Good evening everyone and welcome to SuperPublic Paris. I’ll just say a few words for some of you who are discovering this venue for the first time. First of all thank you to Personal Democracy Forum for having chosen SuperPublic for organizing this evening and to Audrey Tang who we’ve had the pleasure of spending the day with. Just a few words in French and then we’re going to pass over into English. My name is Stephan Vincent. I direct an association called the 27eme région and with others we founded this place one and a half years ago. The place is called SuperPublic because it’s dedicated to innovation in the public sector and more broadly political innovation and democratic innovation. Some of the people who work in this place are here permanently, others are more transitory. They are all very devoted people in the area of transformation of the public realm. We organize conferences, workshops and trainings here. So we’re very very happy to host this evening and to welcome Audrey. Clémence will now say a few words about the purpose of this evening.

Thank you Stephan I’m going to continue speaking in French for the moment. My name is Clémence Pène and this is the director of La Netscouade Matthieu Lerondeau. Together we organize the Personal Democracy Forum (PDF) in France, a conference that’s taken place for the last 5 years. It’s an American conference that’s organized in New York each year in June that’s been running for about 10 years. I’d like to let you know it’s a conference with people who still believe that the Internet can change the world and in fact in 2016 I have the feeling that in Civic Tech this belief is becoming quite fashionable once again.

As far as this is concerned we will meet you on the 10th of June at the Gaîté Lyrique for the yearly conference of PDF France and this evening we’re going to launch a call for pitches. For the first time we’re going to pitch Civic Tech startups. So thank you very much to SuperPublic!

Thank you. I am Matthieu Lerondeau, director of La Netscouade. We support and contribute to the organization of the Personal Democracy Forum every year for nearly 5 years at the opening of the Futur en Seine festival.

For those of you who don’t know Audrey yet. Forgive me for saying that Audrey pretty much became a star of these days in France. For those of you who haven’t met her at the event “Nuit des Idées” (Night of Ideas) it took place on Jan 27th. For those who haven’t seen the excellent documentary “Tous les Internet” on Arte and Claire Richards paper at the end of September 2015 I will give some very simple and short introductory words.

Audrey Tang at 12 years old became a a self-taught programmer and pretty much decided to leave school to code. She created her first start-up at 15 and defines herself today as a retired. That is definitely a young retired person. There are other different and interesting other things: She is notably a conservative anarchist. I would like you (Audrey) to explain what you mean by that.

Also and probably more important for us, people from civic tech in France and all of us civic designers, she is someone who reinvents tools to reinvent democracy. Just to finish up and let you start with your talk I also want to say a few words about g0v.

You joined this group of civic hackers in Taiwan and definitely played a very transformative role during the year that was defining the new politics in Taiwan. The year of 2014 during the Sunflower Movement gave rise to an entirely new breed - you could say - of open government. So we are very happy to have you here today and thank you for giving us the opportunity to have this conversation.

Thank you so much. I am very happy to be here and to spend more than two hours with you. I love this SuperPublic place because it feels like I am speaking with you, not at you. It’s like I am looking at this screen just like everyone else.

I would like to talk a little bit about the format of tonight. Because this is actually four different talks that talk about g0v and about the g0v way of civic hacking. It also is about the sunflower movement which is how we occupied the Congress or the Parliament for 22 days. We will also talk about the so-called post-sunflower politics in Taiwan. Because I don’t know which of those four or five different topics are actually what everybody is interested in, I propose I just go on with my slides and anybody could just raise your hand or just start to speak out without raising your hand.

You can speak in english or french. I will try to understand french and if needed we have excellent interpreters here. We can use the slides as material for discussion and I will just doodle something on the slides. And when people generally look bored I will just fast-forward that particular part of the slides. Like that people can talk about whatever they would like to talk about. Is that ok with you guys? (General noding.) Cool.

The facilitator wants me to talk a little about the “conservative anarchist”. This actually is a very mundane idea. I joined the internet back in 1992. At the time the political system that defined the internet was the Internet Engineering Task Force. Anyone could join, there were no credentials,… People didn’t have voting, kings or rulers. People just sat together or wrote on mailing-lists and they got a rough consensus of where the Internet is going. This was mandatory for the internet because if people - with my computer and your computer - don’t agree on the protocol with which computers talk with another there is no Internet, right? So people have to arrive at some sort of consensus and you cannot really coerce anybody to do innovation. It’s an oxymoron.

So the idea of the internet is that the people innovate however they like, and then they try to convince their peers or neighbors that this is a good idea. If they convince their neighbors their neighbors adopt the same protocol. Then the internet upgrades with more useful kind of protocols like the World Wide Web. So that is how anarchy is by definition. That is how the early thinkers of anarchy define it und put it into practice around the 70s and 80s.

“Conservative” also has a very mundane definition. There is a tradition that is generally good and worth protecting. And as the world changes we try to adjust the tradition very slowly to not break what already has worked before but have it adapt to the current world. That is what conservativism is.

So a “conservative anarchist” therefore knows that the anarchism model has worked since the 70s and tried to keep the same anarchist model that defines the Internet and try to keep it working in different areas and endeavors of human history and of human society. So this is the kind of anarchism that I try to conserve. As you can see this is very much an everyday thing. This is not radical, it is just a way of living.

Central to the idea of the internet brand of anarchism is the idea of “fork”. People who are working as coders or hackers may recognize this word. This word means basically that on the Internet it is free to duplicate somebody else’s projects and somebody else’s designs. “Free” is meant in two different ways: First at no cost or at very little cost. Then again when people relinquish the copyright you don’t have to ask permission. We call that “forgiveness of permission”. So if you see a website or an internet protocol that doesn’t work entirely to your liking you can fork it meaning that you take the same program, you run it on your computer and you change a few things so it works differently. That means you don’t eliminate what has come before but you take it to a different direction. If you also make your modifications public, as early Internet hackers always did, then the original people will become the upstream people. The people you get the fork from may decide to take your contributions. This is how science advances.

This is how the early internet lawmakers released the internet laws by forking each others programs. When I say law I don’t mean jurisdiction, I mean physical laws because the internet protocols, the laws that define what is possible and what is not possible. That is what physical is. So what I will do is share four or five different stories about how we applied this kind of anarchist internet oriented way but put it to governance which has not been a domain where it has been applied before. So that is about today’s talks and again feel free to interrupt me if I’m talking too fast or with too much jargon. Just interrupt me and we will have a discussion.

So I am @audreyt (referring to the Twitter account on slide 2) and this slide is under Creative Commons and already on my Twitter account. People can download it and join the discussion perhaps on Twitter. I just arrived two days ago to Paris and I came from Taiwan which is seven hours in the future. So I’m literally from the future and I am almost done with the jet lag adjustment. So I am very happy to be here and talk about what I have been doing, as the moderator has introduced already, since my retirement in 2013 which is two and half years ago.

Before my retirement I worked in the IT industry for 20 years (referring to slide 3). It’s not a early retirement really, it is just that I started very early. Then I still do consulting work with Apple. I have been working with the theory team on computational linguistics for the past five years, with Social Text Facebook for the enterprise. I also worked with the academics and the dictionary people in the new OUP (Oxford University Press) and with government sector people in the Taiwan National Development Council. But most importantly I work with them not for them. By working on projects that are located in the third sector which means the voluntary sector meaning that the people I work with choose to donate their time and effort and energy to work on the dictionary which I’ll talk about or the vTaiwan rule-making platform which I will also talk about. The result of those two projects, because they are open source, we relinquish our copyright. That is actually how I prefer to work with them. People in the first sector and in the second sector usually just harvest the outcomes and improve their own service and products.

Also as introduced I learned programming when I was eight. It was 1989 (slide 4) and I got my first computer in June. Some other thing happened in Beijing in June that year. My father actually visited Beijing in his capacity as a journalist for the first time. So he actually covered the protests all the way until I think June 1st. He flew back to Taiwan fortunately.

Then he also went on a dispatch to Berlin later that year and again something happened that year (the wall fell). The point here is that in Taiwan we lifted off our martial law that year. The first presidential election was 6-7 years after that. So the democratization began exactly the same year as the internet or as personal computers did which means that digital natives were the first generation who could actually participate politically. This creates an entirely different dynamic compared to what we see in older democracies because (in those countries) digital natives commonly were apathetic to politics or not interested in mobilization or things like that.

In Taiwan this is completely the opposite. The first generation who got on the internet are also the first generation who could protest and who were getting arrested. That’s the background. I lived in Germany for a year and a half because my dad was doing his PhD at the time and was studying the dynamics of the Tiananmen movement in Beijing because a lot of people associated with the movement flew to Paris and to Germany. So we did a lot of interviews and debates with them and we talked about how to do democratization in a way that doesn’t end violently. So that’s my childhood.

When I came back to democratized Taiwan I quit school because in 1994 the World Wide Web (WWW) was launched. I discovered that I could work with the scholars and researchers over the WWW. Because they were also new to the WWW everybody was very enthusiastic. So we worked on a lot of projects and they don’t have to know that I’m only 13 years old. So then for the next 20 years I worked on a lot of free open source software projects I think only the most geekiest of the audience will recognize. I think the Wikipedia is what people will know about. Those projects have something in common. They all create a space that is relatively safe for people to experiment, to write and share what they have. It’s okay to fail, it’s okay to do mistakes or fauxpas or whatever. Because it is a safe space and open source people eventually learn from each others experiences and mistakes and eventually get into something very valuable for the humanity as a whole like Wikipedia.

This is my colleague Lu Chia Hua (Slide 6) who said: “Behind every technology, there is a set of values informing its pursuit.” Personally (Slide 5) my value was just to build a safe space for people to learn from each other and build something that’s workable. So that is about me.

In Taiwan in early 2013 the usage of Facebook has reached about 90%. People online were using Facebook. People used the internet also around something like 85%. This is a lot of people. There is this prediction that by the end of this decade there will be more Taiwanese on Facebook than the population of Taiwan because some people have more than one Facebook account. So this is a very highly networked place.

So naturally wired I did a report on this and then asked one of the leading political thinker and writer and also heavy FB user Zhang Dachun about his take of FB and if we can improve civic engagement. He said: “No, it just feels as if we have participated. We can very easily get tens of thousands of likes but if you want to mobilize people to go to some place then only maybe10 people will go because people were lazy on FB. Anything that requires more than a minute of your time they won’t do it.” So you have to think of something practical that only costs a few seconds to make a civic impact. So we have to describe g0v with just one short sentence. We are a movement that tries everything to create a way for lazy people to engage in real action. That is the g0v motto.

This is a campaign we did a year and a half ago. We made a captcha. I assume everybody knows what a captcha is. It’s a way to prove you’re not a robot. It’s not going to work after last year where a robot solved it better than humans. Until recently it could tell whether you’re a robot or a human. Then we made a captcha where people would just type that they are here and click enter. This maybe takes just five seconds of people’s time. So it’s good for lazy people. But this website that we built said that you’re saving the country by participating in this captcha. That is because those numbers came from the campaign finance records because the taiwanese Campaign Finance Law law was defined in the era of Xerox printers. It says one has to file all the campaign donations and to where the campaign finance has been sent. All this is kept in paper form in this building. Everybody can go in and do a Xerox copy with a watermark to make sure that the tally (record) is not wrong.

But there is no digital download, you can’t download a spread sheet, there is no transparency law because the law was made before the internet. This makes sense, right? But after the internet people keep proposing bills like: “Now we should make it downloadable and digital and so on.” But if there is one stakeholder that will be negatively impacted by this proposed change it is the lawmakers. Even though the bill is always scheduled in the parliament it’s never voted on. It has been going on like that for years. So it is to the benefit of everybody except to the lawmakers.

So we decided to do it ourselves. That means we bring people to the building and make the copies. We used a A4 paper because that is the only paper they have. We printed the records in a 3 to 3 double sided way so on every paper we can bring 9 records out and then we digitally scanned it and then we asked people to digitize it into spreadsheets so that we can do real analysis on it instead of just confirming that the corrective body is doing their analysis correctly which was the only thing they could do when using the paper format.

Now if you try to digitize a page, I tried it, it may take you 3-5 minutes which is larger than the so-called natural limit on facebook. So if we call people to help with digitizing nobody we will come. We know because we tried. Then we made it into a game. We used open cv, a computer vision library, to split it into … which is what we call bite-size tasks. Now this one only takes five seconds.

Making it into a game the key in doing this is: If people have played Farmville or Candy Crush or things like that on FB, if you add a countdown timer and a progress bar and badges people will do whatever you want. They will spend the whole night digitizing those single cells as long as it takes them only five seconds and they get an instant reward “Thank you for saving the country”. Also it would mention: You know thousands of people are playing like you. We had designers who designed everything very prettily.

Slide 16: So the first batch that we brought out which is more than 30 thousand records was digitized by 9700 people within 24 hours. Each cell has at least three people looking at it and three people agreeing. So we are reasonably sure that our digitization of the finance record is ok and people feel like they’re saving the country and they are spending less time than posting a picture on facebook. So this is a win-win situation.

Were all the participants from Taiwan?

Actually no. They were from all over the place. I think 90% were from Taiwan. Because it only requires a very rudimentary OCR processing and you can even skip all the chinese and only help digitize the numbers. So international participants were also among us.

So it’s crowdsourcing but in a very grass-roots way. There is no assignment, there is no personal top-down authority and everything is in the open. So when people from the opposing party questioned this process, we say this is the open source scientific process. You can just download it on your computer and re-run everything to make sure that we are kosher. The corrective yuan, the auditing organ, then issued a press release saying that you cannot be 100% sure because they are bound to be errors. All the three people could have made the same mistake and how do you guarantee that this will be 100% correct. We said okay: “If you pass this legislation you can only release data that is 100% correct.” So this strategy kind of worked actually. They are now changing the rules as of this election cycle.

This changed the dynamic of the civic society with the government. This is not petitioning and not demonstrating. This is demonstrating in a different way. This is a demo, a demonstration of how it may work. And when the government likes it, it just takes it. So now that we have a completely digitized the records of campaign financing we can create tables and correlate their portfolio with the declared donors and the owners behind those donors. Also their is a legislator voting guide (slides 17-20) which correlates all the council members in each district and in each county and their suggested construction budgets and correlates them with their campaign finance records.

So there is a lot of very interesting things that could be done with this sort of raw data when correlated with other data. This actually really changed the voting behavior. The first version we posted had more than half a million people visiting. Under every precinct there is a discussion board so people could just post additional material, they would say something like: There are 22 candidates in my region. After reviewing this website I only have to choose between two. Things like that worked really well. We all see the party finances and so on.

This is not only limited to crowd-sourcing politically. After getting help from the international community we gave help back to the international community. There was this earthquake in Nepal. If you look at Google or Apple maps the only street-level maps are around Kathmandu and the connecting roads. The very small roads - the country roads - were not mapped by street cars because cars don’t drive on them. There is a lack of mapping around that area. After the earthquake everybody needs the mapping otherwise the UN and other relief organizations couldn’t send the supplies to the regions in need. The open street map team worked with g0v and the humanitarian teams in other countries. It is a very international effort to digitize in the same way we did for the campaign finance records. They divide the maps into very small map areas. If you are mapping the first time you receive a very small area. You mark one of the circles: o Is the road broken / o Is there a building / …

Then using the satellite images within 24 hours the mapping was completed and experts reviewed it. For the first time the satellite company donated the imagery on the first day after the earthquake. For the next 24 hours people concentrated on the post-quake images. So on the third day when the supplies actually came they had a real map reflecting which roads were broken and where camps were set up. So it really helped the relief. We can do this in Taiwan. We don’t have to fly all the way to Nepal. Taiwan made up maybe 10% of the mappers because we did a lot of outreach and instructions on how to map it. The president elect also helped a lot in promoting this behavior.

This actually points out how g0v works. It is by uniting three kinds of people who don’t actually work much together in Taiwan or anywhere in the world. We started with the free software people who don’t usually care about public matters. Then we just introduce this idea of the hands-on and fix things you don’t approach. That gained our trust with the social activists who share the same value. But then this doesn’t scale. You have to get the message out to make it cool to participate in this kind of thing.

So we need the civic media like the bloggers or the Wikimedians. Again those people share something with us because they do trust strangers. The key of doing open source is that you trust random people on the internet to help you digitize maps and things like that. This is what Civic media is especially good at and which the traditional social activists are especially bad at. This is very interesting because these three groups of people each like something that the other two other groups of people could provide. Just by organizing things in a way that lets people learn from each other eventually helped converge into something like the digitizing of the campaign finance records which was started by somebody in the Civic media, a very famous blogger. He has this very cynical way of writing about things because he says the country is helpless and hopeless. Then we brought him to our hackathons and proved to him that if you just trust the strangers we can let the internet to do the job.

If you just trust the strangers you can get the internet to do the jobs you want done.

g0v (”g0v”, the Taiwanese civic tech community) has a hackathon every other month that has 100 to 600 people. It’s a very large hackathon. And every OTHER month, there’s a smaller one that’s about 50 people - but still, a lot of people. And then practically every weekend, there’s a smaller one with maybe 5-10 people working specific on a project.

So our large hackathons are just ways for people to start new projects, like an incubator project. We meet at this large venue and when you join for the first time or for any time, there’s a bunch of stickers on a table (slide 25) and you can choose whether you specialize in maybe agriculture or music or a tax jurisdiction or any kind of programming language, and then people take those stickers and put them on their shoulders….

This is about playing musical chairs. At the beginning of the hackathon everybody with an idea goes on stage and presents (sometimes with powerpoint) their idea for three minutes. And they would say, ”I want to make a public finance report website. I would need two coders, one designer, and one storyteller for this task.” Usually there are about 20 projects for each hackathon, many of them new projects that people have never heard before. Then we play musical chairs. With stickers, one can see at one glance that this corner of open space already has their staffing of engineers, so other engineers go to some other groups. So people introduce themselves and just join whatever projects they find interesting that still have vacancies.

And usually a lot of first timers - those deer-eyed people - would stand around and don’t know where to go. People with the bear badge (experienced hackathon participants) would walk up to them and then walk with them very slowly and ask things like ”What’s your hobby? What things do you care about?” and by the end of the walk, they will find themselves in a project.

So this is a very interesting way for people to discover each other’s project and then participate in it. And then people work for an entire day and sometimes two days and by the end of the day every project takes five minutes to present what they have done that day and usually they have a prototype, and they will say ”Let’s meet online at some Slack or chat room” or ”Let’s meet every Friday” or something to make this actually complete.

So this kind of incubation project means, basically, even if I’m already involved in a lot of long-term projects, for one day every other month I forget everything about my existing projects and make myself available to the opportunity of new projects. And of the 20 projects every month maybe 15 will not survive to the next month - maybe they will prove to not be a good idea or they will lack the necessary expertise. Because all the projects are required to be open source or Creative Commons, all the failed projects then become then materials for the next months’ projects that then build on their basis.

So that’s how a lot of ideas, social enterprises, longterm projects, crowd-funding projects, etc., came about in this kind of open space. And because we don’t have an agenda - it’s an unconference - you can meet people from all walks of life. It used to be more than half engineers and designers but now we’re less than 25%. We now have social activists, lawyers, civil servants, everybody….

One of the codes of conduct we follow in the hackathon is called: “Less is more.” This is another way of saying “worse is better”. The worse is better philosophy is especially needed in the East Asian context because people care a lot about face. People don’t really want to throw something that ugly that’s incomplete or that’s shameful or something like that. In g0v we started saying “worse is better”. you don’t think about your face and we demonstrate this as the “zero” (slide 26) in the word hackathon.

The logo of the g0v were designed by two brilliant hackers and top class coders who absolutely suck at design. I am very sure that everybody here can design a better logo than this. They literally only took five minutes or maybe five seconds to put this together. I think this was done with notepad or something. This is a very ugly logo and they had the guts to just print it out on a A1 large sheet of paper and hang it up just like this in the Sinica Academy in an open space for everybody to see. Then a magical thing happened a visual designer Even Wu looked at it and then tweeted that "looking at this ugly logo hanging here if I don’t work on improving it I will lose my productivity for a day or maybe a week. I become so upset." 
(see image: https://medium.com/@audrey.tang/lessons-i-ve-learned-32f5d8107e34)

I guess it made him furious (laughing). After a day of work he came up with this (middle) which is better, right? This is like talking and chatting and voting. It is a little bit better. But then because he also relinquished the copyright other designers could also improve it. They then decided that this is good on laptops but on phones it is not really identifiable. The small dot is not visible, it is often mistaken for “Q”. So we had to register the gqv.tw domain because went to the wrong domain. Then a better design eventually emerged. We can just use the zero as in g0v.tw as our main and visual identity.

With not being afraid of losing face bringing up this ugly design and other visual designers involved non of these developments would have happened. Without relinquishing our copyright it is not possible to build on each other’s work. This becomes kind of the rallying cry and the culture of g0v. It is just whatever you have, even it is a hand-writing, a mock-up, you don’t have to know coding, you do not have to know design, you just through it out and people will get upset enough and join your team and improve your work. So that’s what g0v’s organization is about.

Before I go on to other projects is there anything people want to talk about, any comments, thoughts?

What do people work on during the hackathons?

The subjects? This is a very good question. Re g0v I think the primary innovation is our domain name g0v.tw. All the government websites in Taiwan end with "gov.tw". In France it is “gouv.fr" It is the same idea. So for example for the environmental agency it is env.gov.tw. Now on your browser if you change the “o” to a “0” (zero) you get into the shadow government. That is g0v version of the environmental agency that shows exactly the same data as the environmental agency but with much better visualization and open data. It is a better version of the environmental agency. The same is with the parliament, the legislative organ of the taiwanese government. If you change the “o” to a “0“ you get into the shadow government site of the parliament. It shows exactly the same bills. It is featured as a progress bar like a shopping cart: where the bill is coming from, who is signing the bill, who are the legislators and then they are cross-referenced to their campaign finance and voting records and things like that. So to answer your question: The topics of the subjects are as diverse as the third level domains of the Taiwan government which is everything. It is agriculture, it is education, … you name it. Whenever people want to do something as alternative to the government website, they can then register a domain that is the same as the government. Nobody has to remember those domain names anymore because everybody knows how to reach g0v you just change the “o” to a “zero”. So I think that’s the first hack that we did.

It also says that we are completely inclusive because the government by definition concerns all the walks of Taiwan life or any people’s life. so people who care about the particular part of the government can do innovation around that function. I hope I answered your question.

How does the government react?

My next talk is the one I am personally involved in. It is the educational shadow dictionary. They reacted very slowly but positively. After the Occupy (Sunflower Movement) they reacted very swiftly and very strongly. That will be the other talk. Generally there reaction is a positive one.

Is it open source? What tools do you use?

We use GitHub Pages a lot for server hosting. For most of these pages even though they have half a million views we don’t pay a dime. There is no cost in setting these kinds of things up. Joining g0v is a way to learn about all the free spaces in the internet. We had to use mostly free of charge services because otherwise when you fork my project you have to pay an additional cost which excludes a lot of people out. So by definition we use only tools that cost nothing.

Maybe a short comment about trying to compare our local context. Here in Paris and especially across France we have a lot of actors doing bits and bites of what g0v has succeeded to do. I am thinking of names most of you already know thinking of Regard Citoyen who made an outstanding job in forcing the parliament into more transparency and thinking of OKFN (Open Knowledge Foundation), OSM (Open Street Map) and Voxe.

What strikes me is that they never managed to attain the kind of visibility and the kind of crowds that you’ve obviously managed to gather in Taiwan. So do you have a clue about how you have managed to have so many bloggers and citizens involved?

I think the most important thing is to meet face to face every month in a place that accommodates hundreds of people. This is like a really old organisation tip. Maybe the groups you mentioned already do some of that. But I think the thing was that at one g0v hackathon of 150 people maybe 100 people are the first time there or the second time. They are not very well versed. Basically we went viral very easily by incorporating newcomers every month which essentially doubles the size or the outreach of the existing hacker community in g0v. People who already form long-term projects they would at least send one delegate to the g0v hackathons to both recruit new members and also see what synergies they could play with other people. I think this is really not about Internet but more about just organisation 101. So for 100 people it is Accademia Sinica, a research Institute in Taiwan. For the 50 people people of the smaller hackathons it is someplace very much like the Superpublic (France) actually.

Who comes to the hackathons?

In addition to the weekly small meet-ups and the monthly hackathons we also do a summit every two years. For the summit we pay for the airfare and flight. Many from across the world joined in 2014. There is …, the Pirate Party people from Germany, OKFN from France, the New Zealand Code For All, all the usual suspects. They come to Taiwan and learn about the local context. I think the most important thing was that they introduced the agenda they had planned for their year and then it becomes part of the routine in your monthly hackathons. So we keep thinking about what kind of ties we can strengthen - usually with tools, sometimes with procedures, and sometimes also with policies we can talk about with other international people. This year on May 14th and 15th is our next g0v summit. Vox will be there, Pandemos and a lot of European people.

Where do you find money? Do you take donations?

Yes, we do take donations. Our primary cost is very high quality food. A month after a hackathon you won’t remember any of the people or the projects by you will remember the food. If the food is very bad or very good it gives a lasting impression. This is true internationally. I think I read a comparison study about the occupy movement. It makes or breaks based on the occupy food station. I think it is an exaggeration but it is also somewhat true. In any case we spend a lot of money on food. So we do take donations and the donations gets you a guaranteed ticket because especially for the slightly smaller venues like 50 people or maybe 90 people usually the tickets sell out in a matter of hours. So people would have two ask for transfer tickets and things like that. So we keep increasing the value but then if you donate something like €30 you get a guaranteed ticket for the hackathon.

Also we only ask for money that is exactly equivalent to the cost of food and the mandatory infrastructures for the next months. We don’t keep any cash.

Is it not excluding a part of the population?

The idea of unconference or hackathon or things like that has actually nothing to do with internet. It has its roots in Nonviolent Communication, Open Space Technology,… all these things were pre-Internet things. People were given a space, a timeframe and the option to do whatever they want. This is sort of anarchist self-organization kind of stuff. So our root activist or core activist groups are people of the older generation who did this kind of stuff or at least took this kind of training and learnt nonviolent resistance in the early days of Taiwan where we still had martial law and things like that. I think there were a lot of things that people were not joining because of a lack of familiarity with the tools.

For example I insist on using something like pen and paper because it is much more familiar with people who are not digital-natives. Also we use very large digital whiteboards but it could be used in a way like people who used to use white boards in their Nonviolent Communication trainings and things like that. We try to transport the same kind of organization that people did in the pre-Internet days but by making the off-line space as comfortable as possible to the paper generation of people. Then because we also organize live video and transcripts and digital representations of everything that’s analog we also make the young people aware of what the earlier generation is doing. This may not be an entirely satisfactory answer but this is our core way of operating.

Let’s move to the next part. This is an example of one of the first projects and the one that got me into g0v two months after its founding. Moedict “Moe” meaning very cute. It is a very cute dictionary with a very cute logo. The dictionary, first published online thirty years ago, has already became the only real tool of the primary school level Chinese education in Taiwan where people and teachers especially in rural or remote island areas where they don’t have the same access to libraries and museums. They take their students to the computer lab and use Moedict to teach Chinese. Chinese as spoken in Taiwan, there is mandarin, there’s traditional simplified Chinese, taiwanese … There are very different ways of talking Chinese in Taiwan. Because of new migrants we have Holo, Haka, Amis, Indonesian languages and so on. We also have aborigines who are enjoying a surge of the rediscovery of their own language. There are other Austronesian languages also because of the large influence of Tibetan Buddhism. We have people interested in Tibetan. It’s a large dictionary with a large cross-reference to all the languages that are spoken in Taiwan. It’s been used all around of Taiwan too. The point of doing this dictionary was very typically for g0v.

When I joined the project which was initiated by 葉平, who was a physics professor in the National Taiwan University and then joined Google. He became the head at Google and Taiwans cloud department and then he moved to the valley to work for Google Analytics I think. When he moved to the valley with his children he found it very difficult to teach his children Chinese in the US. It is a very common problem. Teaching Chinese is hard enough and doing that in the US is very difficult.

When we learned Chinese online we used this gopher site which was this pre-world wide web protocol which I’m sure not many are aware about. We used to use this online dictionary that was built by the Ministry of education. Nowadays his children don’t use gopher anymore. The generally only use the mobile devices and the official Ministry of Education doesn’t really have a mobile friendly website. That makes the job extra difficult. … said “ok, I’ll just download everything from the official dictionary website and will redo it with an open API so that people can use it to make mobile websites or whatever they want to do with it. Actually we have done this over the years personally but because it’s violating copyright we never make public of the fact that people have been doing this individually.

When g0v came about, 葉平 said: “Okay, if the 30 people are doing it together then it is called civil disobedience.” It’s not called copyright violation and then it became kind of a legitimizing movement. Then we called people to download the dictionary which is very old. Since 1945 it has been maintained and is of very high quality. If you want to learn about classical Chinese which is like Latin then this is the to-go-to dictionary. Because its website was built in 1996 and hasn’t been updated for 20 years, it reflects a lot the early web. For example you cannot bookmark the URL because then the idea of bookmarking wasn’t invented yet. For example it used a very old legacy encoding. It is like ISO AA59 with which a lot of Unicode characters just cannot be displayed, so they use pictures. That means copy and paste is not possible. Also it hasn’t been designed for the mobile web because then there was no mobile web. If you view the source especially with IE5 or with Netscape 4.7+ (the plus is meaningless because Netscape didn’t continue after 4.7). So it is a very old website that hasn’t been kept up with times. Also it locks you out after 30 minutes of disuse and something would pop out that says: “You are going to be redirected to the home page”. It is true for all the Ministry application dictionaries ever since. But the catch is that there is no login button. So you can’t login but it logs you out after an half hour. It’s ridiculous.

So 葉平 had a vision and then we had a collaborative tool which is like goodledoc which we call hackpad. So he wrote everything that needs to happen and then went to sleep. We then had our 100 people hackathon. People joined and crowdsourced and downloaded everything. Then I created a spread-sheet to list all the pictures so people could fill in their unicode. It brought down Google’s spreadsheet. It will become a pattern that whenever g0v mobilizes somebody, some project on a new platform, that platform will go down. So we are the scalability testers of new services. We had to create our own spreadsheet which is called Ethercalc 2 to complete this work. Within 24 hours everything is downloaded and made into an open API cross-linked to browser extensions, mobile web, everything. It’s like rough consensus. People just have a general direction and then share whatever they have.

But what about the copyright? A few years before we did this, the creative Commons people introduced a new creative commons devise called CC0. The CC0 plays nicely with the g0v. We were the popularizers of this idea. The zero means we abandon completely the copyright, there is not even an attribution right or anything. So it’s as if it enters the public domain the second we publish something. The reason why we do this is because we used a loophole in the Taiwan copyright law. It says a government publication, if it’s used in a non-profiting fair use doctrine, part of it may be reused without criminal penalties. But the problem fair-use is of course the question: “How much is too much?” We are using 100% of the data so we had to relinquish 100% of the copyright.

We are not doing derived work, we’re technically just converting formats for the government. So a complete reuse warrants a complete abandonment of copyright. We then argue under the Taiwan fair-use doctrine that this is fair-use. We are hundreds of people. This is really civil disobedience. Then we wait for the ministry to respond to your claim. While we are waiting for the Ministry we started to discover that when we make things like the PDF copies of the campaign finance records or the paper dictionaries into you know very fine-grained data we could start to build communities around that. This is the idea we call “social object”. For each word in the dictionary we now have a URL. For the chinese word meaning “data” the permanent link is www.moedict.tw/data" So there’s no need to remember its web addresses. Its web addres is the word. itself

With that we can do what Tim Berners-Lee calls “Five-Star Data”. When you mouse over any word in the definition it would pop out using link data format whatever they could find in the same dictionary or in the wiki dictionary or in the open dictionary, anything that is linked to this word. People started sharing those words on Facebook, Twitter, Google+, and on social media. Now when people start writing something like this it is open data workshop. This is a word yet not found in the dictionary, we see a segmentation that shows the definition of “open” and “data” and “workshop”. Then we generate an open graph image that is a beautiful calligraphy of whatever word the person wrote. This solves a very important problem of Internet social media campaigners because it is very difficult to find a photo that fits your message that is also free of copyright. People usually spent an half an hour looking at the proper image because without an image it doesn’t get the same kind of shares in virality. Now people just type whatever they want and use that as a kind of banner for their message to improve virality. You would think caligraphic is not good enough for your message you can switch to any of those open fonts.

This is how we get 7 million visits per month. This is not about people who want to look up a definition in the dictionary these are people who want to use it for creating banners, for protests and for campaigning and things like that. They want to share the message in a very clear way that kind of hacks the Facebook ranking system. A good thing about this is when they click into the definition they can share it again and again on social media. Now with 7 million visits per month we can now call people to action because even though our conversion rate is 0.1% that is still a lot of people. So whenever we put a call to action on the top right corner of the Moe-Dictionary thousands of people come. For example (showing to a slide) this is a aborigine … that was done in the 60s and all we have is this low quality scan. So again we do the same thing as with the campaign finance records. We split things into rows and ask people to type whatever they see. They don’t have to know french or amis they only have to type in Latin characters even if they don’t know the language. Then we finished a thick dictionary very quickly in 53 hours. Then we turned it into an additional dictionary for the aborigine people.

People take pride saying we’re saving a culture just by typing random words into the internet.

The other project that we did was that the Ministry of Education, before they responded to our legal case, had their own exercise in asking people for corrections to the dictionary. That was a very good chance to demonstrate how we work. We had 18 days in that event. We wrote a program that looks at the citations in the dictionary where two entries site the same source. The fragment of the sentence they cite, the example, differ by only one word. These kind of things usually mean a typo when people were first digitizing it. One of the two citations has a typo. The program knows which one is different but it doesn’t know which one is correct. So we put a call to action for people to click on the search button which looks to Google to see whether it’s being used somewhere else in the dictionary or if it is just a typo. We identified more than 5000 typos this way just by crowdsourcing people to do this work. The ministry had received maybe 6000 typos and maybe 5600 were from the Moe-Dictionary.

This proves that we are not just consumers we are also contributors. We can do proofreading and we can do crowdsourcing. The Ministry of Education can decide if this is totally fair use because if they rule against us this not just ruling against 30 hackers or a 100 hackers they are ruling against thousands of people contributing to this way of reclaiming our language. This is a constituency that the Ministry of Education really cannot alienate. They very swiftly decided that this is a very good idea.

To recap quickly: The way the ministry and committees in the pre-internet era worked was usually by coordinated consensus. People have to know everybody in the same committee and when a person joins they have to know everybody else also. The problem was the human wetware. After 20 or more people these people stopped having the same egalitarian relationship. Some people become just listeners, you start to have hierarchy and bureaucracy and things like that. The way we fixed this was by rough consensus. As I talked about that re the dictionary. Every dictionary language involves people whose language I don’t speak. So those indigenous groups they all just take whatever I built from Moedict or built from Moe-Dict then use it to apply to their own material. They don’t have to ask for our permission

And maybe they try something that is a good idea. Then those ideas become adopted by other language dictionaries. Because everything is open source, it is very easy for us to crosslink our outputs together. If some French dictionary tries something that’s really not a good idea they could still do that themselves. It is just that nobody else merged this kind of change. So this kind of rough consensus moves everybody generally toward the same direction and allows people who are usually enemies like competing for the dialect resources on education like the various dialects of Taiwanese like Hakka and things like that happen all over the world, they can now work not as friends but as collaborators by working on their own language and cherry picking each other’s good ideas and contributions.

So that is how we scale, not only among the dictionary hackers but also among the ministries with the dictionary sources. Because when they finally revised their website after 20 years they could now build on our Unicode mapping, they could build on our interface, they could build on our crowdsourced corrections so they don’t really have to spend that much of taxpayer money on infrastructure because the community has done that for them. They could then merge back our work. After two years after the Moe-Dict the Ministry of Education decided to open up all their dictionary data under creative commons license so we don’t even have to argue about fair-use any more.

Again we did the same thing with the Open Data Portal. We had an open data portal that really sucks. It’s data.gov.tw. So naturally we did data.g0v.tw where you can look at the same data, meta data, but it’s done in a very useful way and it allows contributions, feedbacks and open license. Then the National Development Council who is in charge of the Open Data Portal merged back our open data license which is unheard of in Taiwan. This has never happened in Taiwan, in any level of our government. They changed their open.gov.tw license to the open.g0v.tw license. That’s compatible with Creative Commons.

The net gain of this is that OKFN (Open Data Global Index) raised Taiwan from the 11th place to the first place. Because magically all the data that were not open by the open data definition were now open by the open data definition. That is a simple fork and merge thing again.

For example the first g0v project was a visualization of the national budget. Using the PDF and Word files of the national budget we show them as a tree map, as a bubble map. If you click on it you can say you want more, you don’t understand, you want a cut, you want a deletion of this project, and people can have a discussion. If you click into it you see the raw details. Now the national government has not adopted it yet but the Taipei city government last year did. They published all their budget in the compatible g0v format. Then when people click on it and use their Facebook credentials to talk about the budget like: You say you build a stadium here, but I don’t see any construction, or my school needs repairing, or problems and things like that we’re just conversing over individual budget items as social objects. And much to their surprise after a month every office replied on this platform. They bypassed the city council entirely. They just replied to all the objective questions, they replied to the general feeling, they replied to the idea explaining why the budget is done this way. This was a very magical moment where the citizens in Taipei saw that whatever they type randomly on the Internet gets an official response from the city government. After this went over the national press all the other five major cities in Taiwan all signed on to this platform. It has now become just a regular thing in Taiwan politics. So that is again another case of merging the g0v with the community’s contributions.

That was again another 20 minutes. Any questions? So let’s go straight to the next section, which is about the other direction.

We have been talking about using civic technology to make government data more open, more transparent and more clear. But that is like only improving the bandwidth of the downloading speed of your Internet connection. So it is now 10 Mb per second. But you can still only upload two bits every four years. That’s voting. It is very asymmetrical, right?!

So the other part is about making citizens voices heard in a much more scalable way. Academically the problem was not that the government doesn’t provide sufficient information. It is that it only provides it to - I don’t want to say it - the lobbyists in the private sector. They are linked together, they have a natural synergy. They have their own industry chain. So any information that is valuable to them is also available to its vendors and its customers. It is very natural, that is how the private sector works. Then the civil society joins this kind of committee with individuals, scholars, committee members and so on.

The problem is that early-stage decisions don’t actually have the same accessibility to the people in the barricades - sorry - in the streets or the larger civic society. First it’s too professional. The scholars and academics are invited exactly because they could understand the jargon, the technical terms and all the the context that has been going on before and so on. These are not very easily understood by the general civil society. The next thing is that this not presented in a way that only cost one minute or five-minutes of people’s attention. It is the same problem that we talked in re to open data.

If you cannot let be people know how relevant the information is to them in one minutes time then it’s not relevant to them then when they get impacted it’s already passed resolution. There is a fable about this in the Hitchhiker’s Guide to the Galaxy which I will not repeat here. The idea is that it’s just too far away for ordinary people. This is a case which we kind of demonstrated. We called this the g0v sunflower digital camp because this is a demonstration of direct electronic democracy. People demonstrating to the parliament showed how should we talk about trade agreements. The sunflower movement is a prime example of the citizens not noticing anything until it’s too late.

The political context is this. China wants to sign a cross-strait trade agreement with Taiwan. It has very favorable terms because it has a political agenda and the Taiwan administration wants to sign. Many legislators don’t. There is a disconnect. Usually when we sign trade agreements with New Zealand or Japan or whatever other countries there is a process. The legislation sets up a committee, there is a public hearing,… the usual process. But because constitutionally in Taiwan Beijing is part of Taiwan and we haven’t changed our constitution yet, therefore as per the Taiwan constitution this is more like signing a pact with a domestic local city. And any administration pact with the domestic city government doesn’t have to go through the legislation - of course because otherwise the legislation would have to go through infinite things. So with this constitutional loophole the administration argues that the legislation has no say in this agreement because Beijing is part of Taiwan. Then they could sign whatever they want. The legislation could have a public hearing but after a 30 days of inaction it automatically passes because this is a city level domestic agreement.

Now nobody really thinks of things this way, even the pro-unification people. This is entirely a constitutional loophole (J.Y. Interpretation #329). But because of this loophole the legislation is powerless. They said we don’t have the code of law that authorizes us to talk about this trade service agreement. So it becomes automatically passed after a certain number of days.

On that night in the streets there was a large demonstration and I was supporting the internet connection for that demonstration. But before we talk about that particular night we can talk about a more evolved 2.0 form of that same platform which took place in Hong Kong a few months after that - this is Occupy Central, the so-called Umbrella Revolution. It’s called the world’s politest protesters - the same headline used for Sunflower movement.

I was in Düsseldorf at that time and I was typing into Twitter. There was this projector that projected into this Occupy Central building whatever people tweeted. People could feel that others are there. A journalist deployed in Hong Kong said the website of the occupy has got to be the most technological advanced in history. Then his friends said: Wait I have seen this website before. Then Chia-liang Kao (Slide 39) said, co-founder of g0v, this is forked form our git hub. Basically the logistic systems were forked from the Sunflower Movement, it is exactly the same code with some modifications. It provides the live broadcast, the mapping of the news and the logistics. The map shows in real-time to the barricades the police forces, the gathering points, the first medical aid sites.

There are a lot of rumors and there are the actual facts, there is a time-line, there is geographic information and there are cameras that the citizens have set up. People can watch four cameras at the same time and have a chat room next to each of it. Then there is a spreadsheet. Every column is a occupied place, every row is the supply that they need. Just looking at the spreadsheet you know how the occupy is going and what extra logistics and supply is being needed. This is a really nice system. This is what … called a situation or application.

None of this is possible if you only use Facebook and Twitter and Google plus. This has been literally coded on the ground and every day we changed the code to adjust to the need of that day and then used it for the next day. During the occupy this entire system got rewritten a number of times. This prototype was first done in March that year (2013) for the anti-nuclear energy plant protest. We worked with the cable electricity radio g0v to provide the protesters a high-speed Internet link because the year before there were a lot of people - almost a quarter million. So we expected a lot of views for the media, for the Internet and so on. Fortunately on that day there was a large typhoon in Taiwan and only 1/10th or less turns up because it was just raining cats and dogs. So nobody used the internet.

So we had a lot of spare bandwidth to use. It occurred to us that we can just channel the SDI connection of the show on the stage and broadcast it on YouTube. YouTube Live was actually just introduced for general consumption only a few weeks before that. So we were one of the first users. Now we discovered that hundreds of people joined. We didn’t even announce this before hand because it was totally in the field. Quickly people because they felt guilty for not going on the streets because of the Taifun they crowded over the chatroom. Eventually there are more people there than people around the stage. We realized that this was something we could work on with exactly the same equipment.

10 days after that there is the occupied parliament and we didn’t have to change the code. It was just deployed as is. But this is different because on the right-hand side, here (showing a slide), was the original protest where I thought I would just supply the internet connection. Some other people supplied the camera and protested for a night or maybe two days and then wanted to go home. That was the original plan.

While I was doing this broadcast unbeknown to me the person, the student, lent me his laptop and said: “I’m not going to use my laptop anymore so use it for for the broadcasting station.” You look like an university student. What kind of university students says I don’t need my laptop anymore? Well the kind that climbs over walls. Well, because there too heavy the occupiers only use MacBook Air and anything that was heavier than that was left on the streets. Then they just crossed over the walls of the parliament building and then broke into it and occupied the parliament.

Nobody was expecting that and only one or two policemen were stationed there. They (the crowd) were therefore very successful in occupying. The first hour or so we had the camera crew who previously were supporting the anti-nuclear protest deploying their recording devices to capture the first batch of people who broke into the parliament. That footage proved to be very important because on the next day all the mainstream media said they (the occupiers) are mobsters, they damage things, they were junkers, … the usual mainstream media way of doing things. Because we had the live and the first hand footage it became very easy to show that that was not actually the case. For the next 20 days it became a show of force between the civic media and the mainstream media.

The civic media was based on the occupiers. They set up their own live broadcasting station using Ustream.Tv. The two sides of the streets were using YouTube Live. Using the same logistical system we asked whether anybody who couldn’t come to the protest would listen to one of the three feeds and type whatever they heard into this collaborative type hackpad. Then the people who know French or English or Arabic or whatever other languages then took this transcript and then put their translation work on this spreadsheet coordinating a translation task force. That is how people from abroad knew about this Occupy in real time and they could check with their eyes what is actually happening on the live stream and realize that they are actually doing what they are saying they’re doing.

Then we used a crowdsourced bookmark to collect anything and everything the civic media pertaining to this movement. The same designer who designed the g0v logo designed the Occupy logo. All this happened in the first 24 hours. The mainstream media didn’t have time. Therefore the first agenda setting power was set by the civic media. They were saying this was constitutionally absurd. We will retreat when the legislative body agreed to talk about this trade agreement the same way we talk about any other trade agreements. Unlike other occupies we had very narrow goal. It was very reasonable to most of the people because it was made very apparent in the first 24 hours. There was very little the mainstream media could do after that.

Now we ran into the same thing any occupy runs into namely there are a lot of people who spread rumors from people who were already there. Some of them were homeless people, some of them were not exactly homeless people but mobsters and people like that. All Occupy people ran into people like these. The way we fixed this was that when there were rumors spread for example that inside the occupied area they were being attacked by the police. This rumor was designed to get the people on the outside. Or another rumor: They were counter surrounding the police to attack the police and escalate the conflict. Then one of the student leaders had to come out and shout that were not actually being attacked. That doesn’t scale. What we did: I brought 300 meters of Ethernet-line for the CPR experts deploying on the team to make all the three occupy areas a intranet. So this is a very high-tech way of solving a very old problem of spreading rumors. Rumors spread because they were cheaper to spread than facts.

So if we made facts spread cheaper than rumors nobody will listen to the rumors. The way we did this is by introducing very low latency real-time broadcasting equipments and take a projector (beamer) like the one I am using now and project everything on the two walls of the Parliament. Everything that is happening in the parlament was being broadcasted in real time with just 20 ms delay to the two streets. People don’t usually listen to the audio. So we had asked stenographers, people who type whatever they hear, to cover what was said in the occupied area. The live feed of everything that was said was displayed. People could see very easily what is actually going on and people who listened to the broadcast could then correlate this and fact check with our stenographer so that they don’t miss anything. And if they miss anything they could just type it on the hackpad. This is an idea that we call the transparent war. Eventually the occupied area also had two projectors projecting the two streets. I was as if the wars were not there as if the police are not there. The three occupied sites become a single occupy site.

Now the way we code, so to speak, is that we repurpose the idea of neutrality. We provide the service to uphold the constitutional right of communication. In the occupied area there are three neutral roles. The doctors protecting the health of both the police and the student protesters and everybody. There are the lawyers protecting the due process and we said we were the ICT people protecting the right to communicate. People were trapped there don’t have access to high-speed Internet and high speed internet is a human right. So we are just protecting that particular morale without having any particular agenda.

On that day the Supreme Court ruled that flash mobs are legal. They don’t need a permit. Because of that we are asked the telecommunication company to have a 50 Mb high speed fibre-optic link to the streets. This was the first request to the streets that they have received. They granted it because they really also wanted to see the live video. Everybody else was watching the live video. We not only had the live Internet we also had the connection to the outside extranet and that enabled a lot more participation across the walls.

On the 24th a bunch of students also decided to occupy the administration building which was a very bad idea. In any case they got repelled by the police. At that time we had our first cyber-attack on our subway infrastructures. Fortunately we only used tools that cost zero. All it takes is a name change of the domain to a different IP address. It goes online in an hour. We had a recovery plan. Re the physical occupiers of the administration we have people using an iPad and a YMax connection to take live footage of the entire process just like we did on the initial occupy of the parliament.

Now there are two sides of the administration building. On one side all the police moving in here and all the students moving in there was kind of a stand-off, there was shouting, just some breaking of glass, but they were very civilized so to speak. But then this part with no live Internet, no live stream coverage with 1000 of people the police brutality was very brutal. We learned that people behave differently under camera. I’m sure everybody knows about it but we learned very painfully that this is really the case. After that it became a kind of top-priority for us to ensure not only the three occupy areas but practically every street corner in the occupied area and its vicinity that they are being filmed by a least three different angles and cameras. Of course we don’t have that much equipment. You would need hundreds of equipment.

If you upload your photo and add your name it will print you a badge that says you are a reporter. This is useful because just a couple years ago there was a separate Supreme Court ruling that says the reporter is generally protected under the freedom of speech and then we had a QR code that links the badge to a ruling that says any ordinary people as long as they are covering something that is of interest to the general public must enjoy the same rights and permissions as any media institution. This is a very important Supreme Court ruling for us. Because of this we have hundreds of people just printing this and sticking it to the back of their IPad and becoming a civic journalist. The police couldn’t do anything about it. If they asked or they wanted to stop such people they would just tell the police to scan the QR code and read the Supreme Court ruling. That’s how we got a lot of in the field cameras and how after that date there were no injuries or nobody missing. It became completely non-violent.

After people calmed down and there was no violence anymore we started to deliberate, to actually talk about the cross strait administration. The rallying cry was that if the legislators deliberate in this building we will demonstrate for them how to deliberate this kind of thing. How did we do this? We first wrote a program. This is a g0v project which has been going on for eight months and was just completed on the occupy day. Basically what it does is when you enter your company’s number or the kind of work that you do it shows with a pop-up with beautiful comics how exactly your company is going to be impacted once we sign this trade agreement - whether the Chinese people can come here, whether their investors can come here, for how long, it is like a three panel comics that everybody can understand in 10 seconds - just the part that relates to them. Because of that we also cross reference the information because we want people to fact-check us. The cross-links to our company registration categories, to the WTO categories, to the World Bank categories and then to the Chinese categories. So it is a lot of work but it is presented in a very friendly way.

Armed with this kind of data people started to deliberate, to really talk about which part of the agreement was good or bad. In the three occupied streets broadly speaking the separatists and independantists, the green people and the environmentalists, and the left people who would care about the worker’s rights - those are the three main concerns about this trade agreement. The trick is that these three people were not friends before the Occupy. They don’t usually talk to each other and there are a lot of schisms between those three camps are people. But because of the common interest and because of this deliberative framework and because whatever anybody said on one street is then transcribed and viewed on an another street people started getting more and more consensus by the end of the day. After the administration refused to meet with the student’s demands there was a massive protest and by that time everybody in Taiwan has seen our live broadcasts and transcripts. So on that day there was a half a million of people on the street - entire Taipei was full. When this many people came we had to find something useful from them to do. Depending on the side of the street they sat down with the environmentalists debating about the environmental impact. The leftist people started debating about the farmland, the farmer or about other kinds of impacts that this would eventually have and so on. And the separatists talked about the Constitution basically.

You can join any part and then just have a discussion which was then captured on the online forum. That’s how the occupy turned people who don’t usually speak to each other into a coherent rough consensus kind of crowd after 22 days. The legislative body - the head of legislative - eventually said: “Okay, we agree with whatever you said” and we just retreated very peacefully. That was the story of the Sunflower Movement.

About the international circulation of these tools. You mentioned Hong Kong. I am wondering how China reacted to this broadcast? Re Loomio - do you have relationships with South-America and Spain?

I think that there are three different layers. The process and the tools our international by definition. We don’t develop any of these, we just apply them. They got very valuable feedback. For example Hackpad went down five times because of the overcapacity. They had to buy a special cluster for g0v so it doesn’t impact their other paying customers. So there’s a lot of live feedback for crowd-testing like this. That improves the quality of the tools because those tools were just not tested at that scale before. That is the easy part. And the policy part on which I’ll talk about in my next talk which is the Uber-Part I think it has some kind of relationship also with the Europeans cities in particular where some kind of collaboration was possible. It is harder than just the tools. Of course there’s the power structure level which don’t usually collaborate in this manner. The idea of the g0v summit was to just make clear what is the agenda of the civic hackers for the next year. That doesn’t mean that always that aligns with the power level.

You have been traveling throughout the world basically. Do you have a concrete example about a collaboration with another country?

You mean besides Hong Kong, which may or may not be a country depending on who you ask?

We do have a lot of collaboration with the New Zealand people, the developers of Loomio. They always wanted some way of transforming … and not just having an inner group decision where everybody knows everybody which is a kind of coordinated consensus model where Loomio is proven to work. But it hasn’t proven to work with strangers.

The same issue pertains to the Seattle start-up called pol.is which I will talk about shortly. They’re trying to work with the New York Times to have people deliberate meaningfully on policies, but again with strangers or with celebrities who are effectively strangers. With these two we have the closest collaboration by developing both the theories and also running the experiments that will scale them from the individual uses in civil society to something that would comfortably appear on mainstream media and have people which only have five seconds to still engage in a meaningful way. That’s the main direction. Any other thoughts or questions?

Do you know of any other movement that big in Europe?

On that scale there is Podemos (Spain) which has a very comparable scale. At the beginning the Five Star Movement (Italy) had a very comparable scale especially on the city or governmental level too. I think what’s unique about the Sunflower Movement it is not really about civic technology. It serves only as an amplifier for everybody who is against this kind of policy-making and to show them an alternative. The campaign finance story you don’t see this kind of crowd sourced campaign finance digitization project in the UK or in Europe and in the US because they were already digital. They don’t need one hundred thousand people doing this work because the government is doing what the modern government should do. But on the other hand in say North Korea or other countries it doesn’t take 100,000 people. As long as you have 10 people you get imprisoned. It requires both a very active civic tech population and also a very not so modern stuck in the paper-era just recently democratized county or state apparatus. I think this is actually comparable to Spain to some degree. Unless you have this kind of power structure that is behind in time and you have a government versus a new democratization in civil society, then you don’t get the same mobilizing power as in the Sunflower Movement.

Did I answer your question?

Yes.

The Sunflower Movement showed very concretely that as long as we have a mediation space that is built with our ICT technologies, with real-time broadcast, with transcription, with projector, … what we call reflective open space then the private sector can join policy discussions with the civic society in a kind of equal way. After the occupy the Prime Minister resigned, after a watershed loss at the national city level election. The new prime minister is an engineer, his vice-deputy prime minister a Google engineer, so we have technocrats running the country after that election. They know that they only have one year of time because everybody knows the Nationalist party will loose at the next presidential election this year. So they have a year of time and they know they cannot actually take any new directions because they know they are going to lose the next election anyway. So what could they do?

Very wisely they put the goal of open government on their national agenda meaning that all the civic servants, the non-elected people and normal civic servants who are not good yet at listening to the people through the Internet or speaking to people through the Internet must learn this skill. So we have had a lot of training programs and trainings like this one for all the levels of civil servants in the art of using the Internet to speak and to listen.

Because it changes the relationship of the government and its people - one of the primary issues here is the open data policy. Without getting too technical Taiwan’s civil law is modeled after the german legal system. Any contract between the government and its people is therefore actually not a contract. When people want to make the government make everything open data they run into some legal problems because it’s like the government starts giving away its assets. People don’t really want the government to give away its assets because its taxpayer money and so on.

But then because they really want to change to open government they have to change the regulations so that they could legally say that any data that the government produced is open data. Now they have to work with the civil society on it and with lawyers and so on. That became their primary agenda. The new Prime Minister has three agenda points: open data, crowdsourcing and big data. It is a very engineering view of a country. With this agenda they now need ways to engage the civil society largely with the same people who helped the Sunflower Movement to talk about the things that they wanted to change which were not political, that is not affecting the election, or that is not affecting the pro-independence or the pro-unification, those ideological debates.

All of these things are the things that people would care about regardless whether you are on the left or right. These are the infrastructures of the country. They initially didn’t want to talk about Uber or Airbnb but we eventually did. So initially these are the so-called regulations that pertain to the people in a cyberspace.

That was a g0v project. The Minister came to the hackathon as a normal hacker and took three minutes (presentation time) and said we want to solve this problem of talking about policy with people. We knew and were aware that there were initiatives like the Cornell Regulation Room, the Estonian or Island process all academics have written about. Any rulemaking initiative has a lots of barriers.

In short, there are three barriers:

First, normally the lawmakers are not the stakeholders and the stakeholders are not aware of the lawmakers.

Second, people are not used to debating meaningfully online. People are more used to posting cat pictures. That is a fact.

Third, policies are interlinked. If you really want to understand a policy change its overloaded with information. It’s impossible without training in public administration and legal code to know exactly how the change will affect the entire legal system. So these are a very difficult things and this is why most of the e-consultation projects fail.

A very concrete example: One of the laws they wanted to change was the telecommuting law that pertains to people working at home and people who are early stage start-up founders that want to employ people who work at home and want some guidelines for teleworking. After the election when they set the course of the new government they got some zero people to talk with the Labour Ministry. They said: Okay we have this new agenda and we want to hold public hearings. Usually they invited heads of labour unions, associations and of the industry and so on … so please recommend all the representatives of all the teleworkers in Taiwan. There is no such thing. There couldn’t be such thing. Because the coder who works at home, a designer, a musician are complete different kind of people. They couldn’t really speak for other people who work at home. It is not a trade, there is no trade union for this trade. Again for people who are early-stage kickstarter companies also don’t have a guild or association because they have trouble paying the next month’s salary. Like forming a representative unit is not their game.

They couldn’t hold a public hearing in the usual way. If they just invite scholars and people they know in the usual way one would say this is lobbying and this has no legitimacy. They were very afraid that if they work with the wrong people they would get occupied again. It’s like having something always hanging above their heads. They really want some way that is both legitimate and could talk to the legislation which was deadlocked by the two parties. They were filibustering each other. Without ways to do normal public hearings they came to the zero hackathon.

Minister Jaclyn Tsai who was head of the IBM legal Department Asia, she’s a technology lawyer and it was her first time working in the government under the new administration. She says: “Okay let’s think of this as a coding problem, an engineering problem, we want to reach everybody who registered their companies on Cayman Islands and we want to ask them why do you register on the Cayman Islands. Also we want everybody who planned to register at Cayman Islands to tell us if we change the law in Taiwan what parts of the law should we have to change so you don’t have to register in Cayman Islands?”

This is an engineering problem, a social computing problem really. This has a technical solution. I’m aware actually that there is something very similar in France. There is also a shift from the civil law to the US or UK kind of law for these kind of companies. For the scholars and the start-ups people and so on we set up this discussion board. We modeled it after the ITEF, The Internet Engineering Task Force in which anybody who posts anything constructive to this discussion, there are hundreds, are invited to the working group. People who were in a working group are just people who contributed to the online discussion.

We used the online discussion as a way to self select people who would eventually formed the working group committee. After a month we held a consultation. We had like lawyers and academics in both traditions of the law and then the ministries of economy, finance, … and then of all the people who contributed to its online discussion there maybe remained 10 or 20 people. They sat down together and then we used the same Sunflower infrastructure to broadcast everything, to take a live feed and post a real-time transcript. Everything is captured so that the entire nation knows that we are talking about this thing.

A week after the working group formed using input from the ministries and from the academics the tolerance limits were deliberated: How much would they tolerate like multiple votes per share or how much would they tolerate stocks options that are non-capital-based and so on. So using these terms we then made a request for comment, a synthesis document, again in a IETF way that consists of: “We are not lawyers but if the lawyers are going to pass this law it must contain this, it must not contain this, it should contain this, it should not contain this, may contain this, may not contain this, …” - the same habit as we do on the internet.

Now all this was very thoroughly debated over a month and a half. By the time we sent this to the Ministry of Economy of Affairs all they did was to translate these recommendations to legalese. Because of the desire of the system they have two cross-relate each point in our recommendation to their translated legalese. So it becomes the first bill in which every statement is cross-linked to the recommendation where it was brought from. The recommendation then was correlated to the discussion and the consultation points.

When that was sent to the - by that time - deadlocked Parliament filibustering everything this was passed in a week. This was like the only bill they would not dare to block because there is already social consensus. Just like the Ministry of Education would agree that our use is fair-use any party who would say “no” does this against thousands of people who have already expressed their consensus on this matter. Unless the parliament can find additional facts or additional reflections that the working group has not considered there was really no reason for them to block this bill. So it was passed and made law.

The way we do work group meetings is again like we did in the Sunflower Movement. We separate the discussion points into the objective of the facts, the feelings of the facts, then the ideas (based on the Art of Focused Conversation by Brian Stansfield). We used a font that has six different bold levels to show the level of consensus of our discussion. With just one glance you can see what the consensus was about. Of the working group members maybe 10 people could participate here, we would take turns speaking for maybe 20 minutes about one section of the law and then we switch to the hackpad to the e-participation because not everybody lives in Taiwan or could travel to Taipei. So whatever they typed then becomes read aloud like an agenda for another 20 minutes. So the idea is that it’s like telepresence. You’re guaranteed to get a sufficient amount of time whether you are in the same place or whether your are on the Internet. Again everything that everybody says was captured with this my-society tool called “Say-it”. You can link any specific utterance to the statements in the bill.

On the design of the system, it was very important to overcome the three walls of the e-participation by first requiring the Ministry of Economy in this case to create a slide that is viewable within five minutes and we asked amateur people to try and read it and whether they could actually understand the issue at hand. For every specific jargon like “What is a closely held cooperation with a start-up” we provided a definition that was 140 or less characters long like Twitter and then just like in then Moe-Dictionary when you hover over that word you see a definition of the word. This is very important because most of the online-debate was about fighting for the definition of words. When we do a lexicon this way: ”A start-up means different things to different people but for the matter of this law it means this … Please talk in these terms.” This eliminates 80% of trolls (a deliberately offensive or provocative online post) for some reason.

The other innovation I think was that the initial mediator-group who interviewed the academics and stakeholders was designed to be one person from the elected official ministry’s office, one civic servant, who is not an elected official, and then one or two people from the private sector who has a stake in this like the Institute of Information Industries and one or two people from the civil society. This 4 to 6 people team were the initial team who decided the time, the duration, the format, the agenda of the entire consultation and this lends to a very balanced view and lends to the legitimacy of the entire process. Otherwise people would say this is just another way of lobbying or this is another way of protesting. We don’t want either of that.

The other thing of course that we do is a mixed reality debate. This is something very very technical but I think the most important of the four. We use our own discussion forum system. Unlike Facebook this allows a moderator to modify and to edit people’s comments. On Facebook when you’re facing a real comment or discussion or any other forum system 90% are good contributions but 10% are ad hominem attacks. All the editors are faced with a conundrum because if you censor that message then people would call you out and then people would be very angry. But if you allow that to continue the next reply tends to be 20% more than usual another attack and then the next one will be more and more toxic because people respond to the toxic parts. Then after five replies people started posting cat pictures and at that point all discussion is lost.

Before it gets to the cat picture point we created first a code of conduct and terms of service pop-up. Anybody who writes something on a site with a creative commons license you agree to not make these sort of comments. So when people post things like that we delete those 10% posts - usually just five words. Because it’s version controlled people could see the original if they want to. It is like Wikipedia. The normal people who join for the first time don’t have to have their mind polluted by that. Because trolls usually are from people who are just craving for attention. They have some gripe. So this is a way to teach people that only by making constructive criticism or constructive input do they get attention. Anything else automatically gets them zero attention.

Even if they write a long paragraph of just fighting words as long as there is one sentence in it that is a contribution we delete everything else and keep that sentence and then write a private message to them saying “Because you violated the code of conduct we deleted these and these words. This part is great, so we thank you for this contribution.” There is an old XKCD comic that says that the YouTube commentary system would vastly improve the discussion quality if the system read aloud back what the person just wrote. This is our way of reading back aloud the trolls. They reform very quickly and we get high-quality discussion. This is a very minor technology point but I think this is the one that made the most impact.

Could everyone change everything?

Actually no, it is just the facilitators, the initial team of maybe 10 people who built this website. As I explained they come from the three sectors. The logic is basically if you go on this website you have to trust the system operators anyway because it is cross sectoral.

We serve as each other’s peer reviewers. If any editor is going out of hand the other editors will call him out because they serve different interests. It’s like Wikipedia which is very legalistic. These behaviors are okay and these are not and we check based on a long running code of conduct that everybody sees on their first login.

Usually when I talk with civic tech people in other countries the first question they would ask: Is it okay that you are cross sectoral but you are privileging Internet elites. You are privileging people who can make interesting and useful arguments on the internet because that’s your selection of the working group. People having trouble to login into the Internet will be excluded by this process. What right do the Internet people have to decide for the rest of the population?

The usual response which we gave in our very first meeting in the hackathon is that we don’t. We only talk about laws that has Internet users as there only stakeholders. The idea is not deliberative democracy, the idea is a town hall meeting, as if the Internet users are a small town negotiating with the national government about things that concern this town’s development. So basically we don’t take any debate points like “gay marriage”. The jurisdiction ministry very much wants us to talk about gay marriage because they really cannot get consensus on this matter in Taiwan. We refused again and again because there is no correlation with gay and internet use. We cannot say that this is about internet stakeholders. We keep saying no. By the end of the deliberation period a lot more ministries see the power of this model, and we will keep saying no.

The other thing is that of course it is about law change. This is not about rallying things or just raising awareness about things. It has to be something concrete otherwise we don’t have a working group. Then again the ministries when they propose something they have two be open-ended. They cannot just propose a draft and ask the people to sign the draft. They could propose some seed ideas but that is it. It has to allow at least 30 days for a consensus to emerge. So we had a pretty good participation. This is very geeky slide that I usually skip. But we use tools that are mostly free of charge.

The forum system that we built is literally called The Civilized Discourse Construction Kit, it’s for civilized discussions. We asked the ministries to publish their introductory material as a slide and then the draft and other working group materials were kept in GitBook. The lexicon is kept on a Google spreadsheet. Then the entire web page was done on GitHubPages and we used different streaming providers and all the transcripts were kept on a … system.

Then just like any other campaign we sent every month the updates to the working group members and then we used pol.is for Uber and Airbnb which I will talk about. The core technology we use is called “The Focused Conversation Method”. This means that we layer our discussion in a way that talks about all the facts. Before we talk about all the facts we don’t about our feelings. And then we talk about everybody’s reflections. And before we complete that round we don’t talk about ideas. Then we talk about ideas and in the final and fourth layer about decisions.

The reason of that is because human beings have a cognitive limit called “the sticky choices”. When you start to think about ideas people lose the capacity to feel objectively about alternative ideas. One would start to be sticky to an idea. This way we always just focus on these two things. The other benefit of focusing on the facts and the feelings is that it avoids the legitimacy problem that other e-participations have. No matter how many people we could involve, tens of thousands or hundred thousands, it is just one person or 2% of the populace. If we do the voting part people would always ask: “By what right do you vote for the other people who don’t show up.” When we say we are just collecting facts and feelings for the people to talk about, for the civil servants to do research, then nobody asks this legitimacy question. So that is another angle of the tools.

So our improvements of the Cornell University Regulation Room format are for:
First all the ministries have usernames. Next they are all in the forum. So like the Ministry of Economy is literally @MOE, the Ministry of Finance is literally @MOF. So they have handles in that forum. They agree initially as a precondition that anybody who mentions them in any of their comments they will reply officially in seven days. So with a month of deliberation period that allows at least four rounds of back-and-forth on which the civil servants discover the peoples ideas on the platform.

For the civil society this is a way to get facts out of the civil servants. If you write them privately they don’t have the peer pressure of all the other ministries watching. So for some answers that are really face-losing like the one of the e-tax issues (Internet tax), we want to check the custom data for the regular importers and see whether it is possible to build a system that identified a regular importer who evades tax for example.

People asked: What kind of information do we currently collect to the destination address in the airport and in the seaports. It took seven days for the Ministry of Finance to answer: “Sorry we are just a tax collecting agency really. We are not really the Ministry of Finance. This is for the Office of Import-Export to answer.” We say, okay, so we tag that agency. So in seven days they had to answer: “We don’t have this system. We know Japan has it and Singapore has it too. It never occurred to us to build this system. Sorry.”

This is not something that you usually get neither from the parliamentary inquiries nor from the individual civic society lobbying or anything. It’s very face losing for them. Because they signed on this platform with this guarantee they will lose even more legitimacy if they don’t answer timely or honestly. So then after one or two times they started to buy in this g0v motto. It’s ok to be imperfect. “There is a crack in everything, that’s how the light gets in.” (Cohen) That creates opportunity for participation.

The working group members are just people who provide constructive opinions and then we focus really on online mixed mode conversation which lends another layer of credibility because a lot of policymakers and scholars and so on they type very slowly online - they are handicapped if you want to do it in an online only form. So we say: “Okay use your pen, use your voice, use whatever way you feel comfortable with and we will deploy sufficient stenographers or transcribers to bring it to the online space. So use whatever you’re comfortable with. So a lot of policymakers really go on this platform by printing every forum post and by writing their answers and have their assistant typing in their answers or turning the recordings back. There is a lot logistics involved but that is how we are getting to be inclusive.

So I think that’s it for Taiwan. The latest news is that they are trying to make this into a really official national all-inclusive supported by the local government kind of facilitation platform which is the agenda of the next administration. So this is how we get the spirit of deliberative democracy into policy-making.

Before I get to the last talk which I have 15 minutes for - which should be sufficient - are there any questions?

How did you manage to get ministries involved?

Technically this is very easy. Just fork the GitHub repository. Just add water.

Every ministry is different because we build this platform as a opt-in. So they don’t have to bring everything here. In fact to most of the things we say “no” because it is not about Internet people. So we built this very interesting multi-stakeholder dialogue platform that saves the civil servants time and face. Because the Taiwan civil servants at this time are very disadvantaged because they are not completely anonymous like in the UK. So if something goes wrong they get punished. On the other hand we are a young democracy. So whatever they get right the elected officials get all the credit. So this is a lose-lose situation. If they do something wrong they get punished, if they do something right somebody else gets the credit.

What we are trying to to do here is basically saying: All the impossible tasks that the elected officials assigned to the civil servants and they don’t have the research capacity for you can outsource this research two civil society and to the private sector. Because of the professional responses during the consultative period you as a civil servant get the credit over the elected officials for bypassing the parliament entirely. Then they become much more willing. The other benefit is - when the laws and regulations are done this way - that the elected officials in the parliamentary or the administration are not blocked. In many senses like in the early participatory budgeting in Taiwan the civil servants were the main proposers. I don’t know whether it is the same in Paris. A lot of the proposals were done by civil servants because they know what the city needs. This is a additionally legitimating way for their ideas to then become the public policy. So after opt-in and selling in these three angles we get all the very difficult cases like open data of the civil service know nothing about and so it’s very willing to share the risk and to retain the credit.

Whenever there is an issue the buy-in is instant. It takes a day or seven days at the most. We haven’t been forcing any ministry to use this method or anything. So we don’t know if there is resistance and what it takes but this is the opt-in our platform so hopefully we’ll never find out.

How do you verify their identities?

The deliberation is actually in the administration building, so first they have to have an email address with gov.tw which is very hard to get actually for a hacker (laughing).

And then we always do a multi-stakeholder preparatory meeting with the Ministry people. So we get to meet face-to-face and are confirmed that they are indeed the person holding the email address who wants to propose this thing. It is two factor authentication.

We have 10 minutes to talk about Uber. Uber is a challenge for the entire world. This is putting it lightly. The green dots are legal, the red are illegal but they are operating anyway, the pink are partially legal or in contention. The same applies for Taiwan and France (slide 71). For Uber we discovered that the vTaiwan existing process doesn’t work. The entire legitimacy was built on all the domestic stakeholders showing up, agreeing on the working group, so that nobody in the media or in the parliament could say: “No, this is not a consensus. They could not ignore us.”

But if we apply the same process on Uber they will probably ignore us. They are not a Taiwanese company. The don’t even have a Taiwan operation office. Even if we confiscate everything like the French people did in their few offices it would cost them nothing.

So a domestic agreement of the local stakeholders, the commissions, and the industry members would mean nothing even though we get all the right people. It won’t make a difference. We saw that happening in other Asian countries. To add insult to injury there is no ministry that wanted to propose Uber as a topic in vTaiwan. The Ministry of Transport who has been fining Uber for more than €1 million now which is a lot of money. They didn’t want to propose this because in their eyes they are criminals. To deliberate with criminals is nonsense. On the other hand the Uber lawyers say the Transport Ministry has nothing to do with this. We are a e-commerce company, they say, and this has got to do with the jurisdiction of the Ministry of Economic Affairs. So MOEA should propose the task. MOEA says: “This is not really our job and we don’t want the flak that comes from the Transportation Ministry.”

Then Jaclyn Tsai actually thinks that this task belonged to the Ministry of Finance because she cares about insurance and taxation. These two things are about Ministry of Finance but the Minister of Finance has never worked with anything like sharing economy before. They dont’ even want to propose this initial slide that shows the problem definition. The MOEA and the MOTC have a problem definition but they don’t have the willingness to speak it out.

Before vTaiwan got engaged it has been like a year. Whatever happened in France happened in Taiwan too - to a smaller degree. The taxi driver that surrounded the Ministry of Transport there was some kind of occupy going on, there was some kind of strike going on but at a lower scale. People generally think we have to talk about this but none of the ministries wanted to go to vTaiwan to propose this. The vTaiwan deliberators talked about our process and concluded that the consensus making process doesn’t even make a dent to Uber. Those are just the facts.

So what is our strategy? It was to introduce professional mediators that could link together the stakeholders not just domestically and not just the private and civil sector but literally all the drivers in Taiwan. We want to engage all the drivers in Taiwan to the vTaiwan process. This is a very different kind of populace in comparison to other domestic issues I talked about because Uber operates primarily in Taipei and Taichung city. We know that all the drivers basically use a mobile phone. The mobile phone penetration rate is 97 or 98%. So there is less of a representation problem if we would require their participation online.

But the problem with taxi drivers or other drivers is that they don’t really have one minute to look through the slides and make informed decisions or whatever. They were driving or making a business out of driving. So we have two lower the participation threshold. It was one minute but it has to be five seconds or less because otherwise we don’t engage these people. We even had a way of saying it. It must be engaging within a red light of time. So of course we don’t actually encourage people to vote while driving. So when they engage in the vTaiwan process we encourage them to park first. But in any case we want to engage a very different kind people, their passengers, the drivers that own the car that only have the mobile phone. We needed professional mediators and we needed to organize with the association of all the taxi drivers that they send the drivers a sms to go onto the vTaiwan website at the same hour. The same with the Uber drivers and all the other stakeholders they had to all get the same URL on the same hour of the day. The message says: Whatever you say here will become the agenda after a month from now on for all the stakeholders to talk about.

So this actually worked. What they saw when they used their phone on a stoplight were just four very simple sentences saying: “This is bottom up. Everybody wants to talk about Uber and Airbnb and Bitcoin. We have one specific question. Is it ok for private non-professional license people two carry passengers and charge them for it?” So this was a very narrow thing. We didn’t talk about sharing economy. Then we say: “A month from now we will have a national debate. We will invite Uber but the agenda is decided by you guys.” And then the open data is published for independent analysis and we guarantee deliberation.

The interface they see on their phone is this very simple thing. They see one random sentiment: “I feel” or “I think” from their fellow citizens (slide 73) “Passenger insurance is very important” and then they they take one second to press “I agree” or “I don’t agree”. That is all we ask of them. They press “agree” or “disagree” and their avatar changes in this landscape. This is a open space technology simulation online. So basically people stand in their different positions and as you say “yes” or “no” you move in your positions. If you login with Facebook you see your friends hats. Initially there are four different groups. There were drivers taxi drivers, Uber passengers, other passengers and then your task is to convince people neighboring to you to accept your view, Then the system only shows the views with the highest consensus within a group. So this actually worked.

This is what we call “dimensionality reduction”. We have two primary dimensions. Based on the “yes” or “no” the two most contentious topics. The one that is most contentious we use as the x-axis the second is a y-axis. That changes as people propose more sentiments for other people to vote with. So it is a very dynamic thing. We said that this would go on for a month. Then we promised to publish independent analysis material.

So what happened on that hour when the four speaking groups formed of equal people. This is very important because if we didn’t get the URL to them on the same hour you would have gone online and would have seen everybody is a Uber driver and that would have turned everybody down. So it was very important that everybody got on at the some time. Then the consensus at that time of the two groups first said: “We don’t negotiate with criminals.” The second says: “Even if there are a lot of taxis in the street I will still call Uber.” They are both very radical.

That is our first primary factor. If you multiply the numbers are both minorities. None of this had majority across all populace. These were actually minority opinions. They were just majority within their small group. The system rewards only arguments to gain support from your group people. People were forced to invent more moderate ideas to gain consensus from people who think like them.

So after about a week the group one then became: “This is not about Uber. This is about the Ministery of Transport which has a duty to fine any unlicensed driver. The fact that they do this to other people and not for Uber is maybe a problem. It is their duty nevertheless.” That was more moderate and gained more support. Group two evolved to: “All the Tsaiwan taxis in the large cities have to join one fleet or another. This is a corporate thing that limits the choice of taxis. Uber provides an opportunity for a driver to join multiple fleets. This is a innovation.” Just by saying this they gained 2% of the populace. Some taxi-driver jumped to this group just by reading this sentiment.

After a week or so different the four groups started to form into two broadly speaking pro- and anti-Uber groups. Another week has passed before we had what we call a majority consensus which is something that everybody regardless of the 4 or 2 groups agree upon. It starts with some very general reflections because we ask numerous questions like: “The laws will change with time.” Most people who saw that agree with that. Or for example: “Although there are many important topics the security of passengers is the most important one.” Everybody agreed with that.

Then after another week people started to propose concrete solutions or ideas to garner cross group support. This one from Ivan from Mozilla Taiwan, a Firefox developer, said: “We should introduce the same five-star rating system to order taxis because the thing that ensures Ubers quality is its rating system. It is not anything else. So if the government mandates all the independent taxis and fleets to introduce the same system we can get the same quality and then Uber doesn’t become a problem.” Everybody agreed to that even Uber themselves.

The initial statement: “We don’t argue with criminals” which initially had 70% agreement eventually got less and less support. By the third week most people - over 65% - still thought they are criminals but we should still sit down and talk with them. By the end of the polling period most people think: “Okay, let’s talk with them anyway.” So by the fourth week and this is why we need four weeks because experience tells us on the fourth week people finally agree on something that’s actionable. Before that it is just random reflections or general observations. After trying to convince each other so hard because the system only shows the things with the highest score people started to come up with very nice ideas, actually policy ideas like that: “You should be fair to both Uber and non-Uber drivers” or like: “The taxation is very important. Uber must register, their registration must display prominently on their window, or in their cars” or “This is not just a commerce because like medicine and food this is matter of public safety and if people want to avoid tax, if you really want ride sharing you should insure the people. If you go to work and go back to work they should only ride two times. That is okay to evade tax like that. But if you take more than two routes then of course you are actually making a living out of it and you’re a business person.” So that’s one policy suggestion. And: “The taxi driver should be allowed to join multiple platforms, even existing taxi drivers.” These are the suggestions that had over 80% consensus which is our cut-off point and that became the agenda of our deliberation.

Before we did the deliberation we did a comparative analysis using this data and then showed what other countries are doing based on the six demands. So that was it. We showed all those consensus. We asked each stakeholder whether you want to compromise or not. So it is like a progress bar, it is actually written like a progress bar on a white-board. Uber said eventually: “Yes, we will provide insurance terms to the Taiwan government” which they haven’t done to any other non-us country. The taxi fleet says that if you open search pricing then we will introduce the new class of taxis that competes directly with Uber. Then the Minister of Transport said: “Okay we will do that.” Then the taxi association of independent drivers said: “The main problem is that Uber takes a 20% cut. If they take just a 2% cut we will work with them tomorrow.” So it becomes a matter of negotiation at the table.”

So basically the progress bar says: Uber is illegal because of these six things. On that meeting we extracted the promises of 3-4 things. Everybody sees that it’s still illegal because it doesn’t satisfy the other two of the 6 demands that we demand of them. So there is no arguing so the parliament doesn’t do anything. If they agree it becomes a legal company and if it doesn’t it is still illegal.

That’s how we got the three ministries to be not afraid of losing your face and putting the one thing that they care about into the initial pol.is conversation and deliberation and to let people vote and then we get taxi drivers and the associations and the civil society people into this shared space. The additional good thing about this was that the Airbnb people were watching the live stream all the time, the entire process they were watching very closely. They knew actually before us, we worked with academics on the methodologies of how to set up the pol.is issues. We designed three questions about what kind of people you are whether you have a professional driver license or whether you are a taxi driver, have you taken Uber before and about the thing that each ministry wanted to hear about taxation, insurance and safety. Then we wanted to hear about the things the government could actually do something about. Those were our initial questions, but it was heuristic.

But Airbnb figured it out very quickly. They knew that we were going to use Wikipedia data to show the timeline. They then went and edited Wikipedia or people who had sympathy for them. I don’t have any proof if it was an Airbnb employee. When we started the pol.is again with this idea they sent an email to all Taiwan members saying please show your support for Airbnb on this rulemaking platform. Of all the people who have used Airbnb before in Taiwan there are three groups: One group said it has to satisfy some laws still, one group said the landlord must have duties, and one group said at least the quality has to be guaranteed, if it’s just your home then it has to be said so, if it’s actually a legal hotel chain or one person has 30 different homes and each of them has the same photo then it has to be outlawed and Airbnb must answer to this. People also said: “You know this is the internet, the government should stay out of it.” That is the minority actually.

Airbnb sent an email to all its Taiwan members. Before that we had like 3000 people engaged. Then after that it was like 10 times as many people involved. So this is unfair because most of Taiwan people are not Airbnb members. It is just that they are so good at mobilizing. So we said exactly that saying we respect the people who are Airbnb members but they shouldn’t get more representation than people who are not members just because Airbnb is so good at mobilizing because we have our statistics showing how many people in Taiwan have used Airbnb.

So the other group of people who don’t or haven’t used Airbnb before is at least equally important. So these four merged with those three and they became our agenda. The main contention was that we discovered that most people think of Airbnb as a good idea especially for foreigners visiting Taiwan. But if they themselves are visiting other cities in Taiwan, they will not use Airbnb.

The same day we can go from being super public to exchange our service for money to voluntary sector work and so on. This a way for different sectors to meet somewhere and to extract promises out of each other I think is generally useful. It also provides a … and gives the so-called clicktivists - at the beginning of my talk I said these are people who just spent time to click “like” or “unlike” in the internet - something else to do. If they have 10 seconds more time they can share our deliberation link, if they have more time they can engage in questioning which gets a granted answer and then if they have more time and a laptop computer they could do a meaningful discussion, and then if they even have more time they can do a deliberation. And if they have some time to attend a deliberation for two hours they can participate the agenda setting.

So that’s my talk. Thank you very much!