 The next speaker is called Code Freezer. The title is Analyzing the Coalition Agreement TOOTS and Methods to Sussell the Agreement of the Future Government Coalition. He is Enterprise Engineer at Enterprise by the Cankeru. He calls himself a fact-checker who likes to examine things. If you have questions, you can put them on Twitter, or Masterton, or IRC. So, here we go. Hello and welcome everyone in the following 30 minutes. I would like to tell the story of Karala Kitt and myself, how we used quantitative text analysis and explorative reading at this time applied to the Coalition Agreement of the New General Government, which we call CA in the following. My name is Code Freezer. Let's go. On the 24th of November 3, things happened. There was a birthday in my family. The cookbook has been published and the Coalition Agreement has also been published. Some people call it the Yellow Pages. On the 25th of November, this little comic strip with the Cankeru appeared in my tweet feed. I realized that I should really have a look at the Coalition Agreement. There were too many strange things that I heard about it up front. Then the call for papers for RC3 happened. So, let's say I'm going to put something in there and I put something in there shortly before the deadline and it was accepted. So, here we are. So, what do you need to do? When you do a talk, you need to somehow capture people's attention, sex cells, scandals, headlines and so on. We didn't have to look far. We looked at this. 177 pages are the total. 76 of these pages are thin air. So, please don't print this out. Here's the attention. We of course want to change the world. There are just this Coalition Agreement in front of us. So, what do we do now? So, let's take it apart for different perspectives. We can look at it as a source code, which we are familiar with. So, we can try to apply this to the Coalition Agreement. What if we look at it as a form of specification for the next legislation period? Or what if we look at it as an object model? Is there any point in this? So, let's see. And finally, we look at the CIS literature, because I would like you to actually read it, because this is what our government will base directions on for the next four years. So, Coalition Agreement as source code, the first chapter. So, I mean, of course, this is not Java or Golan or Ruby. This is German. German as a programming language. So, if we have a programming language, we need to think about atomic parts of a programming language. So, structural types, the letter is a character limited by two spaces. Sentences usually terminated by a full stop. And line feed terminates a paragraph usually. Apart from that, there are higher text design structures, which are usually structured according to the DKAT principle, which stands for document chapter section and topic. The nine chapters, 22 sections and in total 200 topics in the Coalition Agreement. Counting. So, that's another thing. If you want to count something to prepare it afterwards, page numbers and line numbers are not really that useful, because they're not independent of the device. They depend very much on the formatting. So, lines of code is just not being used anymore today in code metrics, because they can be very different, different style guides. We don't really count letter characters because they don't have very any content. Words can be grouped into nouns, verbs and adjectives without stop words. And this is not such a bad idea as a base for later on. Now, the number of sentences is the most meaningful and reasonable base for me. So, this is comparable to NCSS, which is an interesting slide here, referenced down below. So, what can we do with those metrics? So, for instance, let's look at the number of words, which are grouped into these different groups, nouns, verbs and adjectives. And you can see what are the most frequent nouns. Well, of course, Germany, Germany is high up there. Communion, cooperation, digitalization, country up a bit less. So, if I look for climate, for instance, it's quite small. Climate protection even smaller. Financing is not that large. In the case of verbs, it gets a bit more difficult. So, verbs themselves and the programming language of German, some verbs are simply more frequent, such as to want, must, etc. But if this has any relevance that other verbs are much smaller, this is hard to say without having a clear semantic association. So, adjectives, we can see words such as good, public, digital, quite large, which is great. But even here, I'm not sure if I should be worried that words like transparent or ecological are quite small, so not very frequent. I mean, this is just counting words. Doesn't necessarily mean very much. Here's the link where you can build such word clouds yourselves and the wordvolcan.com. What else can you do when you count in words? So, in addition to the raw numbers, you can look at the relations between words. Some are quite close together, such as Europe and EU, functions, national, strength, protection, which are grouped together here or here to lead to create access regulation, which is a small cluster here. So, based on these distances, you can also create graphs or you can show it in a so-called self-organizing map of words. We will look at this later on. So, here I counted the different sentences in the 22 sections. So, these are the 22 sections in the different chapters and we can see climate protection, socio-ecological market economy is the dominating topic, which groups together ecology and economy. That's also the reason why this is so large. But in my opinion, this is the right focus. And the digital revolution, this is where the three parties agreed on and there's one chapter where there's nothing. Find this. Doesn't have any content? Well, not quite. I mean, what happens here is that the different topics were not grouped into sections here. So, I thought it was quite funny because I think they did this on purpose since it would not have fitted on the pages if they had actually formulated this. So, let's look at the different topics, the 200 topics. This is probably a bit hard to read in the video. So, I prepared something here. Here's an SVG graphic. So, we can zoom in and out. So, let's get the topic of economy and ecology and market economy. This is sorted now by the size of topic blocks. So, they're becoming smaller and smaller from top to bottom. So, for instance, river traffic and car traffic is not so far apart. What if there's animal protection, there's bicycle traffic, two sentences, no pedestrian traffic. But as we see, there are financial topics. So, it's not all that bad. Interesting topics such as money laundering. You might remember FIU, Financial Intelligence Units, something cyber-cyber-cyber. Let's see what happens there. But you can get an overview here about the 200 subjects as they are split across the different chapters. And this is also linked in the slide so you can look at that directly. So, so much for what can I see if I just count around and display what I've counted. Let's look at the coalition agreement as a specification. So, here's a small thought experiment. Let's assume the people are the customer and the state is the enterprise and the goals are called objectives and a legislative period is a PI, a program increment. Does that sound familiar to you? That sounds kind of like safe. And this is indeed following the safe methods or inspired by them. And if you find safe interesting, you can look into that. And it is linked there. It's very interesting variants to bring together various development teams combining many existing things such as Scrum, DevOps, release engineering, reason demand, something. And it's customer-driven, community-driven even. So, you yourself can get engaged and work on this framework as well. It's also a version. So, it's on version 5.1 or maybe a 5.2, I think 5.1 at the moment. And the interesting thing is the colleagues who are working on this and continue working on this, they work according to SAFE, which is also interesting. But we want to look into the coalition agreement. So, there it's about goals or objectives mainly. So, the objectives are the most interesting thing. And SAFE has an interesting offer here. According to SAFE, these objectives should be smart, which means specific and measurable. So, they have to be measurable, achievable. You have to be able to achieve them. My first idea here is spontaneously the thing with the election age. Starting to vote from 16, I am very much in favor of this. And I think they can do this for Europe. But for federal elections, they need a two-thirds majority. So, this might not be really achievable, even though I like it as a goal. Reasonable. So, objectives have to be reasonable and time-bounded. So, they have to be able to achieve within PI. And I did one thing here, which is to look at the year numbers from the coalition agreement and see where are they. So, I took some number with 6 before the legislative period, 80 within the legislative period. And I know, okay, this is achievable within this period of time where this coalition is actually in the government. But there are also goals that are outside of the legislative period. And I think this is not so ideal, because those might not be really achievable. And there is a fair point. You can have larger goals. So, you have to deal with longer periods of time. But I would expect that there is something described for these longer goals, something that you can do within this legislative period to reach these goals. So, I wanted to look into this a bit further, a bit closer, into these 40 goals. So, I built this timeline with all the mentions of years outside of the legislative period. And if you are on the slide deck and you can't read this, then there will be a link to an HTML version of this outside of this slide deck. And it's interesting, there's a positive example here. In 2026, they want to achieve something which is outside of the legislative period, but until then, they want to do something which is within the legislative period. So, it can look like this. So, that's how we can do it and work with it. And if we look at, well, I think this word has gone through the press enough already. I don't need to say anything about it. But you can see there are a lot of things in here. Well, I don't know. It's not really intended that it's achieved within this period of time. And so these objectives aren't really smart. So much for the coalition agreement as a specification. Now, let's look at it as an object model. What object model? So, color could also looked at me strangely about this. Oh, that's architecture. It's weird. Can't you remember all of these discussions we had in the project already agile? Or is it waterfall? Is it understand if our agile manifesto the big upfront or something evolutionary, a zero models versus no plan tall, doctors code, designers code, diagrams is code, architectures code, something as a service, we know all of this. And within this field of tensions, Simon Brown thought of something interesting and call it C4. And C4 is not related to the Cologne chaos club. It's also not the explosive C4 means according to the time in Brown as abstraction level, which are for abstraction levels, which is our kind of missing link in the UML in UML, we have all kinds of artifacts. But we don't have any idea or, well, in system L maybe a bit, but we don't really have an idea otherwise of how we can structure larger object models, according to looking at the level of detail. And so we thought, let's look at these four level of details, context, container, that doesn't mean Docker containers, but that we could talk about chapters and sections here and components, those are the basic packages. And then there's the code. So the text itself, and those four abstraction layers, Simon, things in these levels. And if you want to read about it, I've linked this, of course, again, also about the debate of agile versus architecture, there's a nice slide deck, very amusing to read. And all right. So what do we do with this now, context, for instance, so we could start like this, a very simple diagram. So that's a completely minimalist C4 model on the system context level. If we have the government, we have the coalition agreement, we have a relation between them, which is the government follows, of course, unconditionally, the coalition agreement. And we could split this up further. The government consists of ministries, of course, the coalition agreement consists of chapters, and then there's some relations between them. And I blow this up a little bit. So some information on the side here, PlantUML is often used for diagrams as code to create architectural diagrams like this. But there is for PlantUML C4 plugin. So it's linked here on GitHub, you can download it from there. And then there's this address that you can't quite read, but you can use this and then you can very simply create C4 models with PlantUML. As I said, here we have some ministries, Carl is in the Ministry of Health, we have the Ministry of Finances, or the economy, the traffic ministry, they all want to implement the objectives of the coalition agreement, and they form the cabinet between them. And within this context, Simon Brown talks about the comparison to Google Maps. So if you have continents and then countries and cities, and within the cities, you see the details. And I'd already talked about this self organized word map, and which you can create with Carl Kroeder out of her text. And this might be an idea for how you can create these word maps in this direction. Alright, so now we have the context here, but we also want to manage this object model, of course, and there are several applications and tools for that, where you can manage object models. There's StarUML, there's Federal Paradigm, or something architect, which I quite like personally. And there I can manage my structures in there. But Enterprise Architect, UML, XMI is also smart. Angle brackets, XML is not really so modern. I don't really know why. But today, you like to do it with Jason or Yammel. So I did this as well. So I looked at the agreement as Jason or Yammel. And during all of this preparation and arrangement, I wrote some Python again for some in a while, which was also fun. And I finally also created a useful table of contents for the coalition agreement where you can actually read and see the connections. And there I had the idea. If I want to put anything into an object model, then it might be useful to have all of these parts number by bibliographically. So then I know I am in T2111 in modernizing the public structure, something. And here I'm in subject to section one or whatever, section two chapter two, subject seven. And I always have a context there where I currently am. And then it doesn't matter if I'm printed on a four or read on ebook reader or something else. Alright, and so what do I have now after taking this apart analytically and putting it into an object model? So what can I do with these ideas? Well, I can, for instance, put these D cut elements into and I can add annotations and we talked about smart already in the specifications. So we could introduce a ranking there. You could have acceptance criteria. We could also just associated with the ministers going through an icon or with a party at the colors of each party to annotate it or at the time to read each element. And we could link these D cut elements and create relations between them or literature and code metrics. So that's some ideas that you could do. But also you could put these entities that you put apart place them in new context with these generators. Or maybe you have another idea. So this is where your idea comes in. What could you do if you have text like this in an object model? What could you do with it? I was talking just about smart rankings. And with objectives, if you remember, those are the objectives. And in PI, you have per objective, these BVs, the business values, which means the stakeholder says, what is what business value does this contribute if I implement this objective? And so this annotate before you actually do it. And then after an increment, you the person who annotated this comes back and says, what is the actual value now the AV? So what was the actual value of this that it contributed? So this might be an interesting method to analyze how the objectives were met during a legislative period, or to generate something spaceships of different kinds, for instance, museums, maybe. And here I've found something interesting, which you can look at or listen to directly after this talk, which is the everything exhibition, which takes different structures. I think currently is wiki structures, Wikipedia structures, if I'm not mistaken, and then it generates 3D exhibition spaces. That's very exciting. I'll take a look at that later. Yeah, you see, for 22, for the next year, I have a new hobby. So much on the subject of KV as an object model. So now on to the last chapter, coalition agreement as a form of literature. This of course, specialized text. But it can also be simply read. But why I had problems with that at the start. I didn't quite know why, but when I was looking at the PDF, I was really hard for me to just start reading, maybe they want to keep the secret or don't really want people to read it. So then I looked at the document itself and noticed a few things. For instance, the distance between lines 1.5 really blows up the text. It makes it hard to read. The page margins are quite large, the differences between paragraphs also. But the block alignment on a force really hard to read those long lines on an A4 page is really difficult. Another thing is the numbering or the lack of numbering of the different topics. It's really hard when you just jump into the text and you hit a topic and you don't really know where exactly you are. I mean, is this about international things but economy? So I would suggest you also number the third level, the level of topics. But this is really blown up. I mean, these are the 177 pages of the agreement. So if you simply set page margins normal, paragraph distance normal, single spacing, left alignment instead of block alignment. The result is 101 pages instead of 177. I mean, it doesn't really matter as long as you don't print it out. But imagine, if you print agreement 1000 times, then 76,000 pages would be superfluous. I mean, there's probably enough interested people who will do this. Maybe 1000 is time, but I don't know. So 76,000 pages would be completely selfless. 43% paper, water, COT would have been wasted, which I think is a bit of a pity. So why not in a more advanced manner, more progressive digital climate and energy saving and device independent and saving money using one of those formats here. So I don't want to complain. I just also want to do so. I built different versions in HTML, the way I would like to read it, which is really nice for me personally. I also created an EPUB version, which can read on your EPUB creator. Also, I converted everything to marked down for using kitlab or even atomized the whole agreement with all the different topics and sections, which you can download in the form of a zip file extracted on your console and have a look at it. And you can easily navigate it. So it's interesting to read the collision agreement on a shell. I put something together for you here. So with a work tool called wordliga, I created a playlist here, which says how long it would take approximately to read each section. I mean, you can just jump in there doesn't have to be done from start to finish. And here it says how long this will take approximately to read to the whole thing. So interesting is the idea of the Schneider who reads the entire collision agreement, you can have a look at that on YouTube. So thanks a lot. This was it. One last thing. No session without it. So I mean, I was complaining a bit. But if you really look at the collision agreement, true pink glasses, there are quite a few things which are really good. So paragraph 219 a will be abolished election age of 16. I think it's great legalization of cannabis at last. So DB Nets, the railway company will be a common good oriented nuclear energy has to bear their cost themselves. chances for illegal residents or foreign nationals after four to six years. So this is great, which allows people to enter society who are normally marginalized and illegal right to open data. This is fantastic. So fact the start will be delighted. And I'm really curious if, for instance, the German standards will be open, right for anonymity. It's great. Open source and multi cloud for public procurement. So lots of great stuff. If you want to read more, but this especially cyber topics, have a look at let's politic.org here's a link to however, all the others, these are words that are not anywhere to be found in the coalition agreement. So I've compared the large ones with Google findings. Interesting. Most interesting was the larger one is bicycle. Okay. But it is there's nothing talking about fracking. There's nothing about a limit, a speed limit on highways. It is in there. But it says that it will not exist the temple. So it's not on this map. And I think we have to be careful because a lot of governments previously, they weren't really my government. So I wasn't in the majority. But now you might think like myself, I voted for one of these three parties. So now I feel responsible for this. So I want to check that we want to take this coalition agreement as a promise. It's not yet law. You still have to change a lot of things around it. But we have to be careful that it turns out well. And now I'm finished with all the thematic stuff. Just one more last thing, saying thank you to BleedTrack, for instance, for the quality here, which I loved a lot. Maybe you recognize the drawing style already. I can't even say it anymore for the stage here, which is very, very cyber. Very well done. I was very happy that you accepted my proposal. And there's a lot of my family has been very patient and has some feedback and has not had as much of me this Christmas as I should. But then you can reach the positive rails here. I think you already we have a small Trello where we plan what we might do there. We have the slides, of course, I will try to with some luck. Maybe I've even done it during the presentation to upload the slides so you might be able to find them in the pre talks on the pre talks site. And you can download them there directly. And you can reach me in rocket chat as code freezer, you can send me an email, you can find me in the fattyverse. I'm often also on Slack, because the colleagues are there. And that's really it. Thanks a lot for your attention. And until next time, see you. All right, we also say thank you very, very much for the talk which was very interesting. And I think a lot of people in the audience will agree. And I would not have thought that there is so much being saved. And there's also a lot of questions. So I'll just start with the first question. What was the base of the data for the shown points of interest to reach the talk? Something so the data was the coalition agreement text. Alright, next question. Could you also look into the party programs of the parties and see what you need to look into? Yeah, that's interesting. So the comparative analysis of the election programs to the coalition agreement. I would like to approach this in the next couple of weeks or months, because I like to know who influenced what in which way the election programs themselves. Sure, can have a look at that. And who wants to do that? Just contact me and I'll provide you with them. All right, that's very kind. Then next question already. Is the coalition agreement really meant to be read like this? Isn't it just a window talk? Or isn't more the way to the agreement, the goal that they have the parties form working groups and then are forced to build teams similar to what is what happens with mission statements and companies? That's a very long question. I think we can summarize it as isn't there more talk and it's meant more for team building and self image? Sure, I mean, I totally agree with that, especially chapter one, which was kind of a self exploration journey. And they also those parties have to somehow get together because they used to be quite distinct. And it's, I think it's totally appropriate that they do this in the form of a coalition agreement. They also didn't just write this down so that they can then just have quite time. I mean, we're going to look that they actually can implement this and then just talk around. And one more question. Can you work on working with kids in the next five years? Sure, of course. I also want to include the election programs get and also the coalition agreements of previous years. So I'm going to be updating the repository regularly. Then I have one more question. Did you enjoy working on this? Or is it just tedious after some time? I created also a plan for next year because I was really excited about this. And if anyone else wants to join, just get a touch on me. All right, very, very nice. And if you want to continue working this, get in touch with code reader and thank you again very much for this very cool talk. Yeah, thank you very much also from my side. If you want to leave me a comment, you can go to the pre talks page. There's place for this. You can add some. You can also visit me on my GitHub repository, make an issue, or even a metric request. Thank you.