 I see North viewers. Amazing. There you are. Hi, Tom. So there's two of us in here, but I think separately are the viewers who are sort of out there in some curious way. But there's north of them because there's north viewers. So I wonder whether the URL at the top of this screen is the one that should be shared rather than the URL of the page behind the screen. If in doubt, share both and people can just randomly click on them, right? OK. And Matthew can see me. Now that's interesting because it says there are north viewers, but at least one person can see me. So that's a win for Google. Right, well, if that's the case, I have to assume that there might be up to a million viewers. In which case, I should probably start going. It is four minutes past net. So right, well, it's bizarre when you're talking. You have literally no idea how many people are there. But let's keep going. Anyone who can signify on Twitter or via email or any other way that they can see me, that would be great. Or even on Google+, or who knows what. And maybe a myth maybe you can monitor and see if you can find out. I will do. Right, so firstly, welcome. Thank you very much for joining. Sorry if the beginning here has been slightly confusing. This is all new technology to us. I'm Tom. I'm the director of My Society. And this is a hangout about what Say It is. And hopefully, we'll answer some questions. And hopefully, we'll be able to take some questions from whoever is out there. And we'll be able to answer them as well. So the first thing to say is that while you might have seen Say It as a demo website at sayit.mysociety.org, it might not be totally obvious from that what the long-term goal is in relation to why we as a nonprofit have decided to build it and why I'm hoping that those of you who are listening in might be interested in it. So Say It is ultimately about helping people to learn about and hear about statements and comments and speeches that are made by powerful and important people when those statements promise to directly affect your life in some way. And another more simple way of saying that is this. If relatively powerful people are talking about you or talking about your family or talking about your street or talking about your job, then you should be able to find out very easily indeed. Possibly find out without even actually having to go and look. Possibly find out by just having that fact that someone is talking about you close to you via modern digital means. Why do we believe that people finding out about what powerful people say matters? Or why is it a problem worth tackling in any way? Why build any software that is about the transcripts of what people say? And the answer is because we have some experience within the UK where we started that is precisely about the popularity of just finding out what people said after it's been written down. And that example is on our British Parliamentary Monitoring site. They work few. And this is a site which contains lots of transcripts that are taken from the Parliamentary site and loaded into our system. And we send emails that are automatically generated to people to subscribers where people have said, I want to know when a politician mentions a topic I care about in the British Parliament, or I want to know when my own politician, my member of parliament, talks about anything. And that has proven relatively popular over the years. We send about 30,000 emails a day when parliament is in session that are all custom-eated mail is different because one will be for a certain keyword. We have one. We have a keyword for people talking about. They work few in parliament, for example. And so this is why we believe that there is, as it were, an appetite out there for people who want to know when powerful people are talking about stuff that matters. And in particular that they want this information pushing to them in a way that is convenient and easy. No one wants to go to a database every day of their life and search it just in case something important happened yesterday. What people really like is to say, I just want to know, I want to be notified when something matters to me. So if that's the long-term goal of saying it, which is to mean that people in all sorts of places can find out when powerful people are talking about stuff that affects them, affects their lives or their jobs or their house or whatever. What's our overall approach to doing this? Well, the answer is our overall approach is to build tools that are useful for other people. Tools that can be deployed by groups that are already fighting for the rights or for the needs and services for the betterment of all sorts of different groups. So this is an approach that is contrast with saying, actually, the way we're going to solve this problem is like Facebook and we'll make a single website for the whole world and we'll try and get everybody in the world to use our website. That's not at all our approach here. Instead, what we're trying to do is build a free open-source technology that means that if you currently try and protect the rights of a group in a certain country or you try and advocate for more power for citizens in some city, that you can take this piece of technology and you can invisibly embed it and you can use it so that people in your city or country or region find out more when powerful people are talking about things like, oh, maybe we'll knock down this house or maybe we'll drive a motorway through this suburb or maybe we're going to change the law in some way that people will be affected by. And so that's partly why we're having this hangout today. This is the first hangout we've ever done of this kind, hence the fairly kind of a ropey technical start here because we're trying to meet new people, new organizations, individuals, organizations who think that the people that they try and help and the people that they try and serve might benefit if those people could get much, much easier access to knowledge about what powerful people have said. And by powerful people, I don't just mean politicians, although that's quite often, it could also be powerful officials or people who control important companies and so on or unions or whatever. It's, say it is fundamentally about when the world contains people who make decisions that affect the lives of other people, how can we make sure the people whose lives are being affected get to find out when things are being said that are going to help. So, say it is this piece of technology that is about that. I'm going to pause here for a second because Miff might be better able to tell me now, my colleague Miff is also on this hangout whether anyone is asking anything as I blab away here. No, I'm very gratified to see we have nine people in the room with us, Tom. As Tom explained, the technology is still new to us. So, I'm a little puzzled as to why some of those people appear to be in the room with us and others are simply viewing. And thank you very much to the people who have typed to let us know that they're here and said hello. So, as much as I know, Tom. All right, that's fine. So, do continue to use any channel you can to post any questions or comments, including on the Google Plus page or via tweets or something. Yeah, I'm monitoring all of our social media, so I will pick it up. Yeah, so what I thought I would do now is just some of you are probably already familiar a bit with what, say it is, but I thought I would just show a couple of the features, which I believe I can do through the magic of screen sharing. Let's see how it goes. It might work relatively easily. And what you're looking at here is not to say it in the same way as you'd look at Facebook, which is to say the definitive website, facebook.com, and there is no other website. What you're looking at here is just one deployment of, say it, that we happen to have made as a demonstration. And so, although, as of right now, because the service is very new, all the deployments of, say, that you'll see are running on our own servers, over time we expect that the vast majority of versions of this will actually be on other people's servers, on other domain names, and so on. It's got a pretty simple kind of classic Google-esque sort of homepage because searching through transcripts is the real essence of why transcripts matter. Almost no one actually wants to read a transcript from beginning to end that, so that's clearly not a sort of realistic idea. But looking through for things that have come up in discussions like this is an incredibly common thing. And we know it's incredibly common because so many people look through things that are said in Parliament, in the UK, via our website, they were few. So if you wanted to look through transcripts of this inquiry in the same way, without, say, you would have to open dozens or hundreds of different PDF files that cover this one inquiry. And you'd have had to basically search through all of them. Or you could try a Google search if you knew the right way of doing it and it might work if you're lucky. But what we've got here is a relatively nice, fairly unambiguous thing in this research result for this term through all of the tens of thousands of species that were given in this big, big inquiry that happened in Britain. And if I click on any one answer here in the search, it will take me through where this was said in context. And you might note that I have linked through, and the link has taken me not to the top of the page, but it's taken me to the exact part of the middle, the middle of the long transcript, where the bit we were looking for is present. It's highlighted in yellow. It's all pretty simple stuff, but it's all completely lacking in big repositories of PDF transcripts that you find in most places. So that's an example of the basic sort of value of the fact that search is even possible at all. But what if, for example, we don't want to look through all of the people who are speaking, we only want to know whether one person, one particular politician said something or one particular witness in this inquiry. And so by clicking on the person's name, we get to a page here that has some information about them. And the most powerful tool on this page is the ability to see, did this person ever talk about a certain term? And you can actually see this person talk all about horses. Clearly, their testimony was very, it was very, very horse-sensitive. So that is an example, essentially, of the main functionality of the site. It's basically fundamentally about making it possible to read things that other ones are just basically quite difficult to read. Each speech can be linked to in a context way, so you can put nice links on the internet and you can rely that when people click on them, they won't get them to the top of a page. They'll get sent to the bit you're interested in. And not only that, but you can actually, every single individual speech in the entire, in an entire, has its own individual web page, just like every tweet has its own individual web page. So if you want to refer commonly just to a single thing that somebody said, you can do that. So how do people get speeches in? That's clearly the big question that I imagine a lot of you are wondering. And the answer is that the answer I give today is going to be different from the answer I give you in a few months, because we're going to add more ways of getting data. As of today, the way of getting data into an instance, into a deployment there, is to convert it into the relevant standard. The standard is one that, it's an open standard, it's a free standard to use. It's also, it's a standard that already exists. So it's not a thing that my society has just made up of it. I think that is for, that is for, sorry, it's already been created, it's already in operation in other places. We're doing this because we think it's really important to support other people's standards and not just to reinvent them every single time. And this web page I'm showing you here, which is accessible from the say at home page, on click away, includes essentially technical instructions about how you would go about investing into the format required. Once you have converted some data into the format required, you can then, you can talk to us essentially because at that point, we can help you take the data you've made and room it can be uploaded in one go and it can be live in a version of say at very quickly indeed. For the moment, what we're encouraging people to do while this service is very new is if you're interested in exploring this, we encourage you to talk with us about converting data and then we will upload it to a version of say at that sits on our own servers and runs on our own servers. That's just a temporary thing. And the reason why we want to do that is because it's much quicker if people don't have to worry about both making the data and then setting up the software. It's quicker if people just have to worry about making the data, we will worry about the other half, which is getting online. And then if it's online and it's good and it's looking useful, then we will embark on a second stage with you where we'll work on getting the software to actually run on your server or in your website. The, again, I'm just gonna pause for a second and ask Miff whether we have, whether there's anyone who's asked anything in particular? Yes, but you've pretty much just covered it. So we've had a couple of questions from Will Perrin. Basically, his first question was whether if he had an audio, a very clear audio recording, whether he would be able to upload that. And secondly, he was saying if he had 10 PDFs of important council meetings, would he be able to just copy and paste the text out? And I think basically what you said has just covered exactly those questions. Yeah, there's a bit missing, which is I basically said, as of right now today, the only way to get data onto a deployment of SEA is to have the technical skills to convert them into this open standard. That is not a thing we expect most people have, most people don't have programming skills of that kind. That's not a reasonable thing to expect. Over the next few months, what we will do is we will improve and then release a web page that you can't currently see, which is the web page where you can indeed type in new transcripts or where you can cut and paste in new transcripts. That will both be possible to cut and paste or type in transcripts that are from, for example, a PDF or from a piece of paper. But we also have some technology in there so that it will be possible to upload some audio and then for you to listen to the audio through the web browser while you type. And the main reason we haven't launched it yet is because making that a good user experience so that it's relatively kind of pleasant and enjoyable to sit there with your headphones on and type and stop the audio and rewind it when you miss a bit. That needs to be, we need to work a little bit to make sure that that is not an unpleasant experience before we make it public. It has been used once already. There's a transcript knocking around of a local government meeting that someone did in using that interface. So it exists, but it's probably another eight or 10 weeks before we'll start asking people to have a go. However, if you're interested in having a go, if you have some data that you would like to cut and paste by hand or if you'd like to type up some audio or work with someone else to type up some audio, then we are very interested in hearing from people who say, I want to be a tester of that because the first time we make that available or before we can make it available to the general public, we really need to work with some people who are willing to try the interface which will be a little bit new and it will be a little bit ropey and a little bit difficult. And so it will be good to have people who are interested in being testers. And I hope that maybe if you're on this hangout, you might be one of those people who is interested in being a tester. So are there, again, I'm just going to ask Miff, are there any more questions? No more questions. Will is mentioning that we will need an audio typist's foot pedal. Not sure how tongue-in-cheek he's doing. Yeah, there is a whole world of USB plastic pedals that are set on the floor and that basically mean that you can stop and start the audio and we will almost certainly have to make, say, it compatible with that because that's the tried and tested way of doing this most easily. I was thinking we might even, possibly as an incentive, we might even buy a few and send them out for people to help us test with since they're unusual but they are very cheap, it seems. So I'm just going to say one last thing in case anyone is wondering, oh, how does this actually, how can it really be integrated in my site? How can it ever be mine not like my site even? So this last point will be a tiny bit more technical. Say it is written in the Django framework in Python and it's been structured in two ways that mean that it's possible for it to benefit people who are running their own websites. So the first way is if you are yourself running an application or website, sorry, if your website is built using Python, then you can use, say it as a Django app, that is to say you can essentially have it as a module. We've just launched a South African, or today we're launching a South African parliamentary monitoring site for friends of ours in South Africa who are an organization that monitor the South African Parliament all the time. That website is 90%, bespoke, it's 90% not code for anywhere else, but 10% of it is the transcripts and that is a copy of say it as a Django application running embedded in site. So option one, if you happen to have written your website in Python, then you can build it in. But of course many people haven't done that. And so it's worth mentioning that there's a couple of other ways that we've tried to design say it so that it can be relatively seamless or it can be value and it can be embedded in other people's services. The first is that once data is stored within a say it instance, it's exposed not only as webpages for humans to see as webpages for humans to see, but as of very recently basically as of today that data for any page or any speech or any session, that data is also exposed in that bad open standard that we use. And all you have to do in order to see the data, if you will rather, all you have to do if you want to make use of that data is go to any page on say it at all. So I will, again, I'm gonna show my screen here. So here is a page from a trial that we have made available. So that's a web page. It looks like a fairly normal web page. If I go and I add letters dot an, which stands for the name of the standard to the end of the URL at the top of the page and I press enter, you'll see another view of the same page. It's the exact same page, this time it's formatted not for humans, it's formatted for computers. It's there, it's in a common Toso, which is the standard that we use. So if you run a website using a completely different language, then you can still benefit because you can suck the data out of an instance in a nice format and you can display it in ways that you want. If you don't want to display it in using the interface that we've built. There will also be a third way, which will be super important, which is we will just make it possible to embed bits of, say, transcripts using technologies like iframes. So if you're, for example, a WordPress user and you don't have strong kind of technical skills, we will make it relatively possible for you to have bits of your site that have navigable transcripts on them just by cutting and pasting a few lines of code. And then the formatting will look more like this because you will be sucking in an application from outside. So the formatting will look fairly standardized, fairly vanilla, but you will at least get a, you'll get a nice usable interface that can be run on your site without too much trouble. So some of these things already exist. Like I showed you the data feed in the different format there, some of these things will be built over the next few months. And we hope to sort of to wind up, unless we've got any other questions, I hope that we will move from alpha in which we're really working with people primarily testing and improving the way in which stuff is published. That is focusing all about how things are viewed, loading in data, loading in data, the sort of, as it were, the slightly more difficult way using the standard and then publishing it and improving user interfaces and improving the search engine and so on and so forth. For the next maybe couple of months. And then starting as of a couple of months from now, we will turn our attention away from how transcripts are displayed and searched and we will turn towards the way in which new transcripts are made. And whether that is from audio or whether it's from text that's cut and pasted or any number of other ways. Possibly the very final thing I'll say is I'm sure some of you have been wondering about voice recognition. And voice recognition is definitely something that we will explore and indeed already have a bit. There is no doubt that it's not a magic bullet because if you have an iPhone for example, you'll know that voice recognition on that is not brilliant and of course, Apple have billions of dollars and their voice recognition isn't that fantastic. So what's important to get right first is before we go towards voice recognition that is a slightly more marginal technology that might be harder to get working. Where we're going to focus first is we're going to focus on essentially trying to make it a better system to type audio into than Microsoft Word than other similar things that are used at the moment by people who are often paid to make transcripts for a living. And we want to make it so that say it is a tool that all of those people want to use instead of typing it into Word because it's just so much more enjoyable to do it. So I think that's the end of my little spiel. And I'm going to ask Miff one last time if anyone has come up with any questions. No, just a question again, I think you've covered it already. It was, do we envisage a deployment model where say a council would deploy an install of say it and use that sort of directly for its transcripts? Yeah, so one of the reasons to work on this technology, sorry, one of the reasons to work on the part of this puzzle that is about making transcripts instead of publishing them is precisely because we are aware that both governments and kind of campaign groups and so on don't have very good tools with which to make new transcripts. So yes, if you were, for example, a local government or you're a court or some other organization that already pays people either sometimes or every day to sit and make transcripts what we would say is that you would either make a deployment of this open source software on your own servers, you'd set it up and you'd start using it and that would have nothing to do with my society or if you felt that you needed some help doing that then we will probably supply service contracts for people like we will actually say, you know what, if you want a version that you can reliably type into for the next year and you don't want to worry about hosting and you don't want to worry about anything else technical then we will give you a fairly traditional, I suppose, piece of technology support although the piece of technology we're supporting is a free and open source piece of technology rather than one with a cost attached. So yes, I would absolutely love to hear from a local government anywhere in the world or a central government or a parliament or a courtroom that said we're really, really interested in replacing the tools that people type into and replacing the things like Word that we currently type into and instead we would like people to be typing into our own version of set and then it would go live on our own website straight away. Right, I think unless MIF says that there is any last... No, that's all, thank you. Brilliant, well I am going to press the magic stop broadcast button and we will maybe do this again soon and at the very least we'll call this a learning experience about Google Hangouts. So thank you very much if you came along and I hope it was useful and please do drop us a mail if you want to know anything else about it. Thank you, bye.