 Good morning, everybody. I think we're going to get started. There are still people going to be filtering in, I think, as they get through security. Sessions of video, though, so they will be able to see it. So this morning, we've got unique perspectives from the civic tech coal phase. It's always nice to be chairing a session. I probably want to be at anyway. And first of all, we've got Isabel Ho, giving us the legal story of Gov Zero. Yes. Okay. Good morning, everyone. I'm Isabel from Taiwan. Thank you to be here for such a early morning. I'm glad to be here to share my thoughts about some legal experiments we did in Gov Zero, which I think enabled the growth of the civic technology community. I'm a lawyer focusing on technology and working with the internet startups and NGOs in Taiwan. In 2000, I started working with people from the open source community. And I'm also a participant of Gov Zero since the first hack zone in 2012. Today, I want to share the story of Gov Zero from a lawyer's perspective. In Gov Zero, the participants are always trying to find new ways to get things done, even in the legal arrangements. What I want to talk about is what we have, what happened in the past six years. It's a process of evaluation. The reason I started thinking about what I'm going to share today is because of a question I was asked by a lady in OGP Summit in Georgia last summer. That is the first time I attended an international conference as a participant of Gov Zero. I met this lady from Mongolia, NGO, which do civic tech projects there. And I give her this name card, just the one you see. And when she saw the logo, she said, I know this. And then she said, what is it? I'm kind of surprised because she can recognize the logo some way, okay, at least some way. And at the same time, her question beats me. What is it? What is Gov Zero? Does any one of you never heard about here about Gov Zero? Before this presentation? Okay. Okay. Then you can scan the bar code there. And that will link to our web page. Okay. The Gov Zero starts in 2012. It was a hack song at first. In the past six and seven years, Gov Zero evolves organically because more and more participants getting involved. Today there are, there have been 34 hack songs, three international conference. Some of you were there in Taiwan with us last November. And the 500, more than 500 proposals and more than 300, which is someone who hosts a project, which we speak, come in Chinese. There have been more than 3000 contributors, but there are only, these are only numbers from a point of view, from the point of a lawyer, what is Gov Zero? What is it? Because too many things have been done in Gov Zero. Like so many conference and hack songs. In some people's minds, they would think Gov Zero is well organized, just like this image. But there are people in the ground and God in the heaven. And because some, at the first, the idea of Gov Zero is to forking government. So someone might think Gov Zero is like this. This is the central government structure of Taiwan. And so there are bureaus and the department of different functions. But no, it is not like that. Gov Zero is more like this. There are people gathered during the weekends and don't really know what to do or what is going to happen. It is what Gov Zero looks like at the very first hack song. And it is what Gov Zero looks like today. Only it is a little bit more complicated. It is somehow in some way organized. Okay. First of all, I want to make it clear. Gov Zero is not a legal entity. It is not a company. It is not a foundation. It is not an association. So what is Gov Zero? What does Gov Zero have? There are citizens who become contributors. There are documents on the hack and the hack MD for the collaboration from remote. There are code in Github. After six years, Gov Zero has an approach of collaboration. And Gov Zero is a network to connect citizens of different backgrounds. The contributors can be programmers, designers, activists, NGO members, public servants, lawyers, and that, et cetera. So what is environment which allows Gov Zero to survive and evolve during the past six years? What are the essentially go elements which helped forcing Gov Zero? The essential legal elements. At least my colleagues. I think it is a creative common license, open source license, open data. And these are legal instruments, enable collaboration. These are legal instruments to empower citizens to take actions to make change. Though I think open source and the creative common license are really important. As I was in, okay. Next, I want to talk about how Gov Zero works from three different parts. First is participants, organization, and collaboration. Participants. Okay. The famous quote of Gov Zero is, ask not why nobody is doing this. You are the nobody. We often say nobody cares about. It's kind of passive. It's only compliant. Complaint. But if you are nobody, just like Diddycat, that makes huge change. Diddycat is nobody. Diddycat is about a lot of things. And Diddycat makes huge changes. So that's our participants. Everybody can be nobody. Even a six-year-old boy can be nobody. When my son was learning phonetics, he proposed to use Mon Diep, the dictionary by O.J. Tong to make a Mon Dick phonetic terrorist game. And he proposed this proposal, and in two weeks, he got the game. And the ideas inspire ideas. There is Chinese vocabulary terrorist game after that. And I'm so glad it still works well after six years. Okay. So everybody can be nobody. There's no membership, no qualification to become nobody. This is the easy part. And then I will talk about organizations. It's just what I said. The first hack zone is like this. But after six years, it's more like this. We have different tax force in the GOP zero. So we have this food supply tax force, which is very important because we provide pizza during hack zone. So people will come and enjoy the work and the jinx and the talk and the program together every two months. So this is a very important part. Maybe the most important part. Okay. And then we have the different tax force. Like OCF actually is one of the control, which hosted GOP zero news. And we have V-Taiwan, which is famous internationally, and so on. So in these different tax force, actually they have their own governance model. There are three different governance models. The first one is open to everyone. Like GOP zero hack zone, the grant hack zone, and the cofax, which is a plan to do fact check. And also there are some tax force, like you need to get the yes votes from current members, like GOP zero talk. And when we are got an invitation for talk or interview, then the tax force will try to arrange things. And also there are some tax force by invitation only, like Joe Song. Joe Song is a tax force who hosts the hack zone every two months and also do the grant. We have a civic tech grant program. Then we'll talk about collaborations. It seems very loose community. But we do have some rules. We think to do things. We have the GOP zero manifesto. We have code of conduct during the hack zone. And we have this GOP zero domain registration rules. So some of them are binding. Some of these rules are not. But these rules are discussed openly on hack path. So people should follow them. If they recognize themselves as GOP zero participants. So this is a manifesto of GOP zero. It's simple and clear. And I think after six years we need another version. So I may propose another draft of the manifesto because things change a lot. So what are our decisions in the legal respect that GOP zero founders have made in the early stages? What are these decisions about the shaping of GOP zero? First, they decide not to register as a legal entity. Because at first it costs too much for us. We are volunteers. So there is no entity. And no partnerships. So we just held a hack zone for people to come. Okay. And no GOP zero chat marks. And licensing under these creative comments or open source license is not compulsory. So you can even use proprietary, propose a proprietary project. But just people may not join your project. So, but actually in some tax force, in some tax force, the control, the hosts do have set up actual rules. Because control can decide how to run their own tax force. So I was a member of a tax force called JOSONG. So I'm going to share about JOSONG's governance and experience. JOSONG is a tax force organized grant hack zone every two months as well as three tax grant program. The program is trying to support the project in GOP zero to sustain. Because we have so many proposals. But most of them I died not give a life after hack zones because people are volunteers. But it also opens up for application to projects outside GOP zero. We'll find eventually these projects were included in the GOP zero community. They get more support other than just the funding such as mentor connection, early feedbacks. In past two years there are more than 200 proposals and 24 grants. The funds amounting to 300,000 euros. And these are JOSONG members to raise the money. And actually TDCATS was in charge of the program in the past two years. So TDCATS, the nobody did really huge change. So we have the enforcement rule of GOP zero grant. Because we are giving out money. So we have extra rules for these grantees. First is to be consistent with GOP zero manifesto. And second, it needs to be licensed under open source licenses. So it also needs to meet the freedom defined definition of free culture works. So that's what I share about GOP zero. And also I want to share something about what GOP zero affects the statutes in Taiwan. This is a public donation open data for us. It's in PDF format. It's just horrible to us because you don't do anything about PDF. And the government said it was open data. So they are hackers, contributors in GOP zero. They try to digitalize all political contribution data. They just print out the PDF files and take it into small columns. And they call the people, they put it on the website. And the people can just input the numbers. So it will digitalize all the data together. And so this is what we do in 2014. And after four years, the government decided to amend the statute. So maybe in next year we can see a real digitalized donation data now. So it affects the statute also. So that's what I'm going to share today. So thanks for being here. Thank you. Isabel, and welcome to everybody who's just joined us. The talks are being recorded, so if you really wanted to see that you couldn't get in. They'll be on YouTube shortly after the event. Next I'd like to welcome Matthew from My Society to talk about when civic tech matures. We'll do questions at the end, but due to the late start we might not have quite as much time. But I'm sure there's people if you don't get to, we'll be happy to ask a question. Hi, I'm Matthew Zumbos. Martin said I'm from My Society and I'm going to be talking about some of the unique issues that arise in running projects for quite some time. So My Society has been around in its current form for 15, 16 years. And I've been with it for 15 of those years this year. So we know quite some time. So I'm just going to run through some of the projects I'm going to sort of talk about and how long they've been around and then go into some of the details about the issues. So our first project was launched in early 2005. It's called Right to Them. It's a contact your representative site where you can send messages to your local councillors up to your national representatives and members of the House of Lords. It builds upon the work of previous things, Facture MP and Stand, which date back to 1998 and fundamentally hasn't changed a lot since then. And that's still running now. In the same year, so 2005, we launched two other websites, Pledge Bank with a collective action website. And here for MP was a constituency based mailing list for MPs to talk to their constituents. And both those closed in 2015. So they had a 10-year run. They worked for you, started in 2004, but was not a my site project to begin with, but was bought in-house in 2006 when we also started running the Petitions website for the British Prime Minister for number 10, which we ran until the 2010 general election. Fix My Street was launched in 2007 as a national street reporting service. And that's still running now, as is what they know, which is our national, the UK's national freedom of information requesting site. And then lastly, MAPIT is our internal postcode lookup tool, which was one of the first things I wrote because Right to Them needed a postcode lookup tool to map your address to your representative. But it was internal until our National Mapping Agency Ordnance Survey released the data as open data in 2010. So with all of these and all our other projects, we do our best to automate them and make it as self-sufficient as possible. But you can't automate everything, no matter how small or self-sufficient you can be. You've got to, there will be investment in terms of updating, maintenance and support. So first there are the issues that can make you feel a bit like the Red Queen. I don't know how many people read Alice with the Red Queen last, but the Red Queen runs very fast in order to get nowhere at all. And if you want to get somewhere else, you have to run twice as fast. And running your website for a long time can feel a lot like that. There's just the general maintenance of updates, the framework you write your project in might, will release updates, might stop supporting the version you're using, the servers you're running things on might require updating, server updates like that. When Right to Them launched, we had a server connected to a fax machine on an unlimited fax plan. So because there was a lot of MPs who didn't know about email at the time. And so we would send them faxes, which obviously has its own lovely maintenance issues that you don't really get with the internet. Not sure how true it is now, but a long time ago working with the Facebook API was, if you wanted to keep, if you were doing something with them, they would change it very frequently without notice, which made it quite difficult to work with if you wanted to keep your thing running. So yeah, general maintenance like that. Also people's expectations may change, and so you might feel like you have to redesign your website in order to keep just keep it fresh and keep people using it. So I think Fix My Street and Right to Them have both had one main visual redesign in their lifespan, whereas they were few and what do they know? We've had two, if you ask me now because he's a designer, two, maybe three. I'm not sure that's because they're more sort of Right to Them and Fix My Street are very focused. You go there for a purpose normally, so they're maybe more simple projects. I'm not sure. I'm just researching there. But I find that definitely designing both at the front end and the back end, designing progressively and for resilience definitely helps reduce the burden. For example, Fix My Street has always had a photo upload so people think, well, I like to say we have the world's largest collection of dog-proof photos. It's not the sort of thing you think you're going to be doing in life, but we've always had a photo upload for Fix My Street back to the very beginning. And because we thought people might be on slow connections even back then uploading the photos, we always built in a way that if something went wrong, uploading the photo, the photo upload went fine, but something else went wrong with your submission. It would remember the photo so you wouldn't have to re-upload it again. It would give you the form back. It would have stored the photo server side so you didn't have to upload it again and save you some time, which meant 10, 12 years later when we got the time to make the photo upload much more busy and modern with Drag and Drop and nicely working and stuff like that, that made the whole process much easier because that sort of server-sized stuff was already there. Fix My Street, the maps on Fix My Street have always worked without JavaScript, which Google Maps can't seem to manage, so they're a bit bigger than us. And then with any long-running code base, you also have technical depth, which if anyone doesn't know, that sort of like things that were definitely the right thing to do at the time and no longer, like, as things change, they're no longer right, or you've got to do something you've not got as much time as you would like to do some things that you're forced to compromise on decisions. Example there is the number 10 petition site that we used to run had to deal with a very high number of people signing petitions and getting confirmation emails at the same time. So we ended up writing our own email sending daemon to be very small and quick and not even touch the hard disk of the server it was on, which was fine, brilliant, it worked fine. Years later we're no longer running the number 10 petition service, but we still use the same code for some local councils, and the code runs fine, so it's still running, but there's no need to decide to change it, but there are issues if it ever goes wrong, or there's just like, we need to look something up, because it's a special little thing. Now it's just an extra effort and stuff, so issues like that can arise. So the general issues, they're still going to arise with any project. Now I'm going to talk about relying upon external data sources and how they could go wrong. What I'm going to call managing the flamingos. They're all out of this, so I hope that guys won't know. So this is going to be about they work for you, so if anyone who doesn't know, they work for you. It's a national parliamentary monitoring site for the UK. It began as basically a scraper that scraped the HTML from the official Parliament site, passed it into XML, and then imported it into a database for display on a website. The HTML that Parliament was producing was partly a manual process, still is. It's an amazing feat, they get it together, I think. But that also led to frequent issues as they were like, because they were writing in some word-like programme and just marking up people's names manually. So if they didn't mark up the whole name in bold, sometimes the colon was in bold, sometimes it wasn't. I don't know how they selected it. So it was just the past obviously kept breaking, and rather than make our code more and more complicated over time, we added like a little patch-up programme so we could fix the source HTML at our end, and then that would get it on our website quicker. And one advantage to doing that is once it's in a database, databases are like computers and can run things automatically and run checks. So one morning I had an email from the pastor that just said it couldn't pass the list of people voting in the division, and so I was like, okay, well that happens all the time, there's probably a mistake. The code is fragile. So I had a look and it was like it was perfect, the HTML code was perfectly fine, but it was the name of someone who wasn't NMP because it was someone who died the week before. But they'd been listed as having voted in Parliament. So we were like, oh, we checked. So we just emailed Parliament to go I think they probably didn't vote. Could you check? And they were like, oh, thank you very much because there was nothing automated at the time. So we didn't actually have a lot of work to do getting it to a distance at all. So those things were sort of changing over time. And as I said, part of our aim is to hopefully show how things could be improved and better, and so we could hardly complain if that then happens, even if it creates a massive amount of work for us. So Parliament have relaunched their website in the time that they would use to be running, which completely then broke the underlying XML that Parliament uses to produce the website, which we still had to produce a new pass about hopefully one that might be slightly more stable until they decide to move to a new publishing platform, which again they're perfectly within their rights to do, but obviously can cause a lot of work. Could cause a lot of effort to us. And even then there's still a manual process at their end, obviously, which can lead to only last week our pass got confused because is a bold tag inside a member tag Either way it can happen. And things they have to deal with, they have to deal with last minute change as well. Proxy support was added after a lot of discussion, was added to Hansen at Parliament, the UK Parliament at very short notice recently after an MP delayed their cesarean in order to vote. And so Hansen had to obviously very quickly as well deal with that and put it in Hansen in some way for the very next day to be published that morning. So then our pass was like I don't understand what this line means and so we have to then adapt that quickly enough to get it on our side. I think it's worth it in the end. One more thing is the IDs that Parliament use in the XML that we use it turns out per person as we would like their per roll. So when the Bishop of Birmingham retires and there's a new Bishop of Birmingham the ID gets reused when there's a new Marcus of Chumbly and the old Marcus of Chumbly. I don't talk to me about why the United Kingdom has unelected registries and bishops in the House of Lords or in the Legislature. They do. So in a reverse situation of the dead MP voting for a brief period they would have said a dead Lord had voted which was our fault because we reused the old ID by mistake. So there's obviously all maintenance of making sure that things are right. The Scottish Parliament which a few also covers we've had whole years like 2011 to 2013 we've just stopped updating the Scottish Parliament when they were a few because they changed their website and we just didn't have the resources to get it working again. Eventually after another change they made they agreed we changed our past for a bit but they agreed to keep publishing it in the old format for us so things can break on their own and things can break when little things change and then there's things that don't break but just go things can go out of date if you just leave things they just sort of go out of date so they work for you, have voting policies where we say how MPs are voted and sort of try and group them into categories and describe some votes and that requires manual work to keep up to date and to just do it what do they know, our freedom of information site has to keep a list of who you can contact all the bodies which requires manual work by volunteers and representative details have to be kept up to date they obviously change a lot of election time but between elections and again a manual process, we did some work with the WikiData and Facebook last year to get things into WikiData for national politicians hopefully that will help keep things up to date obviously there's work to them, has local level as well where there's many thousands of politicians to keep up to date elections back in 2005 for updating the overview I happened to my old job to be in Canada on the day of the UK general election which meant the time zone difference meant that I could update they work for you during my day when it was overnight back in the UK which was very handy as the results came in of the general election but more recently I've not had that luxury but there have been more people have popped up like Democracy Club and your next MP which is a community of people to help maintain all this data and the last one which is the one I'm involved in a lot is Boundary Data which comes up a lot in technology whilst I know Brexit might be getting most of the coverage at the moment a few days later on April Fool's Day I don't know why they've picked that but there's some big boundary changes happening in local government in England some councils are merging like all of Dorset Dorset is split into two levels of council but they're all merging into one big council on the 1st of April I know some big news and some other some are set on western Taunton emerging together but obviously for that we need the new Boundary Data in order to for Fixed by Street to keep sending reports to the same to the right place and for right to them to have the new representatives the elections for these new councils aren't until May so there's a month period where no one's really sure who your council is going to be I think we're just going to maybe turn the site off for a month that's my preference but we'll see if anyone notices or complains but there's elections in May for those new councils that come into existence on the 1st of April so I thought maybe they could do it the other way around might make more sense I'll suggest it for next time so the UK is in terms of releasing boundary data quite advanced now it's open data if you're lucky you can get it in advance of the elections so that's good so getting the data you need for other countries is definitely we had a lot of for the Democratic Commons WikiData work we did we had a lot of problems just finding boundary data in some countries and stuff so I've done a lot of technical stuff there but I must also mention the elephant or the dodo in the room which is funding quite often people and organisations seem to want to fund many things quite often you see a lot of people asking for submissions for new projects and ideas and a less likely to fund to keep things going so my side is actually being very lucky in getting unrestricted funding and ongoing maintenance funding a lot over the years but hopefully that can continue and stuff is a big problem for people to get the funding if it's set up as a volunteer thing like Isabelle was mentioning if you're just reliant on volunteers people have got lives and might not have the time if you want to be able to sustain do it sustainably I have to find money from somewhere so what my side does is to run commercial services to help fund the charitable work so I work quite a lot on Fix My Street which we sort of work with councils to help integrate better with their back-end services and provide their reporting tool like a version of Fix My Street for them to help provide funding for the charitable work also we did some work with Channel 4 UK Broadcast a few years ago they were doing a TV series about empty homes in Britain and the scandal of them not being put into use so it was a sort of charitable thing we worked with them to make a Fix My Street style reporting tool for reporting empty homes in your local authority which not only was provided funding but also gave us the ability to add Northern Ireland to Fix My Street which had been lacking until then that gave us the resource to do that so that was really good another one more issue that can arise is our sites have been around for so long that people can start to assume they're just part of the furniture like the Cheshire Cats we just sort of fade until we're just the grin is left as we are and we have to continue people don't sort of realise so when they were few it was set up in 2004 we took the copy of Hansard from the official website without permission Parliamentary Copyright at the time didn't allow you to take copies but we thought that you should be able to so we did it's probably bad PR if anyone tried to do anything about it really and now obviously everything is lovely of the open Parliament license so that's brilliant and great I don't know if we would want a little push hopefully in that direction along with many others but nowadays obviously everyone's like they refuse it's just this thing that everyone uses to check how their MP voted and stuff and people don't think about the people that are behind it the small sort of official body recently the BBC drama Bodyguard I don't know if anyone watched that but at the beginning the Bodyguard becomes Bodyguard from the new Home Secretary and he goes home and he Googles them and they mocked up they work for you a copy they work for you with the Home Secretary on so he could look up rather than the official Parliament website I was like okay so yeah that's good so what do they know freedom of information because the government in some cases are not that big a fan of freedom of information so they can try to push back and one time they try to introduce limitations to any FMI law which is a thank you law but you've got to again be vigilant which requires effort on people's parts then of course there's the sad issue of if you can't maintain everything to the level you might want to to keep it useful and functional you have to make the hard decision to shut things down or something happens that what you'll do is now no longer necessary so examples we have so Pledge Bank which I mentioned at the start was a collective action website where people could like say I will do something but only if other people will help me do it which has some successes like helping set up the open rights group and some charitable things but it predated Kickstarter unbound group on all of them but it didn't have the resources the scale or the commerciality really to be taken further especially with the idea that I mentioned that we thought that things should try and be as self-sustaining as possible maybe in a looking glass world we'd have your group on millionaires here for MP was I said a constituency based mailing list and at the time email was very novel to quite a lot of MPs but nowadays obviously people have got too many ways of being in touch and their representatives vice-versa can get in touch very easily with the people they want to so we thought it was no longer a service we needed to run and then lastly the petition site that we ran the government took that in house after the 2010 election it became an official thing with parliament there was not enough signatures they would debate it in parliament which is great to see especially I don't have to do any user support for it anymore so that's a great thing for my point of view and so the last thing I'm going to cover is if your service become more popular things you might want to think about foresee like remember forwards the white queen there's a poor sort of memory that only works backwards and you get more users speaking selfishly that's not necessarily a good thing especially if they're not paying if you get more users and each one of them is paying you for the service that's generally a good thing I believe for a lot of charitable free services more users generally means more user support more technical issues more storage space needed even we find more enticing to spammers our websites are popular now that you get spammers trying to use them to contact your users through the insight messaging service and stuff link spam and stuff which you don't have to deal with for fix my street growing more popular led to a lot of leads to lots of UX issues as we get more reports showing them on the map in a way that's not confusing that still provides overview and stuff lots of things to think about there making report creation easier when you've got so many like 12 years of reports in the system now and then lastly we spent quite some time making fix my street and out of the telly the what they know code base customizable and installable and it is now installed in many places around the world both of them but now obviously leads to more support to both from the people who reused it to help with stuff which is great but also writing your code from then on in such a way that it is easily like they can do their own thing that still you're doing your thing and so sort of like being able to have the things working together which is a good problem to have but it's not something we did at the start because we didn't know that be where we'd be going with fix my street but it's something to think about like if you might be reusing them to consider how what you're doing can be generic enough that it can be customizable like people for different uses I hope that was an interesting insight into history and thanks for listening thank you Matty it's a shock to me that anything gets done at all I work with Matty every day and there's things in there that I didn't know which make me feel a little more worried if I want to get back home Gemma's let me know that we've updated the schedule on the website now with the new timings due to the delayed start so tiktek.mysociety.org should be able to find it there and finally I'd like to welcome and sorry if I massively your name Walter Pommershofer talking about the impacts of Open Shufa on credit scoring transparency in Germany thank you well it's my pleasure being here and you could also have introduced this nobody giving from Gaff's here it's a nice example so I will shortly give an overview what we did last year on our project Open Shufa we did that together with algorithm watch and I'm Walter nobody, Pommershofer working for the Norwich Foundation Germany which is in my opinion the second best civic tech organization out there in Germany and Shufa just to give you a background even in Germany if you want to rent an apartment or you want to have a new cell phone contract you need a score from Shufa in order to get the things and they have a lot of power, they have a monopoly to charge people at some points where we think it's not fair and the whole thing is not transparent and the whole people are struggling is like cleaning up the Shufa record or sometimes get misjudged by Shufa that was the main idea behind why we did the campaign Open Shufa and Open Knowledge itself we have like three columns we do like accountability transparency and participation nowadays it's called like public interest technology or we normally over say it's like international political technology because we basically also want to try to change something where we go to the lawmaking process in there and the core projects are one is like Jungtakt where we teach young kids ethical programming with Will Hackers there were like six events last year all over the world another part is like Code for Germany that's like code for all where everybody knows where we have currently running 25 labs and another cornerstone is the prototype fund which is basically a fund for funding individual programmers up to 47,000 euros to do for six months their civic tech program and then we have Ask the State Frachtenstadt which is like where we bugging the German government and from that last year we did opening up the company registry which was not available in Germany we had open corporates and another one was like we published the law itself of Germany which was like copyright issue which brings us which is brand new that went online a few minutes ago at 10 o'clock so we had a FOI request for some study on the Klufussart which is like kens inducing pesticide and one of the ministry basically said we had to take down the study because it's like against the copyright issue because it was done by a ministry so basically we are using it to take it down so it goes into the court and actually then it will be like thought out like is it actually makes sense which we think it's completely BS to have that angle that you can take require to take down a study from the government so that's basically a nudge at all open knowledge foundation Germany is doing what I always say is like we are helping Germany moving from painful manual data operations where this big beautiful analogy was like with container shipping to automate the size rollout on data structure once you have open data, open source and standards and we do that in different variations a beautiful project is OPAL which basically is like a standard on how you get data out from the civic hall all over Germany in like one format so you can also create the tools and one project last year was the so called open shufa project so in order to study we did in the beginning in February 2019 a fundraising campaign where we got around 40 thousand euro to create like this platform where people can send in data and we were blessed with having shufa as an opponent because they really were stupid in the sense that in the beginning of the campaign they had like for two weeks on the front main page disclaimer what open shufa is and we were like yes that's it because now everybody knows actually what we're doing and those of them over stretched it so the busy thing our campaign is causing an economic male function in Germany that even like everybody who could think straight was like that's like way above the line so that helped a lot for the momentum and the campaign itself targeted like so you can get as an individual the shufa score they will send it back to you home via mail and what we did at that part so first we encourage people to request the individual score that they get it and then we created a platform where they actually can upload the unique score to our platform with different levels of privacy since they're talking about Germany and privacy is like a big deal there and it's basically like your financial history it was a big problem from the campaign because a lot of like consumer rights organizations basically backed off and said like we were following like how it will work out because it's like too much risk for the campaign for them to be associated with so that what we did and we had another partner which allowed us to like manage in like this whole request process to make sure the taxes send over there you have your ID there and making sure to track it when it's done it's similar like what we do is a fact and start the freedom of information portal and on that portal we just had like 30,000 unique request just for shufa so the other companies out there who doing just credit scoring in Germany the problem is that like not that known and shufa is their household brand name for that stuff and that's what you get back so you have like a different scores basically when you for example want to have a mobile phone you get one score if you want to have a new bank account you get a different score and like parameters so basically it goes up to 100 in theory so 99 point something is really great if you're like below 95 you're like really efficient score and that is basically depending then if you get a rent for apartment or not and what we see here that was the long list in the beginning and that's a short list after the GDPR actually kicked in May 25th so that was one of the big issues that in Germany over like a regulatory authority who should take care of like how the data is in there and since then we only get like that less information which makes the reversing from the crowd source data more complicated the current status is that this should get back to the state before and we are pushing for it but it's like Schufer so it's like a little bit tricky so altogether we had like 100,000 data requests 30,000 for you Schufer the other one were like father credit scoring agencies and from that 30,000 we got actually 3,000 back which were like usable and out of this data set we didn't like as much as we could do reverse engineering as you say like trying to calculate how this score actually works the problem here was also that we have a bias in the data so we have like urban male young people it's like a big bias and what happened there Schufer in the summer basically threatened this journalist that he was up writing fake news so that was really great because the whole campaign we dominated almost like twice in the evening in the completely news cycle once we launched the campaign and once we had the results at the end of November and that was thanked to Schufer because it's like really not really liked in Germany and they're not really up to like how to handle crisis PR and so this is for example what we had like from the data sets back you can say like where's the range of your score and what is clear if you for example were in insolvency or you have an execution proceeding that your score is like from here from 100 like somewhere really below that's okay you can discuss about it if it's like just and what is the whole thing with when you rate some person what we are looking at it it's like where is actually wrong data or bad data and we found like four areas so we had one thing was like people with no negative data had a bad score so we had 20 people who had like perfect like data set but they got like a negative score which means like if they want to get a credit or loan for a house they pay more and giving the size from our sample put up to the population of Germany it's like 100,000 people are affected by some like weird data limbo another thing was that you Schufer is pretending to have a perfect detailed score although sometimes for the majority only have like three or less data points and then they pretending to when you saw the numbers before that it's 99.95 or 99.90 and that little thingy actually makes a huge of difference but there's no data for that to back that up another one was like they have different versions of the score going back to 2001 till like the latest one is 2011 and depending who is using what score which bank you sometimes get a different outcome and it's like you have to understand the problem here is like company is collecting some data then they use that data to bring it to Schufer Schufer is calculating like the score and then a third company is using that score to make a decision and the thing is always then like saying when you say there's something wrong in this data chain somebody says well it's not my fault because I was not collecting the data the other one is like saying I just calculated but didn't make the decision saying it's like how to break it up and find like who is responsible because there are like a lot of individual cases when people try to fix that thing that like everybody is blaming the other ones and there's no pressure on it like to fix that so another one was like where you could argue is like there's a bias on age gender and how many times you moved and for example how many times you moved in their series like the more you moved the more problems you have but some people have to move because of the job and they get a penalty because they're moving from XYZ to XYB because of they are like in whatever tracer stuff like that and you can't clean that up like proactively you don't even know that something is happening and if you want to know it like once a year you can ask for free and otherwise you have to pay to get the information what they have over you where they sell the data and that's like there was for us the whole automation to move that on and that's my favorite one from our data sample what we figured out there's the perfect setup for you to get a higher credit score is you need that's the number of credit cards that's the number of bank accounts and it's the number of mobile contracts and finally it's always like if you have two credit cards, two bank accounts and two mobile contracts your calculation isn't higher and there's sometimes a weird logic behind it some people actually thought that the banking account is weird because if you're happy with one bank why you need a second one but there's a bias towards like what some matured is doing that's then considered to be the best case so on the impact side at the end of October the German ministry from consumer protection and lawmaking gave out the report publication and we were mentioned and it also like framed the whole debate and like you had for automated decision making what needs to be done that the criteria should be there and you also can re-verify the input variables at the end of November then we had a big launch with our media partners by and speak online on the results and again the hilarious part was basically like got asked for a statement and then they put down a nine-pager and this claiming like you can't site out of the nine-pager so that's actually what helped us a lot because everybody knew like why you're writing even nine pages when it's not available to the public so this is a German minister who after the campaign also pointed out that something has to be done especially like the authorities monitoring Schubert have stepped forward that was from the account when the presentation was so it's transparency authorities and you the consumer should know what the criteria is for the exact measurement this will be like a long-term project over the next years to figure that out how that will be done because there's a lot of pressure on it and open Schubert it's not bad at the lobbying on the area calling a lot of like for example journalists up and when they do reporting to make sure it's like onto their line and that was promised for the beginning of 2019 that we from the paperwork when they send us mail mail I actually will provide electronic data we are still waiting for it and that was from my quick presentation and we always say Schubert is the beginning not the end we know that like other credit scoring agencies in Germany out there we also like in Poland now this is a topic and when you think it's like a long-term view I mean that stuff will not stop just by credit scoring when you look for example towards China we have automated decision making on like a human level and I think this is like why we chose that project was because we had it in the pipeline for years but you sometimes need to wait when the time is right and also this is one of the few examples from our point a few where people actually understand what that scoring means in reality because otherwise this is like really dry boring data thing to explain and nobody's like who cares right but giving the great reputation of Schubert in Germany it was a self-running project if you have any questions please let me know and the slides are online, thank you thank you Walter that was absolutely fascinating I'd love to see the same thing on credit scores in the UK so please join me in thanking all our speakers for their talks and their excellent timing the next session starts in 20 minutes and we're not going to have a break so if you will do 10 minutes for questions if everybody's happy with that and you'll have 10 minutes to grab a coffee and a pastry and we'll make to the next session everybody happy to do questions you had one yourself there I have sports questions one for the seven and one for the matches both your talks resonate a lot with me and with Robert's program which is pretty much the French society that exists since 2009 and we have a very similar mechanism of way of functioning as you do except we are an illegal entity we ended up creating a real association but otherwise we are very horizontal and trying to let everybody participate on the only volunteer work which is the recent session with Matthew as we are running a parliamentary monitoring website as well and we have the same issue I'll come back to my question but first I was wondering since you have no legal entity how do you handle things like we have an open cultural foundation we have an open cultural foundation which is organized by different communities in Taiwan because we have very strong civic society so there are many different open source communities and they held an event every year so they decide people from different communities decide to create this open cultural foundation and he becomes an umbrella foundation and he handles the logistics and money things but it's not our only choice we can choose another one just every task force they can choose other entities to co-operate that's what we do for the moment how do you handle administration of platforms basically it's a volunteer it's a wrong volunteer yeah so just like Joseph we raise money and it goes into the open cultural foundation account but we just try to make an agreement between the foundations and to arrange things Matthew I was wondering when you talked about scraping the source is manual so when you work with it you always have some new problems and that you have to fix them I was wondering what kind of process did you set to facilitate this and to let other developers contribute to it so the problem we have is that basically there's a technology that has only two people that can actually work on the scraper and know what to check so we have scripts that analyze the result of the passing and take a look at it but we have some manual human reading outputs just try to find potential errors so I was wondering if you had some good inputs we were the same sadly I'm afraid yeah we tried to spread it around to other people as they joined the team or whatever but it fundamentally was an issue but less so now thankfully but it was in the barn in your exchange with the environment are they constructive are they fixed programs if we spotted a mistake we let them fix and then that would automatically then feedback so yeah do you have a question very interesting Walter I was wondering I know that this type of projects are really necessary to improve the accountability of these agencies but maybe this is naive just checking did they have some kind of mechanism in place to project against a credit decision or get a little transparency into the data or was there really nothing in place so how it worked before it was like you have real big problems once you get declined or something you can make sure that they cleaned up somewhere the data we had cases where it takes like 6 months plus if you really like going for it then can happen that the score is clear on an individual base but for example there's no real way you can go there you have like no tools that for example getting like the damage you had gotten through it while not getting a rent which is difficult at the moment and it's not even like that it's the first step but we will say that you as a being part of like this credit scoring at least an access page where you can look in and you can see what is like my score and for example if it's changed you can look at what is causing it there's a lot of things you have like with identity mismatch so you have like one person with the same name and another person with the wrong entry just to figure that then I'll take some time but sometimes you don't know it if you're for example not even aware of it that it's A happening, B that for example your employer looks at the scoring and if you have like which sometimes can happen paying your credit card, being paid or getting like a real bad mark that can affect like your promotion which are the one thing with GDPR according to our understanding from the GDPR there should be absolutely the necessity and also like a legal base that I as an individual can require that data and I think that's like being overseen by a German this authority should be controlling Shufa I mean that's like a political thing because they're located in one part of Germany and especially in Germany like a really some states have like enforced data protection that one doesn't have the high reputation of people that are really going after them it's also like one of the things was a there was a study on Shufa how they handled it and that study was paid by Shufa and stuff like that on this consumer protection board there are people who worked in on another angle for Shufa so it's a thing a lot of I think a political question at least in our approach to like to raise the awareness and I think we had some really nice impact on this one ministry the line perfectly being Shufa, being Shufa really helped like building the momentum of the campaign because it could have like almost like at my point if you would have like you could have like put it asleep in the beginning to say like well let's meet and make a workshop on it right for the next two years and let's we know but the other ones are even worse being Shufa that helped a lot and so we hope that this is and there will be actually some stuff happening in the near future and then with the bias in gender and age how do you plan for it? well you have for example bias if you're young it's worse than if you're older if you're male it's actually worse than if you're female and same as with the number of moving the question would be is there just because your age and your gender you really can't choose it will of course like this is backed up by data it's the same discussion what you have is in driving insurance and young male will clearly make more exit yeah what time for one more one more question just following on from that a little bit with then the actual selection of people who volunteer to help you with getting all of their Shufa dates were they representative of the population of people living in Germany or were they a bit more like people in this conference were they sort of civic tech people or were they it's like there is a huge bias towards young male IT related urban we try to make corporations with local newspapers so we reach a completely different audience we fail at that point it's has to do something the process you have to sign up asking for the data they send it by mail then we making as slick as possible for that like sending the data but then it's also like Germany especially all the people have like a huge a consign about the privacy which in the one part was interesting in the other part we got like un-black copies from the scoring because they were so pissed and I've never seen anything like that how many people are like really mad about one company and that also sure I have sure like for example other reports which most of them are from the campaign the different news outlets bringing clear cases of like a retired person living programs because of why some Goldman Sachs person moving to Germany get a credit score although he was like really rich and stuff like that to make it clear that there's a systemic problem in how the data gets together and making this like data value change transparent okay thank you everybody thank you