1 00:00:02,100 --> 00:00:15,680 rc3 hacc preroll music 2 00:00:15,680 --> 00:00:21,480 Herald: Welcome to the hackrf channel. Today under the title Patching Democracy. 3 00:00:21,480 --> 00:00:28,130 Today we're talking about the era of digitalization, as well as about the 4 00:00:28,130 --> 00:00:32,780 understanding of the enormous importance of digital tools in both private and 5 00:00:32,780 --> 00:00:41,129 public life, because we need it to reduce the world's complexity to an amount that 6 00:00:41,129 --> 00:00:46,345 we can actually handle. This is something that is very important, for example, in 7 00:00:46,345 --> 00:00:51,809 democracies. Especially when talking about decision making, like for example, the 8 00:00:51,809 --> 00:00:56,140 voting advice application that we have in Germany, the Wahl-o-mat is a very good 9 00:00:56,140 --> 00:01:03,520 example of making parties comparable to common people all over the place. But 10 00:01:03,520 --> 00:01:08,150 those machines are like those Wahl-o-mats are very expensive and thus they are only 11 00:01:08,150 --> 00:01:15,080 available for larger elections. And this is a problem that is actually handleable. 12 00:01:15,080 --> 00:01:19,930 This is what Till Sander does with his voting advice application that is 13 00:01:19,930 --> 00:01:28,848 called... wait a second 14 00:01:28,848 --> 00:01:32,637 with his open election compass and he actually was approached by the 15 00:01:32,637 --> 00:01:38,552 small city of Lüdenscheid to develop something that could actually do something 16 00:01:38,552 --> 00:01:42,710 that the Wahl-o-mat also does. And when he found out that this is actually something 17 00:01:42,710 --> 00:01:46,825 that needs to be provided also for smaller elections and that is actually affordable, 18 00:01:46,825 --> 00:01:54,000 he is actually a web designer, decided to do it in a bigger way. So he wanted to 19 00:01:54,000 --> 00:01:59,132 make it open source and thus created this platform that he now talks about in his 20 00:01:59,132 --> 00:02:06,091 lecture that we provide to you right now. You can also ask questions that will be 21 00:02:06,091 --> 00:02:13,050 answered in the following Q&A and on Twitter, as well as on the IRC. Under the 22 00:02:13,050 --> 00:02:21,980 hashtag rc3hacc and the channel rc3-hacc. Now enjoy the talk. 23 00:02:21,980 --> 00:02:28,430 Till Sanders: Welcome to Patching Democracy. This is a short introduction to 24 00:02:28,430 --> 00:02:32,731 applications like the German Wahl-o-mat and why we might not need to hack 25 00:02:32,731 --> 00:02:38,140 democracy. In this talk, I will demonstrate how we can improve elections 26 00:02:38,140 --> 00:02:44,907 and political education everywhere, thanks to free and open source software. 27 00:02:44,907 --> 00:02:49,806 Researchers of the Friedrich-Ebert- Foundation found that only a third of the 28 00:02:49,806 --> 00:02:56,790 population believes in a brighter future. Less than half of our society is satisfied 29 00:02:56,790 --> 00:03:01,621 with how our democracy works. In parts of Germany, this drops even further to about 30 00:03:01,621 --> 00:03:10,422 a third. Even worse, three out of four Germans feel like politicians don't care 31 00:03:10,422 --> 00:03:18,470 about their concerns. And lastly, many people even agree that it doesn't make a 32 00:03:18,470 --> 00:03:25,989 difference which parties form the government. Studies like these question 33 00:03:25,989 --> 00:03:33,450 the state our democracy is in. Is our democracy broken? Let's take a look at 34 00:03:33,450 --> 00:03:41,930 some other results of the same study. Only 1.3 % want an authoritative figure with 35 00:03:41,930 --> 00:03:47,792 extensive powers to make the law. With the rise of the extreme right this is a good 36 00:03:47,792 --> 00:03:54,770 thing. While 88% of us think that politicians make more promises than they 37 00:03:54,770 --> 00:04:03,735 can keep, the majority acknowledges that politicians do have a difficult job. And 38 00:04:03,735 --> 00:04:08,769 out of several problems, a great majority identified a lack of participation in 39 00:04:08,769 --> 00:04:16,828 elections as the biggest problem. Our democracy generally fails to make everyone 40 00:04:16,828 --> 00:04:23,490 happy and to be fair, that's somewhat the point. But while many people have issues 41 00:04:23,490 --> 00:04:28,069 with our democracy, they also seem to believe that it is still the way to go. 42 00:04:28,069 --> 00:04:34,206 Democracy is not broken. It is just our implementation of it that is experiencing 43 00:04:34,206 --> 00:04:41,699 technical difficulties. Hi, my name is Till, and I'm here to fix this. Not alone, 44 00:04:41,699 --> 00:04:48,479 of course, but I'm happy to be, you know, someone just doing what he knows best and 45 00:04:48,479 --> 00:04:54,690 I like all those numbers might suggest I'm not even a political scientist. I'm, in 46 00:04:54,690 --> 00:05:00,900 fact a designer and web developer. And as such what I enjoy most is the challenge of 47 00:05:00,900 --> 00:05:07,849 making complex concepts easily accessible, preferably with beautiful user interfaces. 48 00:05:07,849 --> 00:05:14,135 I'd like to first introduce you to the idea and short history of voting advice 49 00:05:14,135 --> 00:05:19,229 applications. We will then dig in a little deeper and establish important principles 50 00:05:19,229 --> 00:05:23,860 that make VAAs successful. There's also going to be a little hands on with the 51 00:05:23,860 --> 00:05:28,861 FOSS project I have developed in the last year. Once I've shown you the tools, I'll 52 00:05:28,861 --> 00:05:32,460 talk about how you can run your own election compass and what to consider when 53 00:05:32,460 --> 00:05:41,770 doing so. And off we go! Our story begins in the Netherlands. In 1989, the Dutch 54 00:05:41,770 --> 00:05:45,940 Citizenship Foundation, the documentation center of Dutch political parties and the 55 00:05:45,940 --> 00:05:50,490 faculty of Political Management of the University of Twente start a collaboration 56 00:05:50,490 --> 00:05:55,160 to develop the Stemwijzer. A booklet containing 60 statements found in the 57 00:05:55,160 --> 00:06:02,182 programs of political parties and a diskette. Well, it's 1989. The idea proves 58 00:06:02,182 --> 00:06:05,649 popular and evolves to the first Internet election compass for the Dutch 59 00:06:05,649 --> 00:06:14,125 parliamentary elections in 1998. Although the project can only attract 6500 voters, 60 00:06:14,125 --> 00:06:19,994 subsequent implementations in 2002 and 2003 attract about 2 million voters, which 61 00:06:19,994 --> 00:06:27,020 quickly become 5 million voters in 2006, which is about a third of the entire Dutch 62 00:06:27,020 --> 00:06:35,275 population. Success began spreading to other countries, the first election 63 00:06:35,275 --> 00:06:41,276 compass I myself ever came across was the German Wahl-o-mat, based on the StemWijzer 64 00:06:41,276 --> 00:06:47,180 itself, the Federal Agency for Civic Education, the BPB, released the first 65 00:06:47,180 --> 00:06:57,060 Wahl-o-mat in 2002. It's fair to say that the VAA concept is now well-established in 66 00:06:57,060 --> 00:07:03,528 Germany and other countries. Usage in Germany has increased to 33% of cast votes 67 00:07:03,528 --> 00:07:10,000 in 2017. Think about that for a moment. One in three voters has used the Wahl-o- 68 00:07:10,000 --> 00:07:15,669 mat at some point before going to the ballot. As software projects of the German 69 00:07:15,669 --> 00:07:20,529 government go, this might well be the most successful yet, doubly so if you consider 70 00:07:20,529 --> 00:07:28,099 the costs of some spectacular failures in the past. So what did the first voting 71 00:07:28,099 --> 00:07:32,809 advice applications actually look like? Let's take a look at the first Wahl-o-mat 72 00:07:32,809 --> 00:07:37,400 from nearly two decades ago. The Internet was quite different back in those days. 73 00:07:37,400 --> 00:07:42,370 Many user interface patterns were yet to be discovered or refined and users were 74 00:07:42,370 --> 00:07:47,909 less experienced. On a side note. Technically, this website from 18 years 75 00:07:47,909 --> 00:07:53,849 ago still runs perfectly fine in a modern day browser. Web technologies are amazing. 76 00:07:53,849 --> 00:07:59,247 Anyway, despite these slight difficulties and the Wahl-o-mat being a new concept, 77 00:07:59,247 --> 00:08:05,334 there are very few instructions. This is because the core concept was and still is 78 00:08:05,334 --> 00:08:12,885 incredibly simple. You are presented with a sequence of statements or a thesis. You 79 00:08:12,885 --> 00:08:18,499 can choose to approve or reject or remain neutral to a thesis. If you don't really 80 00:08:18,499 --> 00:08:24,887 understand the meaning or the issue behind it, you can also skip a thesis. After 81 00:08:24,887 --> 00:08:29,362 about 30 statements, you can choose categories that are more important to you, 82 00:08:29,362 --> 00:08:35,709 so they are counting double. The political parties or candidates answer the same 83 00:08:35,709 --> 00:08:41,415 theses. At the end your answers are compared to those of the parties showing 84 00:08:41,415 --> 00:08:50,880 you potential matches. Fast forward to today. The idea is about 30 years old now. 85 00:08:50,880 --> 00:08:55,940 In this time it spread not only to Germany but also to Belgium, Finland, Denmark, 86 00:08:55,940 --> 00:09:02,990 Portugal, Norway, Sweden, Austria, Switzerland and many other countries, 87 00:09:02,990 --> 00:09:08,850 continents even. The teams behind StemWijzer inspired most European 88 00:09:08,850 --> 00:09:11,787 countries and others around the world. 89 00:09:17,940 --> 00:09:22,297 Let's talk about VAAs in more depth. What 90 00:09:22,297 --> 00:09:29,161 are they actually good for? Why do people use them? How do their mechanisms work and 91 00:09:29,161 --> 00:09:38,200 does their popularity make them dangerous? So what do we actually want to achieve? 92 00:09:38,200 --> 00:09:46,180 What is the purpose of voting advice applications? Since their inception, the 93 00:09:46,180 --> 00:09:51,820 target group are actually young, even first-time voters. I guess the reasoning 94 00:09:51,820 --> 00:09:57,390 behind this is that the older people get the more experienced they are with the 95 00:09:57,390 --> 00:10:03,406 political landscape, or at least they should be. The term voting advice 96 00:10:03,406 --> 00:10:10,227 application suggests that the purpose is to advise users on who to vote for. Now, I 97 00:10:10,227 --> 00:10:15,460 must say, I've been struggling with this name. I find it counterintuitive because 98 00:10:15,460 --> 00:10:20,763 from what I've seen, this is actually not the purpose of these projects. And that's 99 00:10:20,763 --> 00:10:26,000 good because imagine for a second what this would mean. Many VAAs are designed 100 00:10:26,000 --> 00:10:31,910 and controlled by government agencies. So who would want to live in a democracy 101 00:10:31,910 --> 00:10:36,803 where the government gives you advice on who to vote for in the election? So 102 00:10:36,803 --> 00:10:42,317 although it's called a voting advice application, the Wahl-o-mat does not 103 00:10:42,317 --> 00:10:47,177 actually want to give you advice. It's even written there on the very first page, 104 00:10:47,177 --> 00:10:54,870 right above the start button. The Wahl-o- mat is not voting advice, but an offer of 105 00:10:54,870 --> 00:11:02,920 information about elections and politics. I found this disclaimer in every VAA I 106 00:11:02,920 --> 00:11:11,189 have come across so far. OK, so the purpose of VAAs is, despite their name, 107 00:11:11,189 --> 00:11:21,606 not to give voting advice. Good. Except, they kind of do, don't they? We don't give 108 00:11:21,606 --> 00:11:26,620 you any advice well, we do, but don't take our word for it. We've warned you not to 109 00:11:26,620 --> 00:11:35,070 take this as an advice. Now go ahead and get not-an-advice. Maybe it's just me, but 110 00:11:35,070 --> 00:11:42,722 I think this is quite german. VAAs have a positive impact on political education. 111 00:11:42,722 --> 00:11:48,753 This might be the main aspect, they have originally been designed for. VAAs want to 112 00:11:48,753 --> 00:11:53,950 have a positive impact on political education. As I understand it, this topic 113 00:11:53,950 --> 00:11:59,650 sadly needs more research. But with the research done so far, we can assume that 114 00:11:59,650 --> 00:12:05,080 this is indeed the case. It appears to be uncertain to what extent exactly, and this 115 00:12:05,080 --> 00:12:11,010 will also depend on the individual VAA. But there is a positive impact. VAAs do 116 00:12:11,010 --> 00:12:15,660 not improve the knowledge about political structures like how the ballot works, how 117 00:12:15,660 --> 00:12:20,350 the allotment of seats in parliament works, etc. But they can improve knowledge 118 00:12:20,350 --> 00:12:25,004 about the policy issues, what the upcoming election is about, what parties there are, 119 00:12:25,004 --> 00:12:30,820 and where the parties stand. VAAs also lead to discussions about these issues and 120 00:12:30,820 --> 00:12:36,970 parties which can also improve political knowledge in peer groups. So as far as we 121 00:12:36,970 --> 00:12:42,649 know today, this claim is true and it is an important benefit of election compasses 122 00:12:42,649 --> 00:12:48,300 or VAAs, because as research shows, most people in Germany are able to place 123 00:12:48,300 --> 00:12:53,180 parties on the spectrum of left and right correctly. But at the same time, many 124 00:12:53,180 --> 00:12:58,567 people are unable to place parties correctly when it comes to policy issues. 125 00:12:58,567 --> 00:13:03,530 So missing political knowledge and misinformation can actually lead to people 126 00:13:03,530 --> 00:13:13,498 voting against their own interests. VAAs promote electoral participation. What 127 00:13:13,498 --> 00:13:17,690 makes people vote? To answer this question, we can take a look at the 128 00:13:17,690 --> 00:13:23,210 reasons why some people don't. And one of the main reasons why some people don't 129 00:13:23,210 --> 00:13:28,470 vote is because they don't feel like their position is reflected by any of the 130 00:13:28,470 --> 00:13:35,352 existing parties. Our political system is complex and our political landscape, our 131 00:13:35,352 --> 00:13:41,320 parties and their programs doubly so. People that have a better understanding 132 00:13:41,320 --> 00:13:47,976 and more knowledge of the choices they have… are more likely to cast a vote. Just 133 00:13:47,976 --> 00:13:52,924 imagine you're helping a friend who has no clue about computers decide on a graphics 134 00:13:52,924 --> 00:13:59,346 card. They either get confused as hell very quickly or they be like, well, I 135 00:13:59,346 --> 00:14:03,563 don't know, do I even need one? My laptop runs fine and it doesn't have one. Can we 136 00:14:03,563 --> 00:14:10,020 get pizza now? You see, being able to make an informed decision can make a huge 137 00:14:10,020 --> 00:14:17,030 difference. And VAAs can help with that. And research tells us that VAA users can 138 00:14:17,030 --> 00:14:25,390 be 2% to 12% more likely to go to the ballot. The last important background 139 00:14:25,390 --> 00:14:30,940 topic I would like to touch on is the matching algorithm. These algorithms are 140 00:14:30,940 --> 00:14:36,162 still subject to debate and some are frequently criticized. I'll spare you the 141 00:14:36,162 --> 00:14:41,000 history, and instead will jump right in because, one, this topic deserves a talk 142 00:14:41,000 --> 00:14:47,320 of its own. And two, that talk should not be held by me. But I'll share with you 143 00:14:47,320 --> 00:14:53,132 what I know. The matching algorithm is responsible for calculating your result. 144 00:14:53,132 --> 00:14:59,808 After you answered all the theses, your answers are compared to those of the 145 00:14:59,808 --> 00:15:06,876 parties. The parties get more points the more you agree with them. Sounds simple? 146 00:15:06,876 --> 00:15:15,848 But how do you calculate this exactly? Say we have an agreement scale of 10 to -10. I 147 00:15:15,848 --> 00:15:23,402 reply to a different thesis with an agreement of 3. Party A is even more into 148 00:15:23,402 --> 00:15:33,216 this and goes for 9. Party B is not a fan of this thesis and answers with -3. How 149 00:15:33,216 --> 00:15:38,904 many points will Party A and B get for this thesis? There are two approaches to 150 00:15:38,904 --> 00:15:44,560 this. The first has been coined the proximity model, and as the term suggests, 151 00:15:44,560 --> 00:15:51,510 it focuses on the distance between two points. In this case, Party A and Party B 152 00:15:51,510 --> 00:15:57,871 are the same distance to my answer, so they will get the same amount of points. 153 00:15:57,871 --> 00:16:03,630 Seems logical at first, but is this really the best approach to this? I might not 154 00:16:03,630 --> 00:16:08,533 fully agree with Party A, but I am on the same side, whereas party B is on the other 155 00:16:08,533 --> 00:16:15,119 side. Wouldn't it be safe to assume that party A is a better match for me? Well, 156 00:16:15,119 --> 00:16:23,970 probably, yes. The idea is called the directional model. It awards more points 157 00:16:23,970 --> 00:16:31,679 if the voter and the party go in the same direction. In our scenario, party A will 158 00:16:31,679 --> 00:16:39,126 receive more points than party B because it is on the same side as I am. Following 159 00:16:39,126 --> 00:16:45,867 these models, one can easily create a matching algorithm. Why not all VAAs make 160 00:16:45,867 --> 00:16:55,525 their algorithm's public, there are a few well known ones. The first is the famous 161 00:16:55,525 --> 00:17:01,379 city block algorithm. It belongs to the proximity model and is still used by the 162 00:17:01,379 --> 00:17:08,159 Wahl-o-mat albeit with only three options, which has been criticized in the past. In 163 00:17:08,159 --> 00:17:13,551 this chart, you can see the users answer in the rows and the party's answer in the 164 00:17:13,551 --> 00:17:21,191 columns. In this area where they meet, you'll find the score for this thesis. For 165 00:17:21,191 --> 00:17:26,817 example, if I choose to strongly approve the thesis, I'm in the first row. If the 166 00:17:26,817 --> 00:17:32,200 party agrees with that we meet in the first cell and the party gets the maximum 167 00:17:32,200 --> 00:17:39,820 score of 1 for this thesis. If the party, however, rejects the thesis, 1st row, 4th 168 00:17:39,820 --> 00:17:48,110 column, it will only get a score of -5. You see, the city block algorithm strictly 169 00:17:48,110 --> 00:17:54,159 follows the proximity model: the closer user and party become, the higher the 170 00:17:54,159 --> 00:18:02,360 score. The classic example of the direction and what is the scalar 171 00:18:02,360 --> 00:18:07,590 algorithm, the direction or side is far more important here and a party cannot 172 00:18:07,590 --> 00:18:13,600 receive a positive score as long as it is on the other side of the user's opinion. 173 00:18:13,600 --> 00:18:19,219 Note also that this must mean that if either the user or the party choose a 174 00:18:19,219 --> 00:18:27,758 neutral position, the score will always be zero. So why both algorithms have their 175 00:18:27,758 --> 00:18:34,559 strengths, our goal is to find a model, thats prediction is as close as possible 176 00:18:34,559 --> 00:18:40,929 to what the user votes for in the end. And there's another group of algorithms that 177 00:18:40,929 --> 00:18:50,460 tend to yield better results. I'm talking about hybrid algorithms that try to 178 00:18:50,460 --> 00:18:56,180 combine the approaches of the proximity and directional model. As you can see, 179 00:18:56,180 --> 00:19:02,771 proximity as well as direction play a role in the scoring. Looking at the colors, you 180 00:19:02,771 --> 00:19:08,311 can see that this now looks a bit like the first algorithm, the city block algorithm, 181 00:19:08,311 --> 00:19:14,230 but the green line fades a little in the center. This is the influence of the 182 00:19:14,230 --> 00:19:25,945 scalar algorithm focusing on the direction. So which one is best? I'm 183 00:19:25,945 --> 00:19:32,549 afraid we don't know for sure. As always, data will tell. It also depends on your 184 00:19:32,549 --> 00:19:38,649 intentions and design choices. What we do know is that algorithms based on the 185 00:19:38,649 --> 00:19:43,720 proximity model tend to favor temperate parties, while those based on their 186 00:19:43,720 --> 00:19:51,206 directional model gently pushes users to the extreme ends. Considering this, hybrid 187 00:19:51,206 --> 00:19:58,299 algorithms should yield more balanced results. We must not forget, though, that 188 00:19:58,299 --> 00:20:04,129 at the end of the day they are still only models, so don't expect any of them to be 189 00:20:04,129 --> 00:20:14,590 incredibly accurate. The OpenElectionCompass is a free and open source software 190 00:20:14,590 --> 00:20:20,899 with a simple mission: making voting advice applications available to every 191 00:20:20,899 --> 00:20:28,218 election to support political education and democracy everywhere. Running your own 192 00:20:28,218 --> 00:20:34,680 election compass can be a costly endeavor. No more. With the OpenElectionCompass, we 193 00:20:34,680 --> 00:20:41,080 have a system that is free, transparent, user friendly and accessible. Unlike 194 00:20:41,080 --> 00:20:46,440 agencies who only run an election compass every few years, a project like this can 195 00:20:46,440 --> 00:20:53,200 focus on continuous improvement. But enough promises. Let's take a look at the 196 00:20:53,200 --> 00:21:00,072 features. The OpenElectionCompass was designed to be easy to use and accessible. 197 00:21:00,072 --> 00:21:06,872 The design is minimalistic, so users are not distracted. Unlike solutions like the 198 00:21:06,872 --> 00:21:14,241 Wahl-o-mat, it makes use of the available screen size with big theses and buttons. 199 00:21:14,241 --> 00:21:19,489 Clear color coding provides visual feedback, and as you can see, the 200 00:21:19,489 --> 00:21:26,128 navigation is not based on previous next buttons, but scrolling. This is quicker 201 00:21:26,128 --> 00:21:31,730 and far more intuitive, especially on mobile devices. To make this experience 202 00:21:31,730 --> 00:21:37,159 even smoother, a big, friendly green button helps guide the user through the 203 00:21:37,159 --> 00:21:43,330 entire process. Whenever it's time to move on, it just pops up ready to take you to 204 00:21:43,330 --> 00:21:50,159 the next step. The OpenElectionCompass is the first VAA software to pioneer this 205 00:21:50,159 --> 00:21:57,369 navigation concept. With great success, I might add. Everything you're seeing here 206 00:21:57,369 --> 00:22:01,650 is also screen reader and keyboard friendly. These things get easily 207 00:22:01,650 --> 00:22:06,999 overlooked. But as I said, being a continuous project, we can focus on 208 00:22:06,999 --> 00:22:14,820 important details like these. After we have finished with the thesis, we are 209 00:22:14,820 --> 00:22:20,611 guided to the selection of the parties. Notice that unlike most VAA's, there's no 210 00:22:20,611 --> 00:22:25,559 additional step here where we would be asked to select some theses that are more 211 00:22:25,559 --> 00:22:30,779 important to us. This is another way OpenElectionCompass is improving and 212 00:22:30,779 --> 00:22:35,799 speeding up the process. We are removing the cognitive overhead of going through 213 00:22:35,799 --> 00:22:43,519 all these theses again. Instead, you can mark the thesis as important, right while 214 00:22:43,519 --> 00:22:49,986 you're answering. Usability wise, this makes a lot more sense. But back to the 215 00:22:49,986 --> 00:22:56,020 parties. We simply select the parties we would like to compare. Again, keyboard and 216 00:22:56,020 --> 00:23:03,059 screen reader friendly, and proceed to the results. Here we have the classic 217 00:23:03,059 --> 00:23:11,037 percentage based result view. With most VAAs, this is pretty much it. Usually you 218 00:23:11,037 --> 00:23:17,068 can go into the thesis one by one and see the statements of the parties, but I 219 00:23:17,068 --> 00:23:23,559 believe this is the most important part and should not be hidden away. Showing 220 00:23:23,559 --> 00:23:31,749 these statements should be the default. So I made it the default, when we scroll 221 00:23:31,749 --> 00:23:38,030 further, we can read the answers of all selected parties in an easy, color coded, 222 00:23:38,030 --> 00:23:49,129 chat like format. And that's it. A simple to use, accessible, beautiful state of the 223 00:23:49,129 --> 00:24:02,185 art and free voting advice application. Now comes my favorite part. How do you get 224 00:24:02,185 --> 00:24:08,540 all this content, theses, the parties, the statements into the OpenElectionCompass? 225 00:24:08,540 --> 00:24:14,220 Well, of course, by using a big JSON configuration file, that's hardly 226 00:24:14,220 --> 00:24:21,131 exciting, but you know what, JSON is simple, but for a non-programmer, this is 227 00:24:21,131 --> 00:24:26,720 a pretty daunting task. And even for programmers working with big JSON files to 228 00:24:26,720 --> 00:24:33,868 manage content in multiple languages is not something particularly fun. Especially 229 00:24:33,868 --> 00:24:40,399 if it involves countless emails back and forth to incorporate small changes. So 230 00:24:40,399 --> 00:24:48,321 guess what? There's a tool for that. Now everyone can read and write the JSON 231 00:24:48,321 --> 00:24:52,326 configuration files using a friendly visual editor called the configuration 232 00:24:52,326 --> 00:25:00,073 editor. It makes adding parties, thesis, and statements a breeze. Simply fill in 233 00:25:00,073 --> 00:25:05,077 the forms and download your ready to go configuration file. It even supports 234 00:25:05,077 --> 00:25:11,789 adding every content in multiple languages and handles images for you. There is 100% 235 00:25:11,789 --> 00:25:17,609 feature parity between the configuration files and the editor. And while this is 236 00:25:17,609 --> 00:25:23,789 only the first step in making the creation of VAAs more accessible, it is a big step 237 00:25:23,789 --> 00:25:33,850 up from any other tool. And there's more. Let me introduce some of the smaller 238 00:25:33,850 --> 00:25:40,799 features that make the OpenElectionCompass special. I really want to make 239 00:25:40,799 --> 00:25:46,470 this technology accessible for everyone, so I took the time to create a single file 240 00:25:46,470 --> 00:25:51,990 deployment solution that would fit the software, your content and images, all in 241 00:25:51,990 --> 00:25:59,014 a single HTML file. Is that the most performant solution? No, but let's be 242 00:25:59,014 --> 00:26:08,820 reasonable. It's perfectly fine for a small town – Definitions: Theses must be 243 00:26:08,820 --> 00:26:14,629 short and precise. Sometimes this makes it difficult for users to understand them 244 00:26:14,629 --> 00:26:20,179 because of words or abbreviations they might not know. To help with this, you can 245 00:26:20,179 --> 00:26:29,210 easily provide small helpers – Solid navigation: As we have seen, the one page 246 00:26:29,210 --> 00:26:34,989 design approach comes with lots of benefits. To make sure no one gets annoyed 247 00:26:34,989 --> 00:26:40,569 by too much scrolling around, an intelligent menu is always right at hand. 248 00:26:40,569 --> 00:26:47,780 That and the big green button helps getting around and no time at all. 249 00:26:49,350 --> 00:26:54,470 Multilingual: The OpenElectionCompass has been multilingual since the very 250 00:26:54,470 --> 00:27:00,744 beginning, and not just the interface, no. You can easily provide theses and answers, 251 00:27:00,744 --> 00:27:06,029 everything in multiple languages, even though this is not a big issue in Germany, 252 00:27:06,029 --> 00:27:10,900 I was thinking about countries like Switzerland where this can be essentially 253 00:27:10,900 --> 00:27:18,239 really – Kiosk mode: You can set up a terminal in a public place and put your 254 00:27:18,239 --> 00:27:23,039 election compass in kiosk mode, this mode will ask users nicely to reset the 255 00:27:23,039 --> 00:27:27,039 application once they are done or will do so automatically after a period of 256 00:27:27,039 --> 00:27:34,377 inactivity. Algorithms: The OpenElectionCompass has a flexible 257 00:27:34,377 --> 00:27:38,161 matching implementation that allows it to support different answer styles and 258 00:27:38,161 --> 00:27:46,659 algorithms. Because we don't know what might be the best fit for you. Privacy first 259 00:27:46,659 --> 00:27:52,649 statistics: The OpenElectionCompass now comes with an integrated tool to collect 260 00:27:52,649 --> 00:27:59,659 statistics in a privacy first design. Users can opt in to submit their answers 261 00:27:59,659 --> 00:28:05,020 anonymously for research. They can also help to improve the quality of the data 262 00:28:05,020 --> 00:28:10,970 set by answering more questions regarding their age, gender, education and more. 263 00:28:10,970 --> 00:28:14,958 I know that this is a difficult topic, so I am taking extra care to get this 264 00:28:14,958 --> 00:28:20,671 right. We certainly don't want to become a privacy nightmare. We want to help people 265 00:28:20,671 --> 00:28:31,213 support science in the most privacy caring way possible. By now, you probably want 266 00:28:31,213 --> 00:28:39,080 to get started building your own election compass. Next up, DIY. There are a number 267 00:28:39,080 --> 00:28:45,120 of principles when creating a VAA, written down in the Lausanne declaration. If you 268 00:28:45,120 --> 00:28:49,196 want to run your own election compass, I encourage you to read it. It's not even 269 00:28:49,196 --> 00:28:56,452 long. Let's go over the most important points quickly. In order to contribute 270 00:28:56,452 --> 00:29:01,740 sustainably to the good functioning of democracy, VAAs should be open, 271 00:29:01,740 --> 00:29:08,862 transparent, impartial and methodologically sound. This is important 272 00:29:08,862 --> 00:29:14,220 because if you're not transparent, there's a good chance that some people or even 273 00:29:14,220 --> 00:29:20,299 parties try to deny your legitimacy or impartiality. You should really follow the 274 00:29:20,299 --> 00:29:27,350 approach: We have nothing to fear because we have nothing to hide. A VAA should be 275 00:29:27,350 --> 00:29:35,380 freely accessible to all citizens. This is fairly obvious, but anyway, make sure that 276 00:29:35,380 --> 00:29:41,330 your VAA does not require any form of payment. This could be the paywall of a 277 00:29:41,330 --> 00:29:46,399 media outlet you've partnered with for a promotion. But this could also be less 278 00:29:46,399 --> 00:29:51,610 obvious, a mandatory collection of statistics. And lastly, keep in mind that 279 00:29:51,610 --> 00:29:57,289 there are probably more people with disabilities that you might be aware of. 280 00:29:57,289 --> 00:30:01,619 The OpenElectionCompass helps you with that as it provides decent screen reader 281 00:30:01,619 --> 00:30:05,968 support and generally follows accessibility guidelines. But you should 282 00:30:05,968 --> 00:30:12,289 also apply these design principles for any other content you might create around your 283 00:30:12,289 --> 00:30:20,169 VAA. A VAA should aim at the inclusion of as many parties or candidates that are on 284 00:30:20,169 --> 00:30:25,759 the ballot as possible. The criteria for the exclusion of parties and candidates 285 00:30:25,759 --> 00:30:31,929 should be publicly available and justified, and also parties and candidates 286 00:30:31,929 --> 00:30:38,899 should not be excluded from the tour for ideological reasons. I think this might be 287 00:30:38,899 --> 00:30:44,850 the most obvious rule, but also the most important. We want to help voters make 288 00:30:44,850 --> 00:30:50,459 informed choices. So we need as many parties to participate as possible. This 289 00:30:50,459 --> 00:30:56,369 might at times be a little difficult when parties don't want to partake. But more on 290 00:30:56,369 --> 00:31:03,549 that later. VAAs should be designed in a simple and intuitively understandable 291 00:31:03,549 --> 00:31:11,115 manner. OpenElectionCompass, this is the reason why the design of the 292 00:31:11,115 --> 00:31:17,327 OpenElectionCompass is not fancy at all, a bit boring even. Because I genuinely 293 00:31:17,327 --> 00:31:22,949 believe that it's how it's supposed to be. Many designers observe that web interfaces 294 00:31:22,949 --> 00:31:28,892 are starting to look alike too much. And there's some truth to that. But this is 295 00:31:28,892 --> 00:31:33,620 mainly because we have established a number of patterns that just work and are 296 00:31:33,620 --> 00:31:40,460 well known to users. So with OpenElectionCompass, I'm in fact building 297 00:31:40,460 --> 00:31:45,228 upon that. If you want an interface that is usable by as many people as possible. 298 00:31:45,228 --> 00:31:53,606 Boring is better than fancy. This is not art. This is design. The Lausanne declaration 299 00:31:53,606 --> 00:31:58,394 holds ourselves to a high standard, but as the original authors stated, it is 300 00:31:58,394 --> 00:32:03,042 meant as a starting point for discussion. There are a few points I would like 301 00:32:03,042 --> 00:32:11,630 to add. VAAs should collect user data only on an unobtrusive opt in basis, you might 302 00:32:11,630 --> 00:32:18,410 want to collect user data such as visitor statistics, answers and polls. There are 303 00:32:18,410 --> 00:32:23,816 good reasons to do so, but it's only ever happened with a clear opt in solution, 304 00:32:23,816 --> 00:32:29,269 preferably near the end of the election compass. A project like this should not 305 00:32:29,269 --> 00:32:39,006 appear greedy. VAAs should collect user data for science, not for profit. The 306 00:32:39,006 --> 00:32:45,691 collected data should be made publicly available. If you collect statistics in 307 00:32:45,691 --> 00:32:52,561 your VAA, do it for science. Let political scientists handle the methodology and 308 00:32:52,561 --> 00:32:58,090 interpretation, not some newspaper. And after the election is done, make the data 309 00:32:58,090 --> 00:33:08,320 you collected and if possible, your research available for free. VAAs should 310 00:33:08,320 --> 00:33:13,789 collect user data in a way that is not possible, that it is not possible to trace 311 00:33:13,789 --> 00:33:19,529 political opinions back to an individual. If you do collect statistics, make it 312 00:33:19,529 --> 00:33:25,161 impossible to connect answers to a name. Not only for everyone else, but for 313 00:33:25,161 --> 00:33:30,639 yourself. If you want to collect contact information for further research, save it 314 00:33:30,639 --> 00:33:43,340 separately from the user's answers. Users trust the VAA, so be trustworthy. OK, so 315 00:33:43,340 --> 00:33:52,669 where to begin? I prepared instructions how to run your own election compass 316 00:33:52,669 --> 00:34:02,759 consisting of 10 phases. Phase number one: Preparing. Organization, planning and 317 00:34:02,759 --> 00:34:08,810 communication are paramount. Before you do anything else, make sure you're all on the 318 00:34:08,810 --> 00:34:14,460 same page. Do you really want to run an election compass? Who is going to manage 319 00:34:14,460 --> 00:34:19,840 everything? This person doesn't have to, and in fact shouldn't do everything alone. 320 00:34:19,840 --> 00:34:25,439 But it is very beneficial to have a single person feeling responsible that everyone 321 00:34:25,439 --> 00:34:32,310 else completes their assignments on time. Write down your own timeline, get a tool 322 00:34:32,310 --> 00:34:41,582 to organize your team like Kanban board or a To-do-app. Phase number two: your team. 323 00:34:41,582 --> 00:34:47,240 You should never run a VAA all on your own, not just because it's an awful lot of 324 00:34:47,240 --> 00:34:52,620 work and responsibility and requires an extensive skill set, but because it is 325 00:34:52,620 --> 00:34:57,540 nearly impossible to do it in a legitimate way. You want to support the democratic 326 00:34:57,540 --> 00:35:03,742 process, so get a team of experts, advisers and supporters working together. 327 00:35:03,742 --> 00:35:09,760 Start with a list of people. This might include political scientists for advice 328 00:35:09,760 --> 00:35:15,300 and possibly in charge of the theses, the marketing specialists managing your 329 00:35:15,300 --> 00:35:22,310 marketing channels, social media, email, etc. A web developer with technical skills 330 00:35:22,310 --> 00:35:28,977 to get the election Compass online. A media designer, enthusiastic citizens, 331 00:35:28,977 --> 00:35:33,366 people with good connections to the administration, newspapers and other 332 00:35:33,366 --> 00:35:42,667 institutions. Someone with great language skills for wording and spelling. Think of 333 00:35:42,667 --> 00:35:48,090 people that might fit into these positions and contact them. Organize the kickoff 334 00:35:48,090 --> 00:35:53,127 meeting for your entire team to present your project. The plan, the structure, the 335 00:35:53,127 --> 00:35:59,230 timeline. Establish your organization tools and communication channel, get 336 00:35:59,230 --> 00:36:07,059 everyone to work, gather to-dos and assign them and set deadlines. Phase number 337 00:36:07,059 --> 00:36:15,155 three: The parties. It is important to get the parties on board. Normally, one party 338 00:36:15,155 --> 00:36:21,061 alone has no choice but to participate. You wouldn't want to be the only party 339 00:36:21,061 --> 00:36:26,270 missing. But if multiple parties aren't interested, you have a serious problem. 340 00:36:26,270 --> 00:36:32,273 You should not run an election compass with some parties missing. One or two 341 00:36:32,273 --> 00:36:38,130 small parties might be tolerable. You can simply ask for a gathering and give them a 342 00:36:38,130 --> 00:36:42,480 rough idea of what you are planning. At this point, it can be very helpful to 343 00:36:42,480 --> 00:36:48,994 belong to a reputable institution whose invitation cannot easily be refused. Most 344 00:36:48,994 --> 00:36:55,400 of the time, parties should welcome your idea, but be prepared for some persuading 345 00:36:55,400 --> 00:37:03,873 anyways. Phase number four: Preparing the workshop. The theses for your election 346 00:37:03,873 --> 00:37:09,220 compass obviously cannot be written all by yourself. They need to represent the 347 00:37:09,220 --> 00:37:15,828 society as a whole. The choice of theses decides over the quality of your election 348 00:37:15,828 --> 00:37:22,000 compass, you need to get this right. Your theses need to cover the most important 349 00:37:22,000 --> 00:37:28,839 matters for the next legislature. They need to be objective and impartial. The 350 00:37:28,839 --> 00:37:34,585 wording of the thesis has to be simple enough to be understood and to the point. 351 00:37:34,585 --> 00:37:40,042 Take this task seriously, it's the most important and the most difficult. To 352 00:37:40,042 --> 00:37:45,570 achieve theses of good quality you should run a workshop with a sample of your 353 00:37:45,570 --> 00:37:52,329 audience. Gather a group of young, probably first time voters, but if you 354 00:37:52,329 --> 00:37:58,180 like, you can also gather voters of all ages. Make sure the group is representable 355 00:37:58,180 --> 00:38:05,126 for your audience. No gender, race or religion should be excluded obviously. Set 356 00:38:05,126 --> 00:38:10,990 a date and find a large enough room with a projector, send out invitations and gather 357 00:38:10,990 --> 00:38:18,403 replies. Your groups should have about 20 to 30 members. Get the political programs 358 00:38:18,403 --> 00:38:24,040 of all participating parties with the help of your experts, gather topics of 359 00:38:24,040 --> 00:38:30,350 political interest from the programs and newspapers, and sort them into categories 360 00:38:30,350 --> 00:38:37,321 like social environment, work, traffic, infrastructure, energy, economy, finance, 361 00:38:37,321 --> 00:38:45,340 tax, security. You get the gist. This is your workshop material. Now plan the 362 00:38:45,340 --> 00:38:51,914 workshop. Help your group of voters discover the topics and create the theses. 363 00:38:51,914 --> 00:38:57,432 What methods are you going to use? Teachers can be very helpful here. What 364 00:38:57,432 --> 00:39:05,273 materials will you need? Whiteboards, pens, paper, etc. Phase number five: The 365 00:39:05,273 --> 00:39:12,879 theses. Use the topics and information you gathered to conduct your workshop. With 366 00:39:12,879 --> 00:39:18,260 your team and your group of voters. In this workshop, you will create a number of 367 00:39:18,260 --> 00:39:26,080 theses. Most election compasses gather around 50 to even 100 theses for whole 368 00:39:26,080 --> 00:39:31,879 countries at this stage. It will take you a few hours at least. Take care of your 369 00:39:31,879 --> 00:39:36,883 guests with, you know, pauses, lunch, snacks and coffee. Collect all theses in a 370 00:39:36,883 --> 00:39:43,590 list, and don't forget to work on the wording. Now, regarding the theses, there 371 00:39:43,590 --> 00:39:48,714 are some simple rules and some more advanced rules. The simpler ones are 372 00:39:48,714 --> 00:39:58,670 these: Can the thesis be easily understood by everyone? Are there words that 373 00:39:58,670 --> 00:40:02,820 not everyone will know? The OpenElectionCompass can provide hints 374 00:40:02,820 --> 00:40:12,344 on those, for those. Might the wording be biased? Does the wording match your style? 375 00:40:12,344 --> 00:40:18,218 Is this a good length? You know, these kind of simple rules. Now, for 376 00:40:18,218 --> 00:40:22,430 the more advanced rules. It can be quite hard to follow these, but you 377 00:40:22,430 --> 00:40:30,285 should at least try or maybe get some help with these. Advanced rule number one: 378 00:40:30,285 --> 00:40:37,162 Theses should not be about ideological values, but actual political policies. 379 00:40:37,162 --> 00:40:43,431 The first statement is completely vague. Voters cannot get any political knowledge 380 00:40:43,431 --> 00:40:48,854 from this because ideologically they most likely already know where the parties are 381 00:40:48,854 --> 00:40:55,342 standing. What's even worse, voters can interpret this thesis very differently. 382 00:40:55,342 --> 00:41:06,740 So be concrete. Number two: Theses should not be double barreled. It is very easy to 383 00:41:06,740 --> 00:41:13,900 accidentally merge two theses. And that makes them hard to answer. Every thesis 384 00:41:13,900 --> 00:41:21,222 should be about one policy and not mix two or more policies. In this example. Voters 385 00:41:21,222 --> 00:41:27,580 might be OK with soft but not hard drugs. So how are they supposed to answer the 386 00:41:27,580 --> 00:41:39,730 first statement? Focus your theses. Number three: Theses should avoid 387 00:41:39,730 --> 00:41:48,140 quantifications. At first, this thesis looks fine, it's clear and short, but what 388 00:41:48,140 --> 00:41:55,469 if you don't think there should be more surveillance cameras? If I reject this 389 00:41:55,469 --> 00:42:00,506 statement, what does it mean? It could mean that I'm OK with the numbers of 390 00:42:00,506 --> 00:42:05,820 cameras or it could mean that I'm completely against them. It's not clear, 391 00:42:05,820 --> 00:42:11,750 and this makes it hard to match partisan voters. It's often difficult to avoid 392 00:42:11,750 --> 00:42:19,924 quantification, but sometimes it can help to get down to the real issue. And in case 393 00:42:19,924 --> 00:42:27,734 of my hometown, this was that some people don't feel safe in public places at night. 394 00:42:27,734 --> 00:42:38,966 Now it's more of a boolean question, so try to go for these. And number four: 395 00:42:38,966 --> 00:42:45,816 Theses should avoid qualifications as well. This is a bit like the third rule, 396 00:42:45,816 --> 00:42:53,360 only this time we don't merge related theses but add more depth to a thesis by 397 00:42:53,360 --> 00:43:00,060 adding an example. This was taken from the Wahl-o-mat of 2002 and while it was meant 398 00:43:00,060 --> 00:43:06,510 to just be an example, it makes it more difficult both for the voters as well as 399 00:43:06,510 --> 00:43:13,070 the matching algorithm. Voters might support gay marriages, but draw a line 400 00:43:13,070 --> 00:43:20,910 when it comes to adoption. So what do they choose? In this case, it might be helpful 401 00:43:20,910 --> 00:43:30,906 to be more specific or even split this into two separate theses. This brings us 402 00:43:30,906 --> 00:43:38,390 to phase number six: The positions. Now it's time to let the parties answer and 403 00:43:38,390 --> 00:43:44,389 positions themselves. First, decide on the algorithm you want to calculate the 404 00:43:44,389 --> 00:43:51,309 matches with. This will also determine how many possible answers there will be. Send 405 00:43:51,309 --> 00:43:57,080 the theses to every party. You'll want to use an online form or similar, as the task 406 00:43:57,080 --> 00:44:02,840 of collecting all answers can get very tedious. Make also sure to collect the 407 00:44:02,840 --> 00:44:09,594 logos in appropriate quality and give the parties two to three weeks to answer, 408 00:44:09,594 --> 00:44:15,182 depending on your timeline. In the meantime, prepare to publish the election 409 00:44:15,182 --> 00:44:22,242 compass. Contact media outlets and tell them about your story. Contact the 410 00:44:22,242 --> 00:44:28,490 administration and ask them if they're willing to put up a link on their website. 411 00:44:28,490 --> 00:44:35,370 Contact schools, teachers, youth organizations and sport clubs and ask them 412 00:44:35,370 --> 00:44:40,231 if they are willing to share some graphics and a link with their followers once 413 00:44:40,231 --> 00:44:48,025 you're done. Phase number seven: Evaluating the answers. You know, I have a 414 00:44:48,025 --> 00:44:53,857 lot of theses and even more answers. The next step is to select the most important 415 00:44:53,857 --> 00:45:00,228 theses. You can do this in another workshop or on your team. Go through every 416 00:45:00,228 --> 00:45:05,592 thesis and decide whether it should become part of the election compass. Ask 417 00:45:05,592 --> 00:45:12,550 yourselves, is this thesis controversial enough? Is it helpful in telling the 418 00:45:12,550 --> 00:45:20,400 parties apart? At this stage, around 25 to 40 theses remain. Too few, and the results 419 00:45:20,400 --> 00:45:26,824 lose accuracy. Too many, and it would take too long for the voters to process them. 420 00:45:26,824 --> 00:45:33,730 Phase number eight: Time for a test. By now you should have everything you need. 421 00:45:33,730 --> 00:45:39,290 Let's run a test, feed your theses, answers and logos to the configuration 422 00:45:39,290 --> 00:45:44,951 editor to create the configuration file. Try it out, give it to your team and the 423 00:45:44,951 --> 00:45:49,900 people that participated in the workshops. Gather their feedback, make small 424 00:45:49,900 --> 00:45:57,278 adjustments until everything is ready for the big day. Phase number nine: Going 425 00:45:57,278 --> 00:46:03,780 public. About 2 to 3 weeks before the election, you should publish your election 426 00:46:03,780 --> 00:46:08,940 compass. Tell your web developer in advance, and when the election compass is 427 00:46:08,940 --> 00:46:18,350 online, tell everyone. And lastly, Phase number 10: Observe. Everything is up and 428 00:46:18,350 --> 00:46:24,390 running? Good. The only thing left to do now is get your election compass into as 429 00:46:24,390 --> 00:46:30,090 many hands as possible. Be available for questions and feedback from the public, 430 00:46:30,090 --> 00:46:36,242 and then wait for the election. Don't forget to vote yourselves. And when the 431 00:46:36,242 --> 00:46:41,702 election is over, archive the election compass. You can delete it, of course, but 432 00:46:41,702 --> 00:46:47,220 if you can, just keep it online. It can still be a valuable resource of 433 00:46:47,220 --> 00:46:52,333 transparency. If you collected any data for research, make sure to share it with 434 00:46:52,333 --> 00:46:58,700 the word. And lastly, please give back to the OpenElectionCompass, give feedback, 435 00:46:58,700 --> 00:47:11,300 write about it or improve our funding. And here we are, nearly done. Undoubtedly, 436 00:47:11,300 --> 00:47:18,720 there are many issues with democracy. Its implementations are incredibly complex and 437 00:47:18,720 --> 00:47:25,112 nothing that comes out of it is ever perfect. It can be frustratingly slow, 438 00:47:25,112 --> 00:47:31,096 inefficient, intransparent and even counterproductive. But it's also the only 439 00:47:31,096 --> 00:47:37,430 form of government that the majority so far managed to agree upon. And it's also 440 00:47:37,430 --> 00:47:43,111 the only form of government that is evolving continuously. We are right to 441 00:47:43,111 --> 00:47:48,297 criticize the system when it appears to be moving in the wrong direction. But we 442 00:47:48,297 --> 00:47:54,590 should not be tempted to hack our democracy. Hacking the system would mean 443 00:47:54,590 --> 00:48:01,678 bending it to our will. We don't want that and we don't need to. We don't need to 444 00:48:01,678 --> 00:48:08,450 hack a system that has the inherent ability to change. We can, however, try 445 00:48:08,450 --> 00:48:14,510 and fix the flaws, and I believe voting advice applications are a way to start 446 00:48:14,510 --> 00:48:23,010 doing this. A way of patching democracy. Thank you. 447 00:48:27,070 --> 00:48:41,196 Herald: Thank you so much Till Sanders not only for your lecture, but also for 448 00:48:41,196 --> 00:48:47,216 the development of this very useful tool, obviously. We still have some questions 449 00:48:47,216 --> 00:48:52,664 that our community posed on our social media platforms. And I would start with 450 00:48:52,664 --> 00:48:58,478 the first one: Who would be moderating the content in such apps? Like would it be 451 00:48:58,478 --> 00:49:03,636 peer moderated? Would it be state election agency or would it be something like what 452 00:49:03,636 --> 00:49:11,130 kind of moderation would there be? Till: That's a very important question, 453 00:49:11,130 --> 00:49:19,946 actually. Um, so in Germany with the Wahl- o-mat that's made by by the BPB, a 454 00:49:19,946 --> 00:49:27,508 government agency. And they develop it in a workshop like the one that I describe in 455 00:49:27,508 --> 00:49:35,970 the talk together with young first time voters. Because that's their target group. 456 00:49:35,970 --> 00:49:41,430 But apart from that, they obviously have, you know, political scientists, people 457 00:49:41,430 --> 00:49:49,810 who've been doing this for two decades now. And if you want to do it yourself, 458 00:49:49,810 --> 00:50:02,150 you won't have access to these kind of resources. So that can be a problem. And 459 00:50:02,150 --> 00:50:13,463 so far it worked well. We did this in in Münster, Cologne, Bielefeld, and Siegen. 460 00:50:13,463 --> 00:50:20,203 We had a team of political scientists who did this. So they had all the expertize. 461 00:50:20,203 --> 00:50:26,940 And yeah, there's no perfect answer for this. Um, not anyone, not everybody has 462 00:50:26,940 --> 00:50:36,160 these resources. Just try to do it as good as you can and maybe get some contacts who 463 00:50:36,160 --> 00:50:43,661 can help you. And we must not forget it's only for political education, not for 464 00:50:43,661 --> 00:50:49,580 actual voting advice. So it will never be perfect. But to a certain degree that's 465 00:50:49,580 --> 00:50:52,070 OK. Herald: But there is a question that is 466 00:50:52,070 --> 00:50:58,550 kind of subsequently to the previous one: This person is wondering, how do you or 467 00:50:58,550 --> 00:51:04,247 how do we make sure that the data is not corrupted? Like that it's not abused for 468 00:51:04,247 --> 00:51:11,290 political promotion, for example, or something like this. Like they refer to a 469 00:51:11,290 --> 00:51:17,870 big removing of Twitter of 20K fake accounts that did political propaganda in 470 00:51:17,870 --> 00:51:22,190 kinds of millions of tweets. And those were from a couple of countries, which was 471 00:51:22,190 --> 00:51:28,067 discovered in April 2020. So how could one prevent this in a way, or is it like the 472 00:51:28,067 --> 00:51:33,524 same that you already stated concerning the first question? 473 00:51:33,524 --> 00:51:43,548 Till: Yeah, well that's also a problem. It hasn't proven to be a problem so far. So 474 00:51:43,548 --> 00:51:48,703 as far as I know, there has been no case where this happened, but it could 475 00:51:48,703 --> 00:51:58,180 obviously happen. And since my tool and a few other tools are open source, there's 476 00:51:58,180 --> 00:52:07,228 no way we can stop this. But actually, that's the case for any projects out 477 00:52:07,228 --> 00:52:16,786 there, not only in this field, so many open source projects can be abused. Lets 478 00:52:16,786 --> 00:52:21,151 look at curl, I think the developer of curl is not allowed to enter the United 479 00:52:21,151 --> 00:52:30,611 States, because they think he is a hacker. The only thing we can really do is educate 480 00:52:30,611 --> 00:52:39,294 people about this topic in general and also improve education on VAAs themselves. 481 00:52:39,294 --> 00:52:51,380 So don't take them too serious and maybe take which ones are not honest. 482 00:52:51,380 --> 00:52:57,460 Herald: Yeah, this is actually a question that bothers or bugs the users a lot 483 00:52:57,460 --> 00:53:02,419 because another question is like: How could we ensure that there is no bias in 484 00:53:02,419 --> 00:53:08,290 the questions which actually connects to the previous questions as well. But they 485 00:53:08,290 --> 00:53:12,076 were wondering, for example, if the questions, the topic are not evenly 486 00:53:12,076 --> 00:53:18,883 distributed, which, yeah, actually tends to lead to some kind of bias in the 487 00:53:18,883 --> 00:53:23,155 questions. But like, this is the same problem, like with open source material, 488 00:53:23,155 --> 00:53:28,099 obviously, isn't it? Till: Yeah, yeah. That's a problem. So in 489 00:53:28,099 --> 00:53:32,430 Münster this year, there was another election compass based on the 490 00:53:32,430 --> 00:53:37,840 OpenElectionCompass, and it was from the BUND-Jugend. 491 00:53:37,840 --> 00:53:46,450 So they focused on environmental issues and it was a completely separate project 492 00:53:46,450 --> 00:53:50,490 and it wasn't even the first time they were doing this. In the past they did it 493 00:53:50,490 --> 00:53:59,380 with like flyers and stuff. And I guess that election compass probably was biased 494 00:53:59,380 --> 00:54:06,576 because it was part of their campaign, you could say. I mean, they're not a party but 495 00:54:06,576 --> 00:54:16,560 still. And I think what's most important is that it's transparent. Who is doing 496 00:54:16,560 --> 00:54:24,540 this election compass. So in Lüdenscheid it was a general youth organization not 497 00:54:24,540 --> 00:54:34,400 affiliated with any parties, funded by the government. And I think that's a different 498 00:54:34,400 --> 00:54:43,690 stand really to do that. On the other hand, the project in Münster wasn't that. 499 00:54:43,690 --> 00:54:48,438 It was biased, yes. But that doesn't make it a bad project. Just people have to be 500 00:54:48,438 --> 00:54:56,020 aware of that. And I hope we can tackle all these problems in the future with 501 00:54:56,020 --> 00:55:02,530 something that's more like a platform as a service. Maybe we can build an institution 502 00:55:02,530 --> 00:55:10,984 around this that can govern all these projects and moderate them a little bit. 503 00:55:10,984 --> 00:55:15,450 But maybe that's just a daydream. I don't know. (chuckles) 504 00:55:15,450 --> 00:55:20,340 Herald: OK, but another question that popped up that I find quite interesting as 505 00:55:20,340 --> 00:55:25,600 well is: Why VAAs can successfully approximate a given user's political stand 506 00:55:25,600 --> 00:55:30,108 in comparison to the party's answers to the thesis. I thought... like the user 507 00:55:30,108 --> 00:55:35,398 thought a lot about another issue. The missing knowledge on a theses' actually 508 00:55:35,398 --> 00:55:40,503 context, like they were wondering if you have an idea how to solve that with a 509 00:55:40,503 --> 00:55:46,850 similar technology driven tool or something like that, actually. 510 00:55:46,850 --> 00:55:52,460 Till: Yeah, that's a question I've had for a while. When I first did the Wahl-o-mat, 511 00:55:52,460 --> 00:55:58,520 that question popped up for me as well. Because there were some questions I didn't 512 00:55:58,520 --> 00:56:05,650 know anything about. And the idea is that you just skip those questions and they 513 00:56:05,650 --> 00:56:11,470 completely, they are not counted. That's how they deal with it. But I always 514 00:56:11,470 --> 00:56:18,804 thought, why don't they tell me more about it, so I can make an informed choice? So 515 00:56:18,804 --> 00:56:25,795 far, I haven't managed to find the definite answer to this, but now I believe 516 00:56:25,795 --> 00:56:31,640 they don't do it on purpose. So they do it on purpose, but they don't do it, you 517 00:56:31,640 --> 00:56:44,669 know? I think the risk would be too high to influence users, because when the 518 00:56:44,669 --> 00:56:48,870 Election Compass itself tells you everything you need to know about this 519 00:56:48,870 --> 00:56:56,960 topic, they decide what's true, what information they give you about this 520 00:56:56,960 --> 00:57:03,360 topic. And I think it's incredibly difficult to be really objective when 521 00:57:03,360 --> 00:57:10,451 creating such content. So I guess that's the reason why they don't do it. 522 00:57:10,451 --> 00:57:14,483 And I think it's a good reason, actually. So people should get information somewhere 523 00:57:14,483 --> 00:57:20,430 else, multiple sources. Herald: So you would not lobby for some 524 00:57:20,430 --> 00:57:24,680 kind of option that you could, you know, expand like you are a open source project, 525 00:57:24,680 --> 00:57:31,300 for example, to cover that one as well? Till: No, I don't think so. Actually, if 526 00:57:31,300 --> 00:57:38,108 you scroll down all the way, you can see the party's answers. And I did it on 527 00:57:38,108 --> 00:57:46,480 purpose that you can... it's more like a chat. So they can actually discuss that in 528 00:57:46,480 --> 00:57:52,820 a way and you can read their answers and then decide what, who you believe. 529 00:57:52,820 --> 00:57:59,320 Herald: Yeah. All right. And you are actually doing this by yourself, you 530 00:57:59,320 --> 00:58:06,920 founded this a year ago, some bit prior because you saw the problems that were 531 00:58:06,920 --> 00:58:12,914 there. But how could one join the party? Like how could one help out? How could 532 00:58:12,914 --> 00:58:20,770 somebody work with you on this project? Till: Yeah, well so I have lots of ideas 533 00:58:20,770 --> 00:58:27,340 how you could improve this project. And so far I'm managing fine to do it on my own 534 00:58:27,340 --> 00:58:35,411 in my free time. And I don't intend to do it full time. It's a good side project, 535 00:58:35,411 --> 00:58:45,855 but someday I think more people should get involved. And there are several ways to do 536 00:58:45,855 --> 00:58:55,410 so. So the project is based on Vue.js and everyone who is familiar with that can 537 00:58:55,410 --> 00:59:05,780 easily join. There are small and big todos that could be done, and many ideas will 538 00:59:05,780 --> 00:59:14,130 require a server side application. So that's something I've been working on in 539 00:59:14,130 --> 00:59:21,445 the past few weeks. So that's also something that could attract 540 00:59:21,445 --> 00:59:29,410 contributions. Herald: OK, and another question I have 541 00:59:29,410 --> 00:59:35,030 left would be: Will the slides be available somewhere like do you have your 542 00:59:35,030 --> 00:59:40,150 presentation somewhere online? So people who are interested, who love the design 543 00:59:40,150 --> 00:59:47,670 and content, like the user who posed this question could still grab it somehow? 544 00:59:47,670 --> 00:59:57,495 Till: Thanks, first. Yeah surely, the project already has a website and many 545 00:59:57,495 --> 01:00:04,033 things in the presentation are also on the website. But everything, I will add them 546 01:00:04,033 --> 01:00:13,070 in the next days, I think. And as far for the slides, I can, I will upload them 547 01:00:13,070 --> 01:00:21,780 somewhere unless I don't know you do that. Um, I don't know. You have like this hack 548 01:00:21,780 --> 01:00:29,560 media site where you, where you post some videos. Herald: So we are allowed to put 549 01:00:29,560 --> 01:00:35,830 them online for you. Till: Yeah sure. They will definitely be 550 01:00:35,830 --> 01:00:39,304 available somewhere. Herald: OK. Amazing. Is there something 551 01:00:39,304 --> 01:00:44,780 left for you to say to that. You just want to get off your heart? 552 01:00:44,780 --> 01:00:52,738 Till: Well, I really enjoyed doing this and I myself learned a lot about VAAs in 553 01:00:52,738 --> 01:01:01,390 the process, that was nice. And I'm just happy that so many people listened to me. 554 01:01:01,390 --> 01:01:06,165 Herald: Thank you very much for being here. For your presentation, for your 555 01:01:06,165 --> 01:01:13,282 work, of course. And for being here, for joining us. We will go ahead on this 556 01:01:13,282 --> 01:01:21,617 channel with air filters. It starts at 8 p.m., of course, and it will be some sort 557 01:01:21,617 --> 01:01:28,180 of an instruction how to build your own air filters that actually get your air 558 01:01:28,180 --> 01:01:35,130 clean and virus free by using your 3D printer. For now, we say thank you very 559 01:01:35,130 --> 01:01:39,814 much and say: See you next time. 560 01:01:39,814 --> 01:01:48,300 postroll music 561 01:01:48,300 --> 01:01:59,520 Subtitles created by c3subtitles.de in the year 2021. Join, and help us!