. We see from our research, it's quite clear that big data offers huge and huge opportunities for new ICT indicators. You can derive all types of new ICT indicators from big data. But more, and then looked at in this specific research, when you found other research, that big data also offers all types of interesting insights for non-ICT indicators. One of my favorite ones, if you look at pork, pig meat, it's very easy to trace data to have the amounts and where it stems from. And big data was one of the, and if you look at pork, it was one of the segments where big data was actually very useful from our perspective. So it's not limited to ICT. And there's a fire hose, and it's always a risk to drink out of a fire hose. And I think that's one of the main risks from big data. And how do you prevent that you start lying under that fire hose and you simply drown by drinking out at fire hose? And one of the things that came very clear from the research we did over here, but also from other research we did, was that if you want to focus on big data, you want to use big data, you have to find a specific nexus. You don't have to think about specific indicators you want to have. You don't want to focus on specific data you want to use. No, you should focus on a specific nexus. What do I mean by the nexus? I mean a certain spot, in this case the ICT ecosystem, where you have only a limited amount of contracts to derive much and much data. And we identified a set of them for this context. Now I'll go to them. One of them, of course, is mobile devices. If you look at mobile devices, you go a little bit deeper. You see, for example, that on an iPhone, a simple iPhone, all types of data is simply tracked. You can fill it in. And you can see, if you look in, I'll show you, if you ask me, my colleague who knows everything in detail. He knows which languages you are using, where you're using it, how often you use it, which type of keyboards you are using, which country you are, et cetera, et cetera, et cetera. And for Android, roughly the same holds. And obviously this data is uploaded to the systems of both Apple and Android servers. So they know what cell phones are doing. So a specific nexus would be mobile phones and cell phones. And you need only two big parties. Obviously there are other systems, but Android and iPhone make the most of the market. And you see lots of lots of data you can derive from that for only two parties. But you also have other examples, of course. The content delivery networks, we see in telecommunication networks, we see quite often, especially in the more developed countries, that are content delivery networks, which are caching information and data that is used quite often. And if you look at these few parties, for example, Akamai is a big one. If you look at them and you ask them, can you give me your data? Also, you will have insights in how quick the data is transferred and which data is typically transferred to which countries. And you have a set of them. I won't go into all of them in detail, especially regarding the time. But social networks, of course, if you have the data of Facebook and also maybe a Russian VKontakte, I think it's gold and a Chinese one, and a few social network sites, you already have a quite a good image of the social network. Telecommunication equipment, the same holes. Cisco, Huawei, and maybe a couple of other ones. And you already have quite a lot of information. Telecommunication networks, this often, if they're used, is the call detail records are often used in countries. They give lots of information, lots of insights. But from a perspective of ITU, it also is quite a challenge because you do have to contact all the telecommunication networks in the world would be quite a lot. You can use the NRAs, of course, at NSOs. But then again, you have another layer in it. Auto-update services is a nice one, of course. If you have an office or your operating system doing an update, Microsoft simply knows where the data goes to, how fast it goes, et cetera, et cetera. Anti-spam services could give you interesting insights. Security vendors, where viruses are, which computers, where are infected, and of course, the mobile payment platforms are an interesting one. Instant messaging is the last one. And especially WhatsApp, of course, is very interesting. So let me go up to the final part of my presentation. I'm back to the conclusion. And I added something, that big data. It offers you opportunities for new ICT indicators. But it does that for everyone. It doesn't only offer the opportunities for you, for the public sector, for the NSOs, for the ITU, and other parties in the public domain who are collecting and presenting data. It offers it for everyone, for all types of companies, for all types of organizations. If they are good in it, if they have the data, if they have smart IDs, they simply have these ICT indicators, and they can present it. And they are already doing it. So I think that's very important to know that, and to see that, that it's an innovation. It's an opportunity, but it's also a threat. A threat for you, and I like to refer to Mr. Schumpeter in his important book, he wrote about creative destruction. And this is a typical example of creative destruction. It could destruct all types of existing structures, both organizational, but also commercial. And it will lead to change. And I think that's an important part for you and a call to action for you, that you need to act. You need to have a plan. You need to go somewhere. You need to have a plan to act together and to do something with big data. Finally, refer to the famous quote of Mr. Darwin. We said, not that the survival of the fittest, but the survival of the one who can best manage change. And even in this case, maybe exponential change. Thank you very much. Reg, thank you very much for getting us off to a good start. I hope you were keeping notes and have got some good questions to challenge him with. Do you agree with his argument that you need to focus on the nexus and not be under that fire hose? I'm not sure I do entirely. So you will have your own questions, no doubt. We're going to move on now and get a UN perspective. I'd like to introduce you to Ronald Janssen, who's the chief of the trade statistics branch of UN statistical division. And he's going to bring us up to date on a number of different initiatives that the UN stats community has been undertaking around big data and how it all kind of stitches in to the post-2015 development agenda. Ronald. Thank you very much. And good afternoon. Yes, I'm going to give the perspective of the United Nations. My background is trade, but as you can see, we now also have big data at UN.org. If you write to big data at UN.org, it will get to my team. I'm going to say a few words on big data, specifically for official statistics and would like to immediately start off with the fact that big data, of course, is used in many different ways. Is used a lot already by private sector for their marketing and other purposes. And the focus here will be on a very limited set, which would be big data for official purposes. So what I was asked when Susan asked me to present something here is if I could say specifically something about the new established United Nations global working group on big data for official statistics, and I will do that. And I will also say a little bit about where this came from and what our objectives are. And I will say a little bit more about also data revolution, which Mr. Lo already spoke about. And then some of the big data sources we specifically looking at. Big data for within the official statistics community, we brought it to the Statistical Commission about two years ago, actually the beginning of last year, where we had both private sector and official statisticians and we were very happy that Parli Lohola was also there in presenting about what big data could mean for policy development and official statistics. Out of that meeting or as an outcome, we wrote a report to the report, as they call it, of the Secretary General for the Statistical Commission. We did this at the end of last year to the commission of earlier this year. It's on the web, give some backgrounds about the kind of issues we're looking at. But one of the outcomes, and I hope I can get there, was a decision to establish a global working group on big data for official statistics, and I'll come to that. We had a first meeting, that group was formed around May this year with about 28 members, about 18 of them from national statistical officers and then about 10 of international organizations. We came very recently, only about three weeks ago, we met in China. We had first a three-day conference discussing a number of data sources and the kind of challenges which we would have with using big data. And then we had a one-day meeting specifically with this group to discuss what our work plans would be especially for the coming year. So this is what I will talk about. This is the report which we are going to submit to the commission in coming March. And the report will be ready December, so we are finishing up on this. We'll have an overview of the conference and then it will go into the details of terms of reference of this group and the program of work which we have set for us. And then I will also, and I will also say something about this during my 10 minutes, about the results of a survey we did on big data projects which are currently ongoing in various offices in the world. At the conference in China, we had mostly a debt meeting, mostly people from the statistical world, a few from private sector, a few academia, but most people were from the statistical systems. We discussed three data sources. One was we went into detail about how mobile phone data could be used. And this is certainly from the point of view of passive data which is location data of where mobile phone data are registered. There was also a little bit of discussion about if we would use active mobile phone data which is with consent of the mobile phone users themselves. Satellite imagery is used a lot for agricultural statistics. Environmental kind of statistics is where there is a lot of interest to see how we can use satellite imagery together with estimation models where the statistical lovers are coming in with academia to use this for those kinds of purposes. Twitter and social media are being used. We had an expert from the University of Pennsylvania giving us some insights about how they analyze data. There are some examples of this and I think my colleague might say something about this. Statistics Netherlands is actively using Twitter and Facebook data to measure consumer sentiment or consumer confidence. And they can do this at a frequency which is higher than the usual surveys. The service is about month by month with Twitter and social data. They can do it almost at a weekly or bi-weekly basis. And so the results are that good that they're thinking actually of switching to using that big data source. We then discussed part of the benefits and challenges which we have with several different data sources. So one of the things we're looking at is more on the challenges side is a lot of the things is about access to data. Access to data has to do with privacy concerns with building trust and we would like to see that we would build with the private sector which is really needed as a collaboration for using data for policy purposes. So get what we would call umbrella agreements. So can we get with global mobile operators, can we have a kind of umbrella agreement about how mobile data could be used for purposes of public good. So those discussions we will have. But again like I said public trust will be one of the key factors to move on with. On building the business case, so what we're also looking at is if statistical office have to invest in experiments, how can we make sure that this is the case for governments to invest in. We think that a number of these big data sources are very useful. There is a kind of a risk of course that it will not work. But we think that specifically for mobile phones, satellite imagery and social media, there's actually a good perspective on this. Here what I wanted to do on the terms of reference is giving you the strategic backgrounds about why would we as a statistical community want to work with big data. So one of the drivers of course, and we've discussed this and I was very happy to hear ITU also indicating the process which is going on at the United Nations. We have the POST 2015 agenda, which is the follow-up of the MDGs. And where the goals and the targets are being set. Referring to the discussion earlier today, in the goals, so there are 17 goals. And as was mentioned, there are quite a number of targets, like 169 targets I think at this point. Some of them maybe not explicitly referring to ICT, but implicitly. And I think when the indicators, which is a little difficult. So you have a targets which need to be monitored. So we need to develop indicators to do that monitoring. I think there is for this community quite a space to bring in ICT indicators at specific paths of the development agenda. And I'll just make some notes on that. Is that there are goals which initially might not look like ICT is a big part of it. But I think ICT as an enabler, especially like for energy, for clean energy, for other kinds of technological innovations. I think ICT should be part of that. And I think this community should try to see if certain indicators could be put in. On the data revolution, Ms. Lawler described that a little bit. I wanted to point out on the data revolution a few things. One of that was the global consensus on principles, which was, so they call it global consensus on data. There was reference made to that. I think that's one of the things where we will work on. There will also be for the UN a global partnership on sustainable development data. And there we will have coming year already a world forum on sustainable data. This is one of the proposals which will take place. We don't know yet where, probably in New York. There will also be a global users forum for data for the SDGs. And there will be much more work on brokering global public private partnerships. So that is the background for which then this UN global working group was put together. They will work in general on big data for official statistics. There will be a specific focus on how this big data could be used for monitoring targets for developing indicators for the post-2015 agenda. Yes? The global working group will need to develop a strategic vision for this. We'll look specifically on practical users. We work with the countries to do this. Within this group, we have a number of Gulf countries, Qatar, Oman, UAE, who want to do some pilot projects in the coming years on big data. And we will do that. And we will probably have the next global conference on big data in Oman next year. That was the data revolution. So within the global working group, and this is the last I want to say on this, we're going to focus on eight streams of work. One will be on training skills and capacity building since there was special skills needed, which are currently maybe not available in statistical offices. We want to link the big data to the SDGs. We will have a group specifically working on that. When we have indicators, when we know which kind of indicators will be there, how can big data help? We will have a group doing advocacy and communication to explain what big data is and what it can and what it cannot do, and also for building public trust. It's like, how can big data be used for public good? Then we will have the mobile phone use, social media, and satellite imagery as specific big data sources we will put attention to. We'll have access in partnerships because access to data is a big issue to building collaboration with the private sector, and we hope also that ITU can help us fostering those kinds of partnerships. And then finally, we will look at some cross-cutting issues on classifications of big data sources, frameworks, and taxonomy. I think I'm running out of time, so I cannot really say very much about ongoing projects. There are about 57 projects currently going on in the world. Half of them are in an ongoing or almost completed stage. The other ones are being planned. But here are the topics, which you could see here, and this will be also on the web, I think. There's not much going on, and the ICT indicators yet, so there might be some room for improvement. And we're working with all kinds of partners in this area. This is not something that a statistical community can do by itself. It has to work with the private sector. It has to work with academia and with others to move forward. Thank you very much. Ronald, thank you very much. We have plenty of time for questions to ask you more about how you get those ICT indicators into the SDGs. But we've been hearing from Mr. Janssen about the need to collaborate between the public sector and the private sector to try and unlock some public goal, some public good. Well, I hope to be able to announce a surprise guest, ladies and gentlemen. Do we have with us, joining from Barcelona, Nuria Oliver, Scientific Director of Telefonica Research. Yes, I think we do. Can you hear us, Nuria? We can see you, but I don't think we can hear you very well. Can we see if we can do something about the sound, please, Thomas? Can you hear me at all? There was a very excellent catering from the people, so you can count on it. I think we can hear you a little better. Let's have another go. This is Nuria Oliver, the Scientific Director of Telefonica Research. Hello, Nuria. Hello, how are you? Today. We're delighted to see you, is what I can say. Nuria, we're still having some problems with the sound. I tell you what, why don't we, we'll ask one of our other participants to speak next, sort out the sound meantime, and then join you in about 10 minutes time. How about that? Hope you can hear us. We're trying to clean up your sound, so do bear with us. We appreciate your patience. We'll join you in a minute or two. So, we were talking about the private sector, collaboration, all that. With us we have, sitting at the very end, Pat Wu from Facebook, social media giant. In what way can Facebook help us with big data? Is Facebook willing to help us with big data here in the public sector? These are questions we can put to Mr. Wu. Delighted to have you with us. Thank you, everyone. I'm privileged to be able to have a chance to speak with you guys and share a little bit about what Facebook is doing with regard to leveraging big data with the goal of increasing internet connectivity around the world. Just to give a little background, what my plan is, I wanna share a little bit about the internet.org initiative that we launched roughly about a year ago, and talk a little bit about the specific initiatives, how we're looking at using data to make better decisions and monitor our progress over time, and continue to get better. As some background, many of you know, the mission of Facebook is to give people the power to share and to make the world more open and connected. Directly from that is the mission of internet.org, which is focused on bringing internet access to the remainder people in the world who are not yet online. And the reason we think this is so important is because we fundamentally believe that connectivity improves lives. It provides people access to information. It connects families. It connects friends. It gives people a chance to have their voice, find their voice, share their voice. All in all, it provides opportunity for people and their family. And the goal is to accelerate connectivity. One of the things that we did early on is we commissioned a study with Deloitte to understand what exactly is the social and economic benefits of expanding connectivity. And you'll see some stats on this page. And this is one of the first things that I wanted to talk about. There's fundamentally a challenge right now to understand what are the right metrics and right methodology to measure impact. We were able to extrapolate some of these statistics with Deloitte, but ultimately these are fundamentally based upon other abstractions. So one thing that we wanna work on is as we continue to seize this goal, how do we find the right indicators with all of you guys that are the right measures to understand both the social and the economic impact so that we know that we are making progress. We know that these are valuable, new jobs, bringing people out of poverty, mortality rate, giving access to education and opportunity. But what's the right way to measure it and truly know that we're making an impact? In addition, we very much focus on the segments of populations that we're looking to reach. We worked with McKinsey on this research and this is a little bit staggering what we were unable to do. We simply wanted to know a little bit more about the segments who were not online yet because fundamentally if you're gonna do any sort of business plan to increase connectivity you need to understand the segments because that helps devise the technology solutions, the business solutions that are actually sustainable. So we looked at the basic stats, right? Who's not online that lives in a rural area? Who's low income, illiterate? Age, gender, all these things matter. Urbanicity is a big factor when it comes to designing technical solutions. Income obviously is a big factor from everything on education level as it's correlated to, but also the ability to have a sustainable business model. A lot of these statistics, we weren't able to truly understand how does this vary based on people who are online or are not? And then we weren't able to find all the details to build the business cases to really tailor our solutions and work with our partners to resolve them. And this is another opportunity that we really wanna work with you guys on. How do we get better statistics to understand the populations who are not yet online so we can better tailor the work that we do together to bridge that gap? In addition, we talked about this a little bit earlier in this morning. Infrastructure is a prerequisite for anyone to come online. For us, one of the guiding statistics we have is looking at what Erickson publishes. The percentage of the population in the world that is under 2G coverage, 3G coverage, or 4G. This is helpful, right? Because knowing who lives under network coverage helps us understand where opportunities that we're focused more on increasing capacity or where our place is where there is no coverage and we need new R&D to expand coverage in a cost-effective manner. But even within these statistics, there's things that we wanna push even further. At strictly 2G 9.6 kilobits per second, that does not support an internet connection. So where is that range when we actually say who fundamentally lives underneath an internet signal so that they can benefit from all the different content and services that the internet can provide? Within internet.org, we primarily look at three major areas. First is partnering with mobile operators and I'll talk a little bit more about that in the coming slides. Second is connectivity labs and the last one that I won't touch upon is the Alliance and that's really the broader collaboration inclusive of research, partnerships, and on-the-ground projects with the broader constellation of life-minded people who are interested in advancing connectivity. Specifically with regard to our partnerships with mobile partners, I think some of you I know that we launched free basic services in Zambia, in Tanzania, and recently in Kenya. And the purpose behind that is we believe that there's a set of essential services that people just need access to and that by having access to it, they'll not only learn what the internet is but they'll have the opportunity to advance their lives and potentially they'll find more value into it and be able to become onboarded to other things beyond the basic services. Currently we have services in five different categories, information, communications, health, education, and finance. One of the things as we were looking into it is in each country, we fundamentally wanted to work with the people within the country to understand what is actually relevant based upon what they're currently using, based upon needs and wants. In addition, we had to know what were the services and content providers within the location that had quality information, quality services that were in high demand. This is also another challenge because as we looked out in the information that was out there, there's incomplete information on what's currently online, incomplete information on what services are localized or in languages that are relevant to the people who live within a given location. And we had to almost go from scratch and work with local and international players to even come up with a set of services that made sense and will continue to iterate over time. This is missing some axes, but as we talk about network access and infrastructure, one thing that's important is, if you just apologize that it's missing the y-axes and the x-axes, but on the y-axes, it's the percentage of population in the world, and on the x-axes, it's population density. And as you know, when you look across, there's very, on the far left, is the rural settings where there's very low population density, very small percent of the population. However, if the goal is to bring everyone online, they need to have network access. And as you continue on to the middle, that's more of the suburban areas where currently we are expanding towers and we're expanding fiber to go ahead and bring individuals online. And on the further right is where there's much higher population density, and we're advancing technology to increase capacity to be able to serve and bring better services. For us, one of the things here is, because a precondition of bringing people online is network connectivity. So as we explore solutions on the terrestrial, on high altitude platforms and satellites, it's very important for us to know information based on the ground, where is existing coverage? Where is fiber that makes the most logical sense to cost effectively increased coverage? And as we think about other solutions that would go non-terrestrial, where are opportunities where that is necessary? That's more cost effective because of the topography, because of the income level of the individuals, because of all the different ramifications that require alternative solutions. So again, this is some place that we've tried to collect the best data that's possible, but understanding a lot more with regard to the demographic information, with regard to existing coverage, with regard to existing fiber, would really help us better understand and tailor the solutions that we go after. In closing, I wanna say again, we believe that connectivity provides opportunity, and our goal is to really get the best data to inform the strategy to increase connectivity. In addition, how do we increase the impact that comes from understanding more what actually is beneficial, both economically and socially, when we increase connectivity? Thank you for the opportunity to share. Thank you very much, Pat. And perhaps later we can find out a bit more about the initiatives in Zambia Tanzania and Kenya. I'm gonna ask my friend Thomas in the far end of the room, if we've had any better luck in hooking up with Nuri Oliver, Telefonica. What do you say, Thomas? Shall we try again? Nuri, we can see your face, but can we hear you? Hello. Oh, wonderful. We can hear you loud and clear. So I'm going to hand over to you because we've been hearing an awful lot about how the private sector and the public sector might be able to work together to unlock some public goods. And I think you're gonna tell us a little bit about possible public health implications. Yes? Yes. Okay. Yes. It's all yours, Nuriya. Thank you. Well, again, thank you so much for inviting me. I'm sorry I couldn't be there in person, but thanks to technology, I can be there virtually. My name is Nuri Oliver, and I'm scientific director at Telefonica, and I'm going to be presenting to you some of the work that we have been doing in the past five, six years in the context of what we call Big Data for Social Good. So I don't have to really give a lot of details about this because you're all very knowledgeable about the fact that the mobile phone is the most pervasive piece of technology ever invented in human history, and we spend more time actually on our mobile phones than watching TV or with our own partners. And what is really interesting is that this is a phenomenon which is global. It happens both in emerging and in developed nations. So the fact that most of us have a mobile phone and the fact that the phone is connected is what has led to the concept of using mobile phones as sensors of human activity. And in fact, last year, MIT Technology Review highlighted this area of inferring human behavior from mobile phones as one of the breakthrough technologies of 2013. So the main question that we have been trying to answer in my research team for the past five, six years is how can we use aggregated and large-scale mobile digital footprints to understand aspects of human behavior that could help make the world a better place and have positive impact. And just to give you an idea of the kind of data we're talking about because obviously privacy is one potential concern. For many of these projects, we are using very basic data such as the levels of activity in the cell towers of the aggregated patterns of mobility between cell towers. So this very first video shows the levels of activity in the cell towers right before, during, and after an earthquake took place in Mexico. And you can very easily see how the levels of activity suddenly increase right after the earthquake takes place. So just by looking at that sensor, we can have an idea of how many people are in different parts of the space. This second video shows the mobility patterns of the mobile phones in the UK. And again, at a very, very aggregated and high level, you can still see what are the main cities and the movements between the cities. So using this kind of data, I'm just gonna very quickly share with you one exemplary project that we did three years ago and that has become very relevant today in the context of the Ebola outbreak. And it was a project that we did with respect to the H1N1 flu outbreak that happened in Mexico in 2009. So I'm just gonna give you a quick overview of what happened at the time and how we could use big mobile data to help answer two specific questions that had remained unanswered. As you might remember, the flu outbreak started in April of 2009, and in order to contain the potential pandemic that could happen, the Mexican government took a number of measures. The first alert level of the first measure was just a medical alert. It was an intervention. And the Mexican government asked citizens to be cautious and stay home because there was a risk of a pandemic. However, the number of cases continued increasing over the next days and the Mexican government raised the level of alert to a second level and they actually did an intervention which involved closing schools and universities, tourist places, churches, et cetera. And this closure happened for a few days. However, the number of cases continued increasing and in fact, the World Health Organization raised the level of alert to the maximum level indicating that a pandemic was imminent and in view of the big global pressure that there was in Mexico, the Mexican government decided to take an unprecedented measure which was shutting down the country for five days, as you might remember, from May 1st to May 5th. And this meant shutting all economic activity except for police, firemen, and hospitals. However, about a month later, the World Health Organization declared that a flu pandemic was underway. The first one in the 21st century. On the one hand, the CDC and the World Health Organization traced the Mexican government for the measures that they took. But on the other hand, the pandemic took place. So there were a couple of questions floating in the air which were, did the measures actually work? Did the measures manage to reduce the mobility of the population and hence reduce the progression of the disease or not? And these are the questions that we answered using aggregated mobile data. So to answer the first question, whether the mobility was reduced or not during each of the measures, what we did was characterize the mobility during a baseline period and then during age of the three alert periods. And we have some surprising findings. The first one was that during the medical alert, there was a 0% of reduction, significant reduction of mobility in the population. So it seemed that people continued with their lives despite the alert. The second finding was very surprising and it was that during the second level of alert, 80% of the population significantly reduced their mobility whereas only 55% of the population significantly reduced their mobility during the third level of alert. So these findings suggest that closing schools and universities and working places during working days is more effective than shutting down an entire country during a holiday period, which was what happened in Mexico. Given that the mobility was reduced in the population, the next question that we answered was, what was the impact that this reduced mobility had on the progression of the disease? To be able to answer such a question, we developed two epidemiological models, one for a population with the normal mobility, that is the mobility without any intervention and another model with the reduced mobility that took place because of the interventions. And what you see here, the orange curve is the percentage of infected agents in our model when there was no intervention by the government. And the red curve shows the percentage of infected agents with the reduced mobility thanks to the interventions. And what we can see is that there were about 10% fewer infected agents and the peak of the infection to place 40 hours later, which in the context of a pandemic is a lot of time gained to mobilize doctors and medications and call for help, et cetera. So this is just an exemplary project of how mobile data can help in the context of public health. But what is very clear and it has been presented before is that if we want to be able to use big data to have social impact, we do need to partner with the institutions and the governments that know the realities of the countries where we want to have an impact. We have right now collaborations with the United Nations Global Pulse, with the data pop alliance, with MIT and with the government of Mexico to see how we can help address some of the challenges that they are facing in the context of not only public health, but also social good. These are just some references to papers. So this particular project of the flu has become very relevant because of the Ebola outbreak. And I just wrote a recent piece in TechCrunch that you can read about how we could use mobile data to help Ebola. Because the reality is that a few months after the Ebola outbreak, we haven't been able yet to really exploit the power of big data to help in the context of the outbreak. And it has really made it clear that there are a number of challenges that we need to address. Just to exemplify with the case of Ebola, I recommend you to read this Economist article by Ken Cooke here, or the situation that has been happening in terms of the inability to be able to use mobile data to help understand the spread of Ebola. So some of the challenges that I think we would need to address collectively are in three categories. First, there are regulatory challenges. I think existing regulation doesn't really contemplate well the use of large-scale, aggregated data for humanitarian purposes. And I think we would need to really update the regulation and also define clear guidelines regarding safe data handling, processing, and sharing for humanitarian purposes. There are also technical challenges because obviously the data is not perfect. There are a lot of issues about how representative the data is, how it can generalize, how we can combine data from different sources. Of course, we're still far in many countries from being able to do real-time analysis and prediction. And very importantly, in many cases, we don't have ground truth, so we need to do interventions to be able to validate the models that we built. And finally, there are obviously privacy challenges that we need to address. Potential privacy risks need to be minimized and need to be understood, and we need to define very clear code of conducts and ethical principles that everyone dealing with this data should follow. But at the same time, I think there are many opportunities as my team and other research teams are showing. We are showing that we can use aggregated and anonymized mobile data to understand human behavior, to characterize mobility, and to help make better decisions in urban planning, crisis management, or global health. So my question to you is, what can we all do to responsibly turn this opportunity into a reality? So when the next Ebola outbreak or any other outbreak happens, or when the next emergency happens, we can very quickly take advantage of the large amount of data that there is to help make better decisions and help save lives. Thank you. Noia, that was very inspiring. I'm not going to ask the audience to put their questions to you right away. Instead, I'm gonna ask you to please stay with us on the line, and then we'll have a general question and answer session at the very end. So please do stay with us and listen to the rest of our presentations. The next one up is very interesting from Statistics Netherlands. Let me introduce you to Gerrit Wassink, team manager of the Culture, Tourism, and Technology Department. And as Ronald Janssen said to us earlier, the Netherlands has really taken a lead in this area. Ms. Wassink. Thank you. Okay. Thank you. Well, I feel honored taking part in this session. My name is Gerrit Wassink, and I'm the manager of Culture, Tourism, and Technology Statistics at Statistics Netherlands. And in the 10 minutes I have, I'm not going to surprise you with nice graphics. There's no time for that, unfortunately. My contribution to this panel involves a number of practical observations I made since we started in 2009 at Statistics Netherlands to search for new data sources for statistics. So, questions I will shortly address are why did we start? What did we do and what did we learn? So, why did we start? Well, some five years ago, using the Internet as a data source was seen as a way to reduce the administrative burdens on companies caused by traditional questionnaires. And at the start, we were funded by the Ministry of Economic Affairs. People from the Ministry thought that it would be a good thing for statistical officers to be a participant in this new type of research. And, of course, we agreed. We were willing to find out what this was all about and were very grateful for the investments by the Ministry. So, what did we do? We have written some reports on what we did where you can find detailed information and nice graphics. And you can find links in my presentation that I hope can be made available. And there is also a paper available on the website on mobile phone data as big data. But to summarize, and not to be complete because, unfortunately, there's nothing about social media. But, well, I have the following things to say. We experimented with the collection of prices for airline tickets and petrol with the aid of internet robots. Although the experiment in itself was a success, it was discontinued because the costs of the development and deployment of internet robots were much higher than the costs of the manual collection of a few prices each month. We also experimented with internet robots to collect information about job vacancies posted on the internet. Unfortunately, it did not yield any satisfactory results. In spite of the high volume of internet jobs' vacancies, the representativeness of these data appears to be rather poor and we were not able to find methods to properly correct for this. We also looked at internet speed measurement using a program installed on the computer of members of a panel. We worked together with the company that was already doing this. Though the panel reached some 100,000 members and almost 16 million internet speed measurements, one of the most important questions for analysis concerned the representativeness of the panel. We did a lot of checks with traditional data, but in the end we had to conclude that we couldn't make it work and that the panel members were far from representatives. Representative, we found out that they were more the people that liked high speed, so not representative. We also worked together with the Technical University of Delft. They carried out a pilot for us. It involved the tracking of 130 volunteer smartphone users for one month on the basis of a research app installed on their mobile phones. The data logs obtained via the smartphones provided good insights into various aspects, such as the frequency and the duration of the use of mobile services. But again, questions about representativeness came up and we had to stop this experiment. We also studied millions of online ads about second hand goods from private individuals on the dominant Dutch website marketplace.nl. We paid for a data set with all these advertisements and the associated characteristics. The anticipated possible uses of this data were to produce an estimate of the sale of second hand goods among private individuals and to search for potential relationships between this data and our economic indicators, such as consumer confidence. We did not really succeed in this, but we did manage to create some insights into the regional distribution in the use of Markplace and the characteristics of the Markplace user. Still ongoing is our research on the use of mobile metadata, which are generated when a mobile phone communicates with a telecom provider, so-called call detail records. We received this data from Vodafone on the very strict conditions. It was supplied as aggregated data, and you can read more about first results in the paper on the website. Other ongoing work is a study of a data file about two million Dutch websites. This data file was given to us by a Dutch company and contains characteristics of these websites. Our hope is to develop a statistics about Dutch websites, but we are far from that yet, but we are hoping that we will succeed. We are usually mostly busy now with cleaning the data. To summarize, our experience over the last years with new data sources show that actual practice is wild. Sometimes at the start of a project, it seems that the benefits are ripe for the picking, while in actual practice, this often turns out to be somewhat disappointing. Usually, this is not due to technology, but much more due to the process required to ultimately to inform the raw data into sound statistics. This leads me to some lessons learned. So what did we learn? We should not so much try to see new data sources as a replacement of existing traditional sources. It should be an aim, but it is very difficult, if not impossible. New sources should be evaluated as giving additional information, partially about new phenomena, and topics that are already measured using traditional surveys. For example, by analyzing mobile phone data, we can follow tourists traveling through the country, and this is new information, but we still need traditional surveys to determine if the tourist is sleeping in a hotel or on the camping site. Working with new data sources is fancy and it is good public relations for statistical officers. But be aware, sometimes expectations are too ambitious. New data sources usually require new funding and rarely lead to the replacement of traditional sources. So it's not a solution for budget cuts. Privacy and methodology remain issues to be tackled. Technology is usually not an issue. I thought methodology is usually the reason for companies to cooperate with statistical officers, as they do not have data scientists on their payroll. Concern about privacy is sometimes the reason for companies not to cooperate with statistical officers. To my colleagues from other statistical officers, I would like to say, be open to initiatives offered by companies with interesting data. But don't pay anything before you know the potential of the data. Don't forget that companies will benefit from publicity that their data is that good that the statistical officers is using it for statistical purposes. Bring this in in the negotiations with these companies offering you big data. Big data also requires new types of visual information, new products like heat maps where every dot is an individual measurement. And we have to start discussing the use of better indicators, implicating indicators with poorer quality. This is important because statisticians are not used publishing results with poor quality. To conclude, big data, yes, it's there. Statistical officers can help in translating unstructured data into sound statistics. So I would really like to encourage other statistical officers to make an effort. At Statistics Netherlands, we see more and more possibilities to produce new interesting statistics. And especially telecom data are very promising. Using telecom data, we hope to be able to come in months to publish statistics about the spread of the population at daytime. This is very important for mobility issues and in case of calamities taking place. With traditional statistics, we were only able to make statistics about the nighttime population. So this is a big step forward. But what we shouldn't forget is to explain to the public what we are doing and the privacy issues are dealt with very well by statistical officers. This involves the identification of objectives, risks, and risk control measures. When we do this in a good way, I'm sure that companies with interesting big data will certainly be prepared to work together with statistical officers. Thank you. Thank you very much, Gerrit Bussing, from the Oxford Internet Institute. And he too is going to put big data into some kind of a realistic context for us. It is not the panacea for all ills and it does not necessarily help in bridging the digital divide. I should also say to you that Mark is a geographer. It's all yours. Thank you very much. Excellencies, distinguished guests. It's an honor to be here today. I just want to dive in and the place that I want to start is with the idea that frames a lot of what we're talking about here. The notion that ICTs can enable radically different patterns of knowledge use and knowledge production around the world. And these are ideas that we can empirically map and measure. And that's what I'll be doing in the next few minutes. Using some big data sources to show you some data on the geographies of codified digitized knowledge that provide an indication of who is and who isn't represented and participating online. Now, when I talk about this topic, I normally come armed with a slide deck full of maps on older or historical patterns of some of these uneven geographies of knowledge production. I was told to keep this talk very short, so I'm not doing that. But you'll have to trust me that there are massively uneven geographies of information and knowledge in the world. And this in many ways happened because of traditional digital divides until recently some parts of the world just didn't have the right infrastructure and so remain separated from the flows and the clusters of information production that are essential to thrive in our global knowledge economy. But something's recently changed, right? Fiber up to cables have been rapidly rolled out around the world. There are now almost 3 billion people online, most of whom live in low income countries. And this has led to hopes that, for instance, all the world's citizens will have the potential to access unlimited knowledge, to contribute to, and to enjoy the benefits of the knowledge society. And so we're seeing hints that there are some changing practices of engagement with the knowledge economy in poorer parts of the world. So just, for instance, there are somewhere close to 100 so-called technology innovation hubs in Africa alone. And these places, like Nairobi's iHub, shown here, they harness ICTs and connectivity and people's entrepreneurial spirit to attempt to change some of these uneven geographies of digital knowledge production. But there are worries that this isn't the main story that's going to be told. This is a cartoon from a World Bank report on, let's think about not just these innovation labs and engagement in this high-end knowledge economy, but also the ways that places are being enrolled to do some relatively low-skilled work through what's been called micro-tasking or micro-work. And this involves people doing very simple tasks like digitizing medical records or classifying images of being paid per click or per task. And research shows that this is, our research shows this is often an important source of income to millions of people, something, interestingly, that often goes under the radar of official statistics. But many people are also starting to refer to it as something relatively or potentially exploitative, something that might create what people are referring to as digital sweatshops. So I see all of this as the starting point for some questions that we can ask about which places are positioning themselves as hubs and which are positioning themselves as peripheries in the knowledge economy. And I'd invite you then to look at a series of maps that I'm about to show you that my team and I made and let me show you these to you now. So a sensible place to start is just by looking at what's actually on the internet. And that's what you're looking at here. This is a map of every domain name and where it's registered. It's a cartogram, so each circle is a country and the size of each country is shaded according to the number of domains registered in the country. It's the 260 or so million registered domain names. The shading of the country indicates, you can't actually see the shading because the contrast is all washed out here. If you were to be able to see the shading, what it would indicate would be the internet population of the country. So just as an example, a dot like India's one, which is, where is India? Over here. That would show us that there's a lot of internet users in the country. India is the world's third largest internet population but very, very relatively few websites. We see that same pattern in a lot of Asia. If we're to look at Africa and South America in contrast, the sort of clusters here and here, we would see relatively small numbers both internet users and internet domains. Let me show you something a bit different though. I wonder if the people on the AV board over there can increase the contrast of the slides. What you're looking at here, let me try and explain it to you, is where these new layers of information on the internet are and we can do that by looking at the internet gateway that most of us use. So this is a map of Google Maps. It's a measure of online content that people are creating about anywhere on the earth and then it gets indexed by Google Maps. So we created about half a million sample points on earth and each one of those sample points. We ran Google queries to get a sense of how much content's created about those places. And a red shade on this map means there's a lot of content about that place. A darker shade means there's almost nothing about that place and you get a sense of this massive unevenness and these layers of information that surround us. It may surprise you to hear there's more indexed content layered over the Tokyo Yokohama metropolitan region than the whole continent of Africa put together. Now this map though, it's not particularly useful because it simply shows you content in all languages. So let me show you something a bit different. This is also a map of Google Maps. It's what Google Maps knows about the world, the amount of content Google knows about the world but divided into languages. So a blue dot means there's more French content about that place. A red dot means there's more English content about the place. You're looking at Eastern Canada and then you see kind of what you'd expect to see. You see more English language content about English speaking Ontario and more French language content about French speaking Quebec. We can do this all over the world. So this is Belgium for instance. We made an orange dot mean there means there's more Google index content in Flemish and blue means there's more content in French. And again you see kind of what you'd expect to see. You see more Flemish content in Flemish speaking Flanders, more French content in French speaking Wallonia. What about somewhere else though? What about a part of the world where there are more let's say unbalanced power dynamics between different linguistic groups, the Middle East for instance. So this is a map of Arabic and Hebrew content in Israel and the Palestinian Territories. So a blue dot in an honest map means that there's more Hebrew content. A red dot means there's more Arabic content. And what this all shows us is while Arabic and Hebrew content tends to annotate the same physical places, there's a much denser cloud of Hebrew content over almost all of those places. So the point of this is that there's not only a paucity of online information about a lot of the world, but of that information that exists, a lot of it's just not accessible to a lot of people. What about more explicitly user-generated content online? I think this gives us a better sense of who is and who isn't creating content, participating and sharing content. Let me give you a few examples. So what you're halfway looking at here is a map of OpenStreetMap. These are all of the hundreds of millions of contributions that are submitted to OpenStreetMap, the world's largest collaborative mapping platform. And basically what we see if we look at that is almost two-thirds of content is created about just five countries in the world. And we can see something similar if we look at all sorts of other platforms. This is Panoramio, for instance, one of the world's largest photo sharing services. Despite there being tens of millions of photographs that we collected here, we again see this massively concentrated geography of information. I won't go into much detail on this one, but just one random fact here is that there's more content here about Italy than all of Africa put together. And I could show you a whole range of maps of user-generated content. This is content of Flickr, for instance. I won't explain this one. This is just a map of Tweets. But I want to focus on one thing in particular. This is a map of Wikipedia. Let me explain why I want to talk about some of these patterns of Wikipedia. Wikipedia is by far the world's largest and most used in psychopedia. And you're looking at a cartogram again. So each block is a country, and the size of each block is sized according to the number of Wikipedia articles about that country. And the big rectangle on the left, that is North America, the giant block in the middle, that's Europe, the top right, that's all of Asia. And then in the bottom left, you're looking at both South America and Africa and those small rectangles on the left. And so again, it's this relative absence of South America and Africa that's again really notable here. There are more articles in Wikipedia about the Netherlands or Poland or the Ukraine than the whole continents of South America and Africa. Maybe even more shocking is the fact, what I didn't tell you, is that this block is Antarctica. There are more articles written about Antarctica than many countries in South America and Africa. Maybe even more surprising than that is that there are more articles written about places that don't even exist like Tolkien's Middle Earth than many countries on our planet. So you might think that a lot of this unevenness here can be explained by unevenness in internet penetration rates, right? That makes sense, but that's not entirely true. So we made this cartogram of internet penetration using ITU data, countries of size according to their internet population. So you see China being the biggest country on this map because it's got the world's largest internet population. And then the countries are shaded according to their internet penetration. So if you get a dark shade like Korea or the UK, it means the vast majority of people are online in that country. So we see there are obviously large inequalities in internet access around the world, but those uneven geographies of access still don't explain all of this unevenness in a platform like Wikipedia that we see here. Now I just wanna quickly expand on this high degree of visibility that's afforded to Europe in particular on this platform. So in this next series of maps that I'm gonna show you, what we did was we mapped every single article in Wikipedia about a human being, about a person. So what you're looking at here is a map of everybody in Wikipedia that was alive in the 15th century. This is everyone in the 16th century, 17th century, 18th century, 19th century, and 20th century. So what you saw here was an incredibly Eurocentric history on a platform that's open to anybody with an internet connection. Now place all of this within one of the motos of the platform, that's to contain the sum of human knowledge, and basically you see it's a dangerous idea to imagine that we're even getting close to have anything like this. And I think the point of this is that we need to keep a focus on some of these significant biases that are embedded into some of this knowledge that plays a key role in shaping our understandings of the world. So because of this, we're asking not just where these Wikipedia articles are but who's writing them, who can access them, what this tells us about global patterns of visibility and voice. I'll leave this map to save time but the very short version without really explaining what you're looking at is that in Europe you get lots of content written in local languages but you only see that in Europe. Every URL in the world there's far more stuff written in European languages. So for instance, the only Arabic-speaking country where there's more Arabic content than French or English content is Syria. Everywhere else has more English or French content. So a pattern of high income countries self-defining and low income countries largely being defined by others. And then because we can figure out where these articles are described in and we can figure out where the edits to them come from, we can ask what I think is the most interesting question here. What percentage of content about any given place come from local people? So this map took a lot of analytical work. We had to look at every edit of every article but the dark shade countries on this map are ones where most edits about a place, most stuff written about the place is done by local people. The light shade, so that the yellow shades that you're looking at, that means fewer than 5% of content written about that place are from local people. So the yellow countries on the map mean that hardly anything written about those places comes from locals. That it mostly comes from outsiders. I'm gonna show you one very final chart. This one shows you which part of the world are writing about each other in Wikipedia. So each region on this graph is given a color and the percentage of local edits is listed under the region name. So if you look at Europe and the number 75, you see that in Europe 75% of edits coming from Europe stay within Europe. They're used to write about Europe. 73% of edits from North America are used to write about North America. But then look at the Middle East at the very top of this graph. It's only 36%. And then the colored lines on this graph, they show you where the edits from each region go to or are about. So the red lines, for instance, that's the content coming from the Middle East just to stick with this example. So you can see that if you look at the Middle East, there's not only what we've seen so far is there's not only many articles about the region. There's even fewer articles in local languages. We've seen that not only there are very, very few editors, but even of the edits that exist from this place, most of them are flowing out to write about other parts of the world that already have rich amounts of data and content about them. So these global informational cores are exerting some sort of informational magnetism, the presence of information, creating a virtuous cycle of informational richness, the absence of it being part of a more vicious cycle of informational poverty. So I think the question then is why? Why when the world's getting wired, when internet penetration rates are rising rapidly, why are there still these massive absences? Why have we seen a reinforcement of these global patterns of visibility and representation and voice that we're used to in an older pre-digital world if I had more time, I'd give you examples from some of our research that illustrate how barriers to participation in the global knowledge economy are often about much more than just simple internet connectivity. But I won't do that. Instead, I'll very quickly sum up some of them by pointing to the fact that these digital divides that we're seeing be reproduced can't just be explained away by a lack of connectivity. Connectivity's unnecessary, but not a sufficient condition. So are things like a broader ecosystem of information and educated and tech-literate population having reliable infrastructure, not excluding half of the population? In other words, women having the internet be trusted rather than highly surveilled and having the critical mass for local language tools and platforms and communities. We often forget about a lot of these things and the enthusiasm about connecting the disconnected, which is only ever the first step to achieving some of these goals related to engagement with a knowledge economy that we often hope to achieve. So what this also points to is the fact that we need to be extremely careful when working with some of these social big data sets if we're trying to use them to tell us something about broader social and economic and political and environmental trends. Big data has uneven geographies and the insights that we get from the data or the methods that we use with the data in one context or place aren't necessarily transferable to another. So maybe I'll just end by saying that the first thing that we should probably ask in our enthusiasm about new data sources for new insights is not just what do these new data sets tell us about the world, but what don't they and who, what, and where do they leave out? Thank you. Thank you very much, Ma. You've given us an awful lot to digest there. Fascinating geographies of inequality and real eye opener. Thank you very much. Nuri, I hope you're still with us. I'm about to throw the floor open for questions to the audience. I just want to remember that we have Nuri or Oliver right there at the bottom of our screen in Barcelona who's also open to answering your questions. Who would like to, I think the lady from Japan, please do just hit the button in front of you, the mic in front of you. Tell us who you are. And if it's a general question or if it's aimed at someone in particular. Okay. Merci madame pour le présent, à la présente. Je sais à tout ça qui, merci les présentations marbeilleuses. J'ai un question. Peux, troisième question. Premier, je crois que le gouvernement est nécessaire pour les autres, les problèmes en association avec l'utilisation des big data. Par exemple, le manipulation des données personnelles, etc. Et puis, nous avons besoin de développement de ressources humaines qui est la connaissance des statistiques nécessaires pour l'utiliser des, encore, le big data. Particulièrement, je vais te raconter les opinions du secteur pour les bays, si accepté. Monsieur Patifu. Thank you, madame. Thank you very much for the interesting presentation of all participants. But, I mean, however, I have a question to the keynote speaker, Mr. Reich. As you mentioned that big data offers huge possibilities in many sectors, even in all ones. That's why my question is how to make use or to utilize big data efficiently and effectively in order or to be able to do exponential changes in the sector. Did you get it, I guess? Should I answer it a bit? Should I? Should I repeat? Thank you very much for your question, first. Well, in my opinion, if you want to derive data from which sector or whatever you want to derive data from, you need to have a good understanding of the ecosystem and you need to know what are the main actors and where is data stored. And if you want to have information, for example, on water or on gas or on meat or whatever, you need to know the ecosystem. You need to know who are the most important parties. You need to know who is gathering information. So the complex answer is that there are no easy answers and it depends from sector to sector for ecosystem to ecosystem. But each ecosystem, if you study it, will have some of the nexuses where you can plug in and you can derive certain indicators. And to finally add to that, I think just what Minister from the CBS also said, that it's often, it is not a substitution for current indicators, but it can offer all types of new insights and more a push approach than a pull approach. On that question, from what I could make out from what Mark Graham was saying, if you focus too much on nexuses, you could very well end up losing a lot of rich data because you end up disenfranchising people's voices. Is that a fair comment, Mark? No, I think you should add some more to that thought. Okay, thanks. Maybe I should just build on one thing. You sort of ended with this idea that big data can offer huge possibilities for everybody, right? And maybe, I don't necessarily disagree with that, but what I might add to that is well to build in one of your, the thing you started with, which is this idea that there could be economies of scale here, so it's not that it doesn't offer huge possibilities for everyone, it's that it offers huge possibilities for some people more than it does to other people, and maybe that's something crucial to remember when we're thinking about how we invest scarce resources. And could I put a variation on that question to Nuria Oliver in Barcelona? You were talking about the potential to possibly used mobile telephony data to try and track the Ebola epidemics better, but in countries like Guinea, for instance, where the networks are not very strong, I imagine, isn't that a limitation, therefore? Aren't we, isn't that a limitation to the use of mobile data in this way? Yes, well, actually, I would say it's a power because if there is any technology being used, it's probably a mobile phone, more than social media, obviously, or any other kind of technology. So I agree with you, and I mentioned it as one of the challenges on the technical side. The representativeness of the data is something to obviously keep in mind, and it's not the same to make inferences about human mobility or human levels of activity in a country where maybe only 80% of population or 50% of population have a mobile phone versus a country where 100% of population has a mobile phone. So that's definitely a factor to take into account, but I think something that we need to understand is that if there is any passively collected big data that we could leverage in these situations, it's actually the mobile data because mobile phones are truly pervasive worldwide. Okay, thank you very much, Nuria. Ronald, please. Yes, I would like to come in here also. One of the, let's say, complaints with monitoring the targets under the development goals was just either very untimely data, so the very old data or lack of data at all. So I think what we're looking at here is we try to get data on issues and then add on with big data. And especially also I agree with Nuria that mobile phone data, given its penetration in developing countries, is something which we really would like to look at as something that will add on to it. Also because of the higher frequency. So you need a basic data of official statistics from other sources, but then mobile phone data or other big data can add on to it so that you have more frequent data because policy, you can't base policy on two data points. You would have policy, you want to have it on frequent data and that's where I think mobile phone data or other kinds of big data can help a lot. Thank you. Okay, thank you very much. I think we've got another question in the audience and I think our speaker's going to be speaking in Spanish so we're going to try our interpretation service and say to everybody, please get your headsets on if you don't already have them on, sir. Hi, my name is Rui. I go to spelling Spanish. Hoy es un importante para nosotros de Cabo Verde. Precisamente estamos hablando de datos, de estatísticas y estamos hablando de la utilización de la tecnología, nomeadamente el telemóvil. Digo que es importante porque en ese momento, desde ayer hasta hoy, está ocurriendo una erupción volcánica en nuestro país. Y ahora hace cosas de 30 minutos pude consultar que se está utilizando la tecnología móvil precisamente para atender a la población que vive cerca en ese lugar. Las comunicaciones a través del Facebook también están fluyendo de forma muy considerable y la gente está masivamente adaptando a ese nuevo acontecimiento que hace 25 años ocurrió, pero de forma diferente porque hoy tenemos tecnologías implementadas que permite que toda la población pueda tener acceso y que más de la mayoría tenemos 97% datos de 2003, más o menos 97% de acceso de la población. So I have to interrupt you to say that I didn't get any of that. So I don't know if the rest of the hall is doing better with the translations. I feel very bad, we've been talking about not disenfranchising large chunks of the world and we're doing exactly that in this hall, so I do apologize to you. But we're really struggling to get the translations together. So let's see if we can extend this session a little bit and come back to it. I do apologize to you both. We're not communicating well. Would you like to ask the question? Dr. Kim? Yes. Let me ask Mr. Woznik, Netherlands statistics. In Korea, my lady president proposed creative economy based on ICT, especially promote big data and IoT. And then we have achieved much progress in IoT. But as you mentioned, we have big dilemma in big data because of privacy. Definitely protection of personal information is very important. And when you go on researches, there is any descent from the NGOs, any other peoples, to protect privacy, something like that. Do you have any idea or do you have any alternative and experience about that? I'll take those questions. OK, we'll put it to you too, Noria, in a second. Well, on questions on privacy, usually the statistical officers are more into that than the people that are being monitored. So at Statistics National, we have strict measures for reporting about, well, every research we conduct with this where there is a possible problem with privacy. So we are more strict by ourselves in doing that than the people that are being monitored. It's not that big an issue for them. But we have not had any trouble or something like that. But we are looking at it ourselves very well. OK, Noria Oliver? Yes, as I mentioned on my presentation, definitely privacy is a potential concern. But I really want to highlight that completely anonymized and aggregated information such as the levels of activity at the cell tower period can provide extremely valuable information that we don't have right now about human activity and about density of people in different areas. So while obviously it is critical to take all precautions possible and have very clear code of conducts and ethical principles when dealing with personal data, it is also very important to realize that there is large-scale, fully anonymized and fully privacy-preserving data that can give huge value for public health or other purposes that we are not really using right now and we could be using. OK, thank you. Mark Rowe? I just want to add that it's often very easy to slurp downloads of data and then start thinking about what we do about privacy and that it makes a lot more sense for everybody if we have that question and discussion about privacy before even touching the data. And then we know what exactly it is we want to collect, analyze, and share. So the two things that I'll say is obviously at Facebook, we take privacy very seriously. But as we approach it when we work with people, it's really about what goal you're trying to strive for. So if it's about the Department of Tourism, it's many countries have put a lot of time in putting into their Facebook page. And there's a lot of content that's very much open to help people discover the country. And even it's interesting for Korea, just how the soap operas have driven people to visit certain places. So there's just different ways by which people have been very public about what they're sharing. And in turn, people from different countries have been also very public about what they're interested in, where they've gone, what they've tagged, and what they've shown. So there's just pockets where, at least from our platform, because you get to choose your level of privacy and who gets to see the information. There has been wonderful opportunities for us to work with different departments of tourism to figure out what has worked. And this is really the power of the individual has already chosen their privacy settings. That was a positive point of view, but there are still huge trust issues, I think, that you face. Let's have another question from the gentleman from Bangladesh. I'm Sami Kobid from Bureau of Statistics of Bangladesh. And my point is, whenever any survey is conducted, at the same time, government conducts a survey and many private organizations or non-government organizations conduct the same survey. And whenever it's a large-scale survey across the whole country, in that case, in many cases, the outcome varies. Suppose government outcome is one, and the private organization survey that has produced different results. And sometimes the inconsistency becomes very high. Especially, it happens in our country, in some cases, government and non-government survey result. So these big data are intended to facilitate the budgeting and planning of the state level. Based on these data, the budget is made, and as per the budget, we execute the development works. So in that case, in terms of inconsistency, how can we minimize this much inconsistency and bring a balance? That is my point. Thank you. Who'd like to respond to that? Ronald Janssen? It's a challenge. Is it? Which one is on? This one's on. Thank you. Oh, yeah, this one's on. I did not fully understand. So you're posing here a methodological question, but I don't know exactly where the big data came in. I heard about your surveys having inconsistency. So that's a kind of methodological question which is of a more general issue. So I don't know if you use or if you intend to use the big data to improve surveys. So in that sense, I have difficulties with. I can, and I just can follow up a little bit with what was said previously, big data can supplement existing surveys. Something that I think definitely is a role which should be used and how it can be used. And then another one, and then I come just back to the previous discussion which I wanted to add in. I think we should work as a statistical community much more with private sector like Facebook and so forth to make apps available. I think if you want to avoid some of the consent issues with privacy, if there is an app provided by big private sector companies like Facebook or others with which you can access data for specific purposes like tourism or others, I think that would help a lot and also in building up public trust for using data in this area. Thank you. Ronald, thank you very much. Ladies and gentlemen, excellencies, I have to announce that we're out of time. I do apologize for the technical issues we've had, especially the translations to our intervention from Japan and also from Kabul-Berdei. I do apologize to you that we weren't able to include your questions to our panelists. Let's give a round of applause to our extremely helpful panelists. Nuria in Barcelona, Pat Wu, Redbrunner at Mark Graham, Herit Vassink and Ronald Janssen. And with that, I hand back to the chair, Mr. Kashibatje. Thank you. Thank you, Ms. Pillai. It's a great honor to me to be the co-chairman of 12 World Telecommunication ICT Indicators Symposium. I'd like to thank Ms. Pillai for moderating our session. Thanks, panelists, for your interesting presentation. Now we have a coffee break for half an hour and in half an hour we will continue our sessions. Also, we have a group photo and please use this opportunity to make group photo together for the history. Thank you very much and thank you, thank you all participants for the conference. Thank you. Thank you. Group photo will be in the lobby near the registration desk. Thank you. Herit, he's on over. Matt Margos, they're on the spot. Yes. You're from Pesicium also. Herit, you're also from Pesicium because you're from Pesicium. I'm doing good. He's in Malaysia, I don't know. That's an interesting topic. Interesting that there was so much coffee. I'm actually tomorrow morning. Thank you very much for your patience. Yeah, I couldn't, I couldn't. I couldn't, I couldn't. I couldn't, I couldn't. Thank you, sir.