 Rhae, ac hynna'n fwyaf am y gallu gwybodaeth i'w bwysig am ymargrifennu yma, ac ydych chi'n gwneud yma. Mae'n ddweud o gweld i'r unrhyw o'r hunain. Felly, roeddwn i'n gweithio'r rhannu ymlaen yn cyfathoriadau, cyfathoriadau a cyfathoriadau ymlaen. Mae'n gweithio'n cael eich cyfathoriadau ac y cyfathoriadau yn cefnog, ac mae'n cefathoriadau a'r cyfathoriadau, ac mae'n gweithio'n cefathoriadau, i fynd i am gweithio'r gweithio'r rong o'r lleol yn ddiolch yn y gefnogi. Rwy'nellawer o'r professor am y bus cambio y Oxford a'r ysgwrdd ym ddiwodraeth wedi'u meddwl o'r Oxford Internyd, rydych chi ar eich cyfnod rwy'n ei gweithio. rhaid oherwydd lu gyda Rheidraeth, sydd ydym yn erbyn yn iawn lle iumpwyr. Felly eu hwn yn mae playlistiaeth. Ond rydyn ni'n gwelfwyr, pwynta THISYST, gydym yn ymlaen o ganfodol, gynaedau, ddechrau, yn aethrapeuys, unrhyw symffymau, dweithio os ydym yn iawn. Rydyn ni'n amlwg bod pan oedd unrhyw llwyddiadau ac rydyn ni. Rydyn ni'n ceisio ar y program oedd oedd sgwyllwyr o rhaid o'r internet a'r Polnwyrwyrdau. Mae'n gael ein ysbryd ynglyniadau multimethodol, mae'n gwneud eithaf euthanogroffau, mae'rardıad o'r O Oxford Internet Survey, ac mae'n gweithio'r awrddant o'r gweithio'r gweithio'r gwahanol sy'n gynghlu iawn i gyd-gweithio'r Internet a'r social media. Ac rydyn ni'n gwybod i'ch cymdeithas o'r cyst-rhoo. Fy yw'r cyllidau yma ymddech chi'n hefyd ar y cyfrifau. Mae'r gweithio'r cyfrifau digital a'r edrych yn iawn i gyhoeddus yn gweithio'r blaidion herselfi, yn cael ei gwyfyrwyr ychydig, yn gweithio'n gweithio, maeodol wedi cael ei gweithio'n gweithio a'r gallu i'n gweithiodiol mae'n rhaid i gwybod yn dyn ratwch gfionedd, sydd yn gweithio'n gweithio cyfans i'r personne plidigol yn gallu archwilio yn ei hunain. Ieg yw, yma diwg ar gyfer o gyfer ymlaen, ac mae yma'r gobeithio'n gobeithio ar y maen nhw. Mae'r ddysgwyddiadau digymaeth sydd yn byw. Mae'r ddysgwyddiadau digymaeth wedi'i ddysgwyddiadau cilio i gael gydag a hynny'n gwneud y ddysgwyddiadau cymrydol yn gwneud a ydych chi'n bwysig i'r ddau'r ddysgwyddiadau ac yn gilydd i'r ddau'r ddau'r ddau'r ddau. Yn amlwg, mae'n amlwg wedi'n gwybod hynny a'i ddysgwyddiadau digymaeth. Mae hynny'n ddysgwyddiadau digymaeth. went on a Goverment site, and there is a screenshot from Google Analytics, 737 visitors on the site. That went up by one when that person visited the site and everything you do, any interaction with government, is going to generate something like that. But it is not just Government or business or university sites that generate digital traces. It's also the case that social media, social media of every kind, also generates digital traces. People across the developed world, but also the developing world, spend growing proportions of their time on social media and other internet-based platforms. All the ones there and many, many more when people go there, as they increasingly do, to shop, work, entertain themselves, educate themselves, date, socialise, bank. You name it. There's a chance that people are doing it online. They are leaving digital traces, and that is also generating large-scale data. Just quickly, I'm sure in Ireland, as we do in the UK, you've been talking about big data for a while. It's the hottest new trend. It's not new anymore, but it's the hottest trend in the corporate world. Government is perhaps being rather slower to discuss the kind of positives and negatives of big data and what might be done with it. But what do I mean by big data? I need to define the term because there are many definitions. Basically, I think what's interesting to be about big data is that the kind of data that you might generate from social media or from government information systems or that businesses are generating is its real-time transactional data about what people really did. Not data about what people think they did or think they might do or think they like, which is what survey data is. As a social scientist, I find that a very exciting development because this is a new kind of way to understand the world. The survey has been the traditional staple of social science and now we're getting all sorts of examples of new sorts of data and I want to show you some examples. What is big? How big is big data? I don't actually think the traditional definition of big data is that it's outside the capability of a normal desktop or a computer environment or is one official at a workshop we held in Boston last year on big data said you basically mean data we can't handle and she had a point. But it doesn't have to be that big as long as data gives us some sort of real-time transactional view on some kind of whole population. That could be not so very big as some of the data that I'll show you will come into that category. I think this kind of data has big potential for understanding both institutions and individual behaviour. So what about governments? How might we use big data to understand government? Governments, after all, for the last 50, 60 years governments in industrialised nations have been heavily reliant on huge-scale information systems. With widespread use of the internet, with the development of electronic interfaces, then in some ways the electronic bit of government is the bit that most people interact with in most of their interactions with government. For many people, it's the only bit of government they really see of executive government that they really see. So it's a very important window on government. In some ways, we don't know that much about it. We see government websites. We hear about government websites and government transactions and government interactions. But looking at it in the round is not actually something that we do very often. Even academic researchers or indeed governments themselves. In fact, I showed somebody from the UK government one of the images I'm just about to show you. He pointed at it and said, I want that. It's his electronic interface. But we don't take stock very often of what it looks like. This is actually some data from a very big by any standards. I think it's around 30 terabytes of compressed data. It's the entire dot UK domain for the last 15 years. So it's every web page in the dot UK domain for the last 15 years. Collected by the Internet Archive for the British Library. We have a research project at the Oxford Internet Institute to try and sort of open up this data set and do interesting things with it. See what it can tell us about a country on the web over a sustained period of time. This shows the growth of domains. It shows that back in 97, there was quite rapid growth of these four overarching domains. .co.org.ac.gov. But it shows that although government and universities were recently quick to start growing, they've sort of platted out in terms of the growth. As you might understand, we don't expect government to experience massive growth. But in that period, .co.org domains have grown pretty, have grown much more rapidly. This is a log scale on the side. If we look at sort of proportionally the relative sector size on the web, we see there perhaps slightly disappointing news for universities and governments because the .ac and .gov sectors were really a big chunk of the UK web. And I'm sure this would be true in many other countries back in 96, 97, but now a tiny, tiny proportion of the overall size of the web. Still vitally important, of course. I'm sure we'd all agree, but a much smaller percentage than they were. So this is one way of looking at government on the web. This is a crawl of the entire UK government that we carried out in 2010. And this can tell us interesting things about a government, I think, things that would be quite difficult to do in the offline world. We can see how visible government is. We can count the number of inlinks that are coming into government, which is some kind of idea of how visible it is to people using Google, for example. We can look at how navigable it is from one bit of the government to the other. We can look at how you can go into government. There's sort of immediacy, the depths of links, how deep you can go in terms of information. We can measure all these things with what are called web metrics. And we can see the extent to which government looks outward, how the extent to which it looks out into other sources and draws people into other information sources, or whether it's kind of introverted, just referring to itself. You see an interesting characteristic of the UK government in an online context there. That big red bob is direct.gov.uk, now being replaced by the portal.gov.uk. But the point is, you're looking at quite a centralised government there, this centralisation of content and the running down of departmental websites, which is a long-term trend in the UK digital agenda. There, just as a comparison, this is very rough. We've only just done this map, so excuse its roughness, but what this basically shows is a sort of opposite kind of scenario where the ministries are the big blobs there and the little like green.gov.jp there in the middle is actually small in comparison with the ministry, so that's a very decentralised government. So interesting differences between governments that we can get at with this kind of big data. What about citizens then in the digital world? How is individual political behaviour changing? And how can we understand it with this kind of data? Well, I have a certain sort of argument about that, and we've just had lunch with some of you here, and I think I may not be in total agreement with all the people who were there at lunch, or at least they may not be in total agreement with me. So I'm interested to know what you think about this argument, but this is our perspective on the way that individual political behaviour is changing, and I'll just run through the argument and then give you some examples. So basically, as citizens go about their lives on social media, they come across political issues on a regular basis in perhaps ways that they don't so much in the offline world. And by that I mean that people can make micro donations of political resources, time, effort, money. And they can make these micro donations have become possible because the transaction costs have sunk so much. What do I mean by micro donation? It could be signing an electronic petition, joining an email campaign, and banging out an email. It could be clicking like, sharing a video about a political issue, maybe downloading a video, sharing a news item, tiny acts of political participation, very small acts, which previously just wouldn't have been possible. I'm sure in Ireland, as you can in the UK, you can watch the television and there'll be an advertisement saying, text the word blanket and thereby contribute £3 to a campaign for Syrian refugees. It wouldn't have been possible to donate £3 to a campaign like that before very easily because it would have cost too much money to give the £3 to make it worthwhile. It's this lowering of transaction costs that is making very small acts of participation become possible. I know that this is a slightly controversial thing to say. I think I got a sense from that at lunch that in Ireland, just as in the UK, in the UK we very much have a culture of, it's a sort of politics as pain principle. Politics is supposed to be painful and if it's not painful, then it's not real politics. So if it's just little, it's no good. You've got to go to a boring meeting or a long meeting in a cold place or something like that. As Oscar Wilde said, the problem with socialism is it cuts so dreadfully into the evenings. So if it hasn't cut into the evening, it's not proper politics and I think that is quite a perphasive thing about many established political cultures. It's certainly true in the UK. But my argument is that these very small scale actions scale up to large scale mobilisations which are interesting sort of mobilisations. They are mobilisations which can get going without leaders in the traditional sense and without organisations in the traditional sense. They can form quickly. They can fail quickly. They're very unpredictable and chaotic. And that's leading to a new democratic model. We call it chaotic pluralism in the book that I've just written with my co-authors, Peter John from UCL, Scott Hale and Taha Useri from the Oxford Internet Institute. And that's what we mean by it. Micro donations of micro participatory acts which scale up to large scale mobilisations of a certain kind. Now, we can use data, the kind of big data that I was talking about at the beginning, to understand these mobilisations and to kind of test some of my assertions just there. So, here's people sharing petitions on Twitter, for example. So, people are signing electronic petitions and they're sharing them on Twitter trying to sort of draw people in to signing a petition. There's a participatory acts in one place are spreading to acts in another place in this kind of way. Now, once these do scale up and we've seen that most famously in the Arab Spring, of course, I'm not ascribing the whole Arab Spring to social media but I think it certainly is the case that in Egypt, for example, where this image comes from, websites like We Are All Coloured Saeed were crucial in kind of gathering enough likes and, sorry, that's clicking over automatically, enough likes to kind of tip resistance against the regime over into critical mass. I think it is right to say that they played a crucial role there. As I said, these kind of mobilisations can get going without the normal organisational trappings of revolution or large scale demonstration. When the Brazilian president, Dilma Rousseff, asked to talk to the leaders of this demonstration in Brazil, she was told, there are no leaders, you can't. There are no leaders here. We're seeing this kind of phenomenon also in liberal democracies with the very small participatory acts that I mentioned. Here's the UK Petitions Platform. Is there a petitions platform in Ireland? You should get one there. Anyway, this is the petitions platform of the UK government. Any citizen can put a petition on here and start to try and raise signatures for it. Now, this site, as you can see, the number of signatures are there. We scrape this site every hour to create a really, what I would call a big data set, which is all signatures on petitions. It's all anonymised and not with people's names or anything, just numbers over the last three years. We do that in the UK, the US, and we're starting to do it in other countries too. That means that we can look at every single petition and the kind of rise in signatures for every petition and have a look at all these growth curves to try and understand this sort of mobilisation. One really robust finding across countries is that most mobilisations fail. Most petitions fail completely. 95% of petitions fail to get even the 500 signatures that you used to need for an official response on the UK platform. 99% fail to get 10,000, which is what you need for an official response now. 99.9% fail to get 100,000 you need for a parliamentary debate in the UK. Now, this is a significant number of people take on this act. I'd say around 10% of the population at least have been signing petitions and that number is growing. Petitions have long been one of the most popular participatory acts outside voting, but they are becoming more popular. The other thing we're finding is that petitions on the same issue can be incredibly successful or can bomb out completely. We tracked three petitions in the UK if you want to look at any sort of political activity. If you look at a sort of a cute animal then the chances are that you'll be able to find quite a lot of political activity going on. We looked at three petitions on almost exactly the same issue with very similar wording to kind of save, to protest against the culling of badges to the recent policy development. We found petitions which completely failed to get even 500 signatures. We found some somewhere in the middle in that sort of bright red block there on the graph and we found one that actually achieved in getting a parliamentary debate. It started at more or less similar times, same issue, radically different fates. The fact is this is quite unpredictable. These are unpredictable kind of mobilisations. Most fail. We don't know so much about why the ones succeed. We do know that the first day is absolutely crucial. We've modelled this data to show that really if a petition hasn't made it in 10 hours it's going to be digital dust. But we can do, because each of the kind of acts that I'm talking about generates digital traces, generates big data, we can do all kinds of things to understand this kind of political mobilisation. This is the linguistics patterns in a Twitter network generated by Scott Hale who works with me at the Oxford Internet Institute. We also might start to be able to model this in new sorts of ways. I mentioned at the beginning that we're a multidisciplinary institute and indeed one of the... Scott Hale is a computer scientist working with me, a political scientist. We also, the other member of my media research team is a physicist. And these kind of data, these kind of mobilisations are exhibiting many of the characteristics of what we call, what scientists call a chaotic system, the weather being a classic example. And some people have argued, I mean the fact that basically meteorologists have got better at predicting the weather. The fact that that's Oxford, well I would say it was Oxford a month ago, but it was Oxford a couple of weeks ago as well, or Oxford six weeks ago. Pretty much Oxford looks like that quite a bit these days. But the fact is that ability to predict what the weather that led to those floods and made Oxford look like that was very helpful in being able to put up the flood defences and to mitigate the effects of this dramatic weather. And I think we might be able to see the same sort of possibilities for political activity. Here is a model developed by two researchers at OII, one of which is Taha Ysari, the physicist, using Wikipedia page views to predict the Iranian election result, the most recent Iranian election result, more successfully than traditional methods of predicting. And I think that's quite, that's a very interesting possibility. The kind of political superstar of big data, Nate Silver, who I'm sure many of you have heard of, who's just written a book called The Art and Science of Prediction, has argued that just as meteorologists have got better at predicting the weather, we ought to be able to get better at predicting electoral politics using things like Wikipedia page views, would be one possibility for doing that as we have done. But it also might be possible to do all sorts of other things. I mean, as a political scientist, when it came to the financial crisis, I used to enjoy saying to people I knew who were economists, you know, we didn't see that coming, did you? But of course, when it came to the Arab Spring, as a political scientist, it wasn't such a good joke for me, I mean. Because of course we didn't see it coming, and very, very few people saw it coming even after the Tunisian Revolution. So there might be a possibility here to think about new ways of predicting political activity, but more than that, of understanding the sort of data that I'm talking about, the big data generated by digital governments and digital citizens to make government better. What do I mean by that? Well, for example, there's a lot of free information out there on social media about what people think about government, what they think about new policies, what they think about services, how they experience it somewhere. And there's a lot of willingness among citizens to actually express their opinions about that. How many, you know, what percentage of the Irish population use TripAdvisor, for example, or other feedback systems when they're dealing with a private sector? Why wouldn't they do that about government? And some people on discussion forums, like Mumsnet, that's a big one in the UK, but there are many, many others, are participating in discussions, they're talking about government, they're giving their opinions for free, giving information for free, and that could be used to generate data. We've just done a little piece of work with the Department of Work and Pensions in the UK to work out the feasibility and data about what people are thinking and worrying about when it comes to universal credit and benefits changes and how we might use social media data to inform us about that. Of course, there's challenges to doing that. I mentioned multi-disciplinarity a couple of times. I never thought, as a political scientist, that I would be working with a physicist and a computer scientist, and that can be really hard. I have an email from somebody else in my research team and thinking, oh gosh, do I understand it? Maybe I do, maybe I don't. Actually, I have a mathematical background, and that's a big challenge for government. We really are talking about using multiple methods and multiple disciplinary perspectives to do this kind of analysis, this kind of policy analysis. But I think it could be really worth it. The Snowden revelations of the summer and the new insights we got into what intelligence agencies, principally the US and the UK, were doing with data, gave a big knock to the reputation of big data when it's got anything to do with government. In some ways, I think governments have almost got a sort of responsibility to try and look for ways of using data for good rather than just monitoring for bad. During the whole discussion of what the intelligence agencies were doing with big data last summer, policy makers were rushing to say, oh, it's too big to use, it's too big to understand, we throw most of it away, we don't use it. But there might be an argument for actually using that kind of data, for gathering that kind of data and using that data to make government better, to make it more in line with citizens' needs and preferences. Of course, there are major ethical challenges here. This kind of data has to be, it has to be anonymised. We have to develop a whole new ethical framework for using this kind of data. Many social media platforms that I've mentioned have very stringent regulations on what sort of data you may draw down for them which you may not, and for very good reasons. We have to be sort of wise to the kind of intellectual challenge that that presents as well. It's actually easier to draw down data from Twitter than it is from anything else so we don't just want to look at the world according to Twitter. But I think if you can, if we can sort of face those challenges and develop the right kind of ethical frameworks for dealing with this kind of data, there is a real possibility for enabling government to be more agile and responsive. I think the internet platforms and social media really are part of the democratic weather now and governments have got to, it's not going to go away and governments face the challenge and they've got to sort of get used to this new environment. Now, I don't think it's going to be possible to completely satisfy the demands of these snowmen against global warming, for example, but it may be possible to kind of understand a bit better what their concerns are and to allay their fears and to be more accountable to them.