# PolitiX
{{< icon-button "https://politix.mesu.re" "link" >}} PolitiX {{< /icon-button >}}
{{< icon-button "https://github.com/PierreMesure/politix-sweden" "github" >}} Source code {{< /icon-button >}}
## X/Twitter has become a danger to our democracy
Researcher Carl Heath has written extensively[^fn:1] [^fn:2] [^fn:3] [^fn:4] about the problems that arise when X/Twitter is still used by our politicians and media, even though the platform has been hijacked by tech billionaire Elon Musk. Musk actively uses it to undermine our democracy, influence elections, and train his AI service Grok, which turns AI-revenge porn and deep fakes into a commodity.
## Yet most politicians remain there
Despite these developments, a surprisingly large part of the Swedish political establishment remains on X (ex-Twitter). It continues to be used as a primary channel for announcements and debate.
But what do the numbers actually show? Have Members of Parliament begun to seek out decentralized alternatives like Bluesky or Mastodon, or are they stuck in Musk's ecosystem? I wanted a clear overview so I decided to build **PolitiX** to find the answer.
The tool reveals that many politicians remain on X, and a significant portion is still active. Only 10% have joined Bluesky and 2.5% Mastodon, but very few remain active on these platforms. Most on the latter two platforms belong to left-leaning parties, but far from all. Overall, over 72% of the governing "Tidö" coalition remain on X, while less than 63% of the opposition have an account.
## How does PolitiX work?
PolitiX consists of several parts:
- A script that fetches data from Wikidata about who the current MPs are and what social media accounts they have. This allowed me to build on a dataset that was largely complete, although I needed to add about 50 accounts and review all data. The best part is that my contributions can be used by many others and everyone can help keep the data updated in the future.
- A "scraper" that visits each profile page on X to check if the profile still exists and when the user last tweeted. Since Musk shut down Twitter's API for researchers and journalists, it is unfortunately difficult to retrieve such information without paying a high price. Difficult, but apparently not impossible, as I succeeded after a few hours.
- An interface for browsing the data, filtering it by several interesting angles I chose, and visualizing it graphically in a way that resembles a parliament (using a fantastic JS package: [parliament-svg](https://github.com/juliuste/parliament-svg)).
[^fn:1]: [The threat of X and why Sweden cannot wait until the 2026 election (in Swedish, carlheath.se)](https://carlheath.se/hotet-x-och-varfor-sverige-inte-kan-vanta-till-valet-2026/)
[^fn:2]: [How Elon Musk can influence the Swedish election (in Swedish, carlheath.se)](https://carlheath.se/sa-kan-elon-musk-paverka-det-svenska-valet/)
[^fn:3]: [X – the platform that is itself the perpetrator (in Swedish, carlheath.se)](https://carlheath.se/x-plattformen-som-sjalv-ar-forovare/)
[^fn:4]: [Is our democratic conversations doomed to be shared on digital rollercoasters? (in Swedish, carlheath.se)](https://carlheath.se/maste-vara-demokratiska-samtal-vara-domda-att-delas-pa-digitala-tivolin/)
# Whisper-Web
{{< icon-button "whisper-web.mesu.re" "link" >}} whisper-web.mesu.re {{< /icon-button >}}
{{< icon-button "github.com/PierreMesure/whisper-web" "github" >}} Source code {{< /icon-button >}}
On Valentine's day 2025, the Swedish National library's AI lab (KB-labb) released their own fine-tuned version of OpenAI's Whisper models. They retrained the models using thousands of hours of audio from both SVT (the Swedish public service) and parliament debates. Unfortunately, to avoid legal issues and to stay focused on their mission, the lab only provides the [open weights of the models](https://huggingface.co/KBLab/kb-whisper-large).
So I took a few hours to build a web interface to use the models as simply as possible. I called it Whisper-Web.
The first time you give it a file, it will download an open AI model and perform the transcription locally in your browser. This means that your audio file never leaves your device. It also means that the transcription will be slow or fail if your computer/smartphone is not powerful enough to perform it.
In the settings, you can pick among different models and various quantisation levels. A smaller model with a lower quantisation will be faster but make more mistakes. By default, Whisper-web uses small models but you can try a bigger one and see if it works on your device.
In the app, the user can choose between the Swedish models, Norwegian ones (from the [Norwegian national library](https://huggingface.co/collections/NbAiLab/nb-whisper)) and OpenAI's models that are the best in all other languages.
This project's code is actually a fork of a demo created by Xenova that I updated and improved. Both are available on Github. Feel free to reuse or contribute to it.
# Keeping track of Swedish government agencies
**THIS ARTICLE WAS ORIGINALLY RELEASED ON [MEDIUM](https://medium.com/civictechsweden/vem-har-koll-på-sveriges-myndigheter-dc8ca8e9dab7) (IN SWEDISH).**
A few weeks ago, the Swedish Agency for Public Management (Statskontoret) made a surprising announcement: they had found 25 new government agencies that no one seemed to know about.
This anecdote echoes a problem I have wrestled with for several years in various projects. If you want to analyse what agencies say, what they do, or for example what they spend, you must first know who they are and what they are called. But every year new agencies are created and others are shut down. Some change names, others are merged.
An example I am working on right now in my role at the Swedish National Financial Management Authority's (ESV) datalab is a data analysis of agencies' consultation responses in a project funded by ESO. We fetch responses as PDFs from [regeringen.se](http://regeringen.se/), but this dataset also contains responses from municipalities, civil society, industry organisations, and even agencies that have changed names or ceased to exist. So how do we find the agencies' responses in these 40,000+ documents?
My naive guess when I tried many years ago was that it would be as easy as downloading the agency register (*myndighetsregister*) from Statistics Sweden (SCB) and matching against these names. But it's not that simple. Here are some examples of problems you encounter:
* the name "Jordbruksverket" gets no hits in the register (the agency is officially called Statens jordbruksverk)
* The Data Inspection Board (Datainspektionen) recently became the Authority for Privacy Protection (Integritetsskyddsmyndigheten), but SCB's register only contains current names
* Many organisations' names contain typos, both in our dataset and in the agency registers.
In addition to this, SCB's agency register is neither complete nor updated. So when I read the article, I wasn't surprised, but it made me want to learn more about the problem and develop a solution that would make the situation better.
## Where do you find information about agencies?
We start with a mapping of the so-called agency registers and other lists that exist. Here are the ones I found:
* SCB's [agency register](https://myndighetsregistret.scb.se/)
* Statskontoret's [Facts about the public administration](https://www.statskontoret.se/fokusomraden/fakta-om-statsforvaltningen/myndigheterna-under-regeringen/)
* ESV's [register](https://www.esv.se/rapportering/myndighetsregistret/) of agencies included in the state's accounting system and the [list](https://www.esv.se/statsliggaren/) of agencies that have received an appropriation direction
* The Swedish Agency for Government Employers' [member list](https://www.arbetsgivarverket.se/statistik-och-analys/staten-i-siffror-anstallda-i-staten/staten-i-siffror-om-arbetsgivarverkets-medlemmar/) (*Arbetsgivarverket*)
* [The Legal Database](https://beta.rkrattsbaser.gov.se/) which contains the agencies' instructions
* Wikipedia and its little sibling [Wikidata](https://www.wikidata.org/)
* [Handlingar.se](http://handlingar.se/), an important service that lists organisations covered by the principle of public access to official records, including government agencies
All of these have their own goal for mapping the state administration and their own perspective on the problem. But what happens if you fetch data from all these sources and merge them? The potential benefit is great:
* Analyse agencies in a smarter way based on the data mapped by different organisations
* Integrate this data into more systems thanks to a standardised machine-readable format
* Identify agencies based on their various names and abbreviations
* Detect errors in the original sources to ensure quality increases and the different registers become more consistent
## Myndighetsdata, everything in one place
That is why I started the project [Myndighetsdata](https://github.com/civictechsweden/myndighetsdata): key information about agencies in one place. The dataset contains information from all the lists above, including extinct agencies, old and alternative names, and attempts to merge the information in an automated way.
Here is an example with data for the Swedish National Financial Management Authority merged:
```json
{
"name": "Ekonomistyrningsverket",
"name_en": "National Financial Management Authority",
"short_name": "ESV",
"department": "Finansdepartementet",
"org_nr": "202100-5026",
"website": "www.esv.se",
"phone": "086904300",
"email": "registrator@esv.se",
"cofog": 112,
"cofog10": 1,
"structure": "Enrådighet",
"has_gd": true,
"postal_address": "BOX 45316 104 30 STOCKHOLM",
"office_address": "DROTTNINGGATAN 89 113 60 STOCKHOLM",
"other_names": ["Potential other names..."],
"sfs": [ "1998:417", ..., "2016:1023"],
"wikidata": "Q7654780",
"wikipedia": "https://sv.wikipedia.org/wiki/Ekonomistyrningsverket",
"start": "1998-01-01",
"employees": { "2024": 159, "2023": 136... },
"women": { "2023": 95, "2022": 94... },
"men": { "2023": 64, "2022": 63... },
}
```
You can access the project here on [Github](https://github.com/civictechsweden/myndighetsdata). The data contains machine-readable versions of all sources with standardised names for most fields, enabling comparisons and analysis. There is also a merged list that matches the same agencies across all registers where they appear, but it cannot be considered correct until it has been manually reviewed. As I explain further down, the data is too dirty to be fully classified automatically. The project's source code is of course open and can be reused to fetch data directly from the various sources.
Ultimately, my goal is not to create yet another list but to be able to compare, merge existing ones, and put the information on Wikidata so it can be used even more as a reliable source.
Below I report in more detail what I learned from each source and the various shortcomings I discovered. This may be important for understanding what can be created with the data.
### Notes on the different sources
#### SCB's agency register
Since 2007 ([SFS 2007:755](https://www.riksdagen.se/sv/dokument-och-lagar/dokument/svensk-forfattningssamling/forordning-2007755-om-det-allmanna_sfs-2007-755/)), Statistics Sweden (SCB) has been tasked with maintaining an agency register with basic information. As the ordinance states, the register should contain courts, public enterprises, and foreign agencies (embassies and delegations at international organisations) but not committees or special investigators. Since it was written in 2007, there was only an instruction to make the register "available to the public through a public website".
It is a bit unclear why SCB, one of Sweden's largest and best agencies for open data, which pumps out high-quality statistics every year, has chosen to fulfil the assignment at such a minimum level. The best example is that many agency names are written in UPPERCASE. Several administrative agencies and boards found in other lists are also missing, but it is unclear if SCB has chosen not to include them or just forgotten them. The quality varies between the different categories, which indirectly shows that some must be updated more manually than others.
The worst quality can be found in the information about foreign representation which probably comes from the Ministry for Foreign Affairs (UD). Some examples: "Svergies ambassad Santigao", "Sveriges Ambassad Nur-Sultan" (Kazakhstan's capital is now named [Astana](https://en.wikipedia.org/wiki/Astana) again) but also Sweden's representation to the OSCE or the Swedish Dialogue Institute for the Middle East and North Africa which seem to have been completely forgotten (aren't they agencies?). The safest source is then UD's website, but they do not publish a list, only information pages for Swedes abroad. SCB also publishes a file with parliamentary agencies (which was not requested in the ordinance) but it is unclear why it only includes 1 of the [9-10 boards](https://www.riksdagen.se/sv/kontakt-och-besok/riksdagens-myndigheter-och-namnder/) that the Riksdag lists on its website.
Besides quality, it is also surprising that SCB blocks you if you download too many files. It becomes difficult to build a data infrastructure on top of it, and there is little reason to limit the download of 6 static files from a technical perspective. On the website, SCB also publishes [a list](https://myndighetsregistret.scb.se/Ar) of closed and new agencies for each year since 2008. Unfortunately not as a structured list, so I have to [scrape](https://en.wikipedia.org/wiki/Web_scraping) and clean it before merging it with the register.
#### Statskontoret's Fakta om statsförvaltningen
Statskontoret is tasked with following and describing the development of the public sector and publishes a list of agencies in an Excel file in connection with its qualitative reports. It is undoubtedly the most reliable source, not least because the agency receives targeted assignments that complement the overall picture of state administration. It was during an assignment to map board authorities that 25 small boards were found and added to the list. Statskontoret is also very keen to define what they consider to be an agency and what is included in their mapping: foreign representation and parliamentary agencies are excluded, as well as temporary ones.
The list contains many details that must have been collected manually, such as alternative names, abbreviations, and management form, and it is so complete and detailed that it could be likened to a work of art. The disadvantage of manual collection, however, is that it becomes difficult to keep it fully updated and error-free. Some minor problems I could detect are [a court that changed names in 2022](https://www.domstol.se/nyheter/2022/11/hudiksvalls-tingsratt-blir-halsinglands-tingsratt2/) and [an agency that did so in 1977](https://sv.wikipedia.org/wiki/Universitetskanslers%C3%A4mbetet#Myndighetens_f%C3%B6reg%C3%A5ngare) (Statskontoret still uses the old spelling without "s"). A bigger problem is that the list has no ambition to be a register and that the data is not produced to be machine-readable. The file has no permanent URL, a trick must be used to automate the download. The format and Excel formulas in some cells make it difficult to extract the information. In some fields, comments occur instead of values.
#### The Swedish National Financial Management Authority (ESV)
ESV (Note: my current employer, but this post is not endorsed by the agency) has a fairly good grasp of Sweden's agencies for various reasons:
* ESV manages the state's accounting system (HERMES) where a larger part of the agencies report in,
* ESV collects the agencies' annual reports,
* ESV manages the IT system Statsliggaren where the Government Offices produce the agencies' *regleringsbrev* (steering letters).
It may seem confusing, but these three sources do not yield the same lists of agencies. From the first, ESV's agency register is created, but it is not complete because many agencies do not report in the state's accounting system. Nor does every agency have an annual report or receive an *regleringsbrev*; the smallest only have an instruction (*instruktion*).
Despite this, ESV is an interesting source because the agency publishes a lot of historical data, in some cases dating back to 1999.
Regarding data quality, there is not much to say. Some fields contain strange data such as "?" and since the agencies themselves enter their information in the system, several have forgotten to keep them updated (e.g. [PRV changed English name in 2020](https://www.altinget.se/nyttomnamn/0-myndighet-byter-namn) but has the old name left in HERMES). The last two lists must be scraped from the website while the first is available as a spreadsheet.
#### The Swedish Agency for Government Employers
The Swedish Agency for Government Employers is the state's employer organisation and an agency in itself. They manage a list of their members which both lacks certain agencies and includes other organisations. Despite this, they are an important source for HR statistics. Their open data is unfortunately very limited, but it is possible to download a list of members since 1980, which includes many agencies that no longer exist. Since some members are not agencies (e.g. the Church of Sweden (*Svenska kyrkan*) which was a member before 2000), they must be manually removed from the list. Format-wise, they have chosen to offer their statistics using Tableau to visualize them. This makes it impossible to automatically download data and the file is not entirely easy to convert to structured data.
#### The Government Offices' code of statutes
Not per se a list of agencies but perhaps the most important source because it contains all agencies' instructions. Not all agencies have a *regleringsbrev* or an annual report, but basically everyone has an instruction. The vast majority of instructions have a very clear name "Förordning med instruktion för XXX" or "Lag med instruktion för YYY" for parliamentary agencies, which makes it very easy to find the vast majority of agencies ever since the 1960s. Unfortunately, some agencies share an instruction so it is not possible, for example, to know how many [local safety boards at nuclear facilities](https://beta.rkrattsbaser.gov.se/sfs/item?bet=2007%3A1054) there are without linking it to another source (like Statskontoret's list).
Otherwise, the database is of very high quality, with a few exceptions ([Formum för levande historia](https://beta.rkrattsbaser.gov.se/sfs/item?bet=2002%3A795)), and the new beta website that the Government Offices have created has a very good API so it is possible to download the entire Swedish code of statutes very easily. The API is private and undocumented so it may not be enough to say that the Government Offices have completely started with open data, but it shows what a modern [regeringen.se](http://regeringen.se/) could look like in the future.
#### Wikipedia and Wikidata
Everyone probably knows Wikipedia, but fewer may have heard of its sibling project Wikidata. The database contains much of Wikipedia's information in a structured format (SPARQL), which makes it possible to fetch all agencies and a lot of metadata (if available). Thanks to the project's many volunteers and initiatives like [Govdirectory](https://www.wikidata.org/wiki/Wikidata:WikiProject_Govdirectory) and [Wikidata Riksdag](https://www.wikidata.org/wiki/Wikidata:WikiProject_Sweden/Swedish_Riksdag_documents), there are many Swedish agencies on Wikidata, and much more!
In the long run, I hope my little project allows us to gather even more information there and keep it updated.
**EDIT:** After I published this blog post, I used the data collected in the project to clean and complete hundreds of Wikidata items and Wikipedia pages. I hope it can help more people in the future.
#### Handlingar.se
Another unofficial source is [handlingar.se](http://handlingar.se/), an important initiative from Open Knowledge Sweden to modernise Sweden's freedom of information act (*offentlighetsprincipen*). It is a platform that makes it easy to request public documents, see public actors' responses, and publish what is released as open data. Elenor Weimar, Mattias Axell and others do a great job with this service, which is based on open source and exists in [many countries in the world](http://alaveteli.org/deployments/), and as part of this project, they release a list of organisations covered by the freedom of information act, which they publish as open data (CSV). The list contains more than just agencies, but it is possible to filter on certain categories.
# g0v.se
{{< icon-button "g0v.se" "link" >}} g0v.se {{< /icon-button >}}
{{< icon-button "github.com/civictechsweden/g0vse" "github" >}} Source code {{< /icon-button >}}
{{< icon-button "github.com/civictechsweden/g0vse/tree/data" "github" >}} Data {{< /icon-button >}}
I arrived in Sweden in 2016 and I quickly started to build services building on government and legislative data. The Swedish parliament (*Riksdagen*) publishes open data since 2010 on all its activities and production. Unfortunately, all the activity of the government as well as preparatory work for legislation is controlled by the government chancellery offices (*Regeringskansliet*) and they have very little interest in open data, transparency and democratic innovation. They are also generally struggling with digitalising themselves.
I know that because I have met with them countless times and tried to get them to improve the situation, to no avail. So I tried to make information from their website ([regeringen.se](https://www.regeringen.se)) more accessible through projects such as Din Riksdag, OpenRemiss. And with each one of them, I gained a better understanding of their systems, the way their data is organised. On the way, I also met countless journalists, researchers, companies and even government bodies struggling with the same problem.
And so in 2024, I decided to put all that accumulated knowledge into g0vse, a project that downloads all information from the government's website every night and makes it available as open data in a format that people can actually reuse.
You can read more (in Swedish) on [g0v.se](https://g0v.se) and about the technical aspects (in English) on [Github](https://github.com/civictechsweden/g0vse).
# 5 years exploring democracy
**THIS ARTICLE WAS ORIGINALLY RELEASED ON [LINKEDIN](https://www.linkedin.com/pulse/5-%25C3%25A5r-p%25C3%25A5-en-demokratiresa-pierre-mesure/) (IN SWEDISH).**
On May 24th, I turned 30. It was not a bitter celebration. On the contrary, I feel happier with each passing year and in many ways more secure in my life journey. Around the same time, I also celebrated 5 years as an official employee at Digidem Lab. I actually started a little earlier, but it was the funding we received from Vinnova for our Democratic Cities project that paid for my first salary in April 2018.
Digidem Lab was founded earlier in 2016 as a Heritage Fund project by Sanna Ghotbi, who had completely lost hope in our representative democracy after 4 years as the City of Gothenburg's youngest council member, Anna Sanne Göransson who had seen the need for more participation in Bergsjön after several years as a social worker, and Petter Joelson who realized already then how digitalisation would affect our ways of organizing ourselves.
I had a slightly different entry point. Freshly graduated as an engineer, I wanted nothing more than to move abroad again after a successful semester in Norway and a few months in England. While looking for my dream job from my teenage room in the Paris suburbs, I spent my time at meetings and meetups in the then-blossoming civic tech community. It was a time when apps were invented every week to visualize the legislative process, collect petitions, or plan a political campaign. I was also deeply interested in the new wave of hackers entering the public sector to reform it with more openness and participation.
In 2015, the Law for a Digital Republic had been written on a digital platform with more than 100,000 contributions and introduced new principles into the law such as openness by default. Instead of requesting public documents, one could instead ask authorities to make them available online, preferably as open structured data. And all public organisations needed to immediately start opening everything that could be opened. As a logical consequence, France took the presidency of the Open Government Partnership and organized its *OGP Global Summit* the following year. Paris welcomed the world's openness activists and I started to feel that maybe this was something I wanted to use my powers for. At the same time, the capital's participatory budget returned with a pot of money worth over 1 billion SEK that Parisians could decide directly over. On every corner, there was a voting station or a pallet with individuals pitching their project.
Don't get me wrong, my homeland's democracy suffers from many flaws and that same autumn I was also at demonstrations against a labor market reform that the government pushed through without a vote in parliament. Even today, these problems persist. But then the need for new ways to organize the state and redistribute power felt even more acute.
At the same time, I mostly wanted to go abroad and I was starting to get depressed looking for jobs from my old childhood home. I bought a ticket. It would be Stockholm in the end, and I would have to look for work and housing when I arrived.
Both were surprisingly easy to find, but I started almost simultaneously to look for the same things that had inspired me so much in France. Sweden and the rest of the Nordics were praised worldwide for their superior democracy, there was surely participatory budgeting in every village and deep involvement of citizens in all decisions that concerned them. Sweden was also the country that invented the principle of public access to official records (*offentlighetsprincipen*), the country where everyone's personal data was public online and where Mona Sahlin had to resign because of a chocolate bar, so I also expected open data everywhere and total transparency from public organisations.
I walked around different meetups in Stockholm, traveled to Visby during Almedalen Week, and knocked on many organisations' doors to understand where these innovations existed. I started meeting the enthusiasts pushing for them from the inside, but I also clearly noticed that much remained to be done.
And after about a year, I found Anna, Petter, and Sanna in the little ship that had recently started sailing: Digidem Lab. I barely realized then what adventures I was saying yes to when I came aboard.
As mentioned, I started with funding from Vinnova and our first project with Swedish municipalities. We had managed to find enthusiasts in both Gothenburg and Stockholm to try to test new ways of involving citizens.
Much of what we tested with them we imported from the various countries we studied. In Paris, we met many civil servants, politicians, activists, and citizens to understand how the city's participatory budget worked. On our blog, we wrote several posts that would, for example, lead several housing companies in Sweden to get started with the method with their tenants (the so-called tenant budgets). At Madrid's Medialab Prado, we met new democracy friends from all over the world and developed with them new analog and digital ways to influence. We imported open tools like CONSUL and Decidim which quickly started spreading in Sweden.
Pretty soon our small pilot projects grew and we needed to recruit new stars: Annie, Steph, David…
In parallel, we started the network Civic Tech Sweden with allied organisations and individuals in Sweden to organize meetups, hackathons, and easier find and support each other. We had the honor of inviting guests from many Swedish CSOs, journalists, authorities, but also activists from Taiwan, Canada, Estonia... 😊
In 2020 came the pandemic and our whole world collapsed. How would we organize participatory processes on the ground in Biskopsgården, Skärholmen, Helsingborg when all municipalities advised against physical meetings and the whole country moved to Zoom and Teams? We were not alone with that question and many organisations from all over the world started contacting us to solve these new challenges. We started working with New York City's democracy department, supporting participatory budgets in Chicago's vulnerable areas, and planning the EU's major Conference on the Future of Europe with the EU Commission.
All these projects were also occasions where we got the chance to collaborate with other wonderful democracy labs around the world. Even if we can sometimes feel a little alone in our industry in Sweden, there are siblings in every country that we could share experience and collaborate with: Open Source Politics and Missions Publiques in France, DBT and We do Democracy in Denmark, Deliberativa and Platoniq in Spain… Too many to mention them all. Everything we have learned we have received from them and I want to take the opportunity to thank them for the endless support we have received. Because of that, it has always felt obvious that everything we create should be available to everyone. Knowledge, methods, code…
In 2022, we finally got the chance to work with a method that we had been trying to promote for over 5 years: citizens' juries. First with the City of Gothenburg's transition strategy, then with the Swedish Food Agency which hired us to conduct Sweden's first national citizens' panel. It was an incredibly educational year and I felt so grateful to be able to perform this assignment with such a great team!
I hope that this first attempt leads to many more citizens' panels in Sweden and I feel confident in Digidem Lab's capacity to continue supporting these future processes, to continue pushing the boundaries of what the word democracy means and to stand by the side of everyone trying to make our country and our world better. 😊
With increased polarisation, an acute climate crisis, and a systemic dismantling of our welfare, Sweden needs more than ever to find new ways to come together and create a society for all.
For my part, a new chapter begins in October, where I hope to also be able to use my competence to influence. More about that coming soon.
I want to take the opportunity to thank everyone who has supported us, who has supported me during the last 5 years. And the biggest thanks to my brilliant colleagues! You are stars!
# Hej Ulf!
{{< icon-button "webperf.se/articles/hej-ulf/" "pencil" >}} Original article (webperf.se) {{< /icon-button >}}
{{< icon-button "hejulf.mesu.re" "link" >}} hejulf.se {{< /icon-button >}}
{{< icon-button "hejulf.mesu.re/beskrivning-klagomal.pdf" "pencil" >}} Complaint to IMY {{< /icon-button >}}
{{< icon-button "github.com/PierreMesure/hejulf.se" "github" >}} Source code (website) {{< /icon-button >}}
{{< icon-button "github.com/PierreMesure/hej-ulf" "github" >}} Source code (scraper) {{< /icon-button >}}
## Hej Pierre
2022 was an election year in Sweden. Political parties were campaigning at national and regional levels and following an ongoing trend, a larger share of the campaign was now digital.
At the beginning of September, I received an e-mail from Ulf Kristersson, then prime minister candidate with the subject "Hej Pierre". It asked me to open a video with a personalised message. The video featured Ulf Kristersson seated at a desk looking at me and reciting some points of his manifesto. But several things caught my attention: Ulf Kristersson was also starting with "Hej Pierre" (which startled me like I guess many others who received this message) and at some point in his speech, Ulf Kristersson pointed to a tablet on his desk where the mayor of my city, Solna, had recorded a message encouraging me to vote for their party locally.
> Freaky, I thought. Why am I receiving this? How does Ulf Kristersson know about me, my name and my address?
## Analysing the ad campaign
So I decided to investigate. I started by exploring the website on which the video message was embedded. I discovered the company behind the ad campaign, the technical method used to stitch video segments together, creating a personalised illusion when most content was identical for everyone. I also managed to algorithmically recreate several hundreds of thousands of unique links corresponding to as many personalised messages that were sent to other people: "Hej Annika", "Hej Waldemar", "Hej Anna". It seemed that Ulf Kristersson had spent half a day recording over 2000 *hälsningar*, and you could even see how the light from the window changed between the snippets.
When I talked about it with friends, I realised some of them had received other personalised variations of the video. All had the same reaction: spooky, and is this even legal?
I gathered a list of GDPR violations from Ulf Kristersson's party (*Moderaterna*) and their technical partners and I discussed it with several specialists. I got valuable feedback and decided to submit a complaint to the Swedish authority for privacy protection (*Integritetsskyddsmyndigheten, IMY*). Unfortunately, it is infamously known for its reluctance to investigate complaints and I knew that my efforts would likely end up in the bin as well.
## Building a counter-campaign
But I also knew that many people got similar video ads (from my investigation, I think they were over 400 000) so I decided to give these people an opportunity to stand behind the complaint.
IMY doesn't accept group complaints so I created a simple website where anyone could send their own complaint with a few clicks (it used to be at hejulf.se but it's now moved to [my domain](https://hejulf.mesu.re)). I also wrote a [blog post on webperf.se](https://webperf.se/articles/hej-ulf/) and I posted everything on Twitter and LinkedIn with a funny video. I waited until after the election night both because I wasn't completely ready before and because I wasn't sure of the reception in the final days of the campaign.
{{< video src="https://github.com/PierreMesure/hejulf.se/raw/refs/heads/main/public/hejulf.mp4" type="video/mp4" preload="auto" >}}
It spread quicker than I imagined and I rapidly got hundreds of messages from people who had also received the message. About 150-250 managed to send a complaint despite some technical issues on my website. I also received a lot of hate messages on Twitter from political supporters on the right.
The story got picked up by specialised media and I got interviewed by [Dagens Media](https://www.dagensmedia.se/medier/digitalt/moderaternas-sms-utskick-massanmals-efter-kampanj-kande-mig-krankt/)/[Resumé](https://www.resume.se/marknadsforing/strategi/moderaternas-sms-utskick-massanmals-efter-kampanj-kande-mig-krankt/).
## What happened since then?
Six months later in March 2023, IMY decided to [start an investigation](https://www.imy.se/nyheter/imy-inleder-tillsyn-av-moderaterna/) (*tillsyn*), which was in itself a victory since they only did so for [less than 5%](https://www.datajurist.se/imy-lyssnar-inte-forsta-gangen/) of the complaints at that time. Most news outlets talked about it but none went into the details of what happened (like [SVT](https://www.svt.se/kultur/myndighet-inleder-tillsyn-mot-moderaternas-videohalsningar), [DN](https://www.dn.se/sverige/myndighet-granskar-moderaternas-valhalsning/), [Aftonbladet](https://www.aftonbladet.se/nyheter/a/zEVjeq/myndighet-granskar-m-efter-valet), [Altinget](https://www.altinget.se/artikel/myndighet-ska-granska-moderaternas-valsatsning)...).
Then, this investigation disappeared into some kind of black hole before IMY finally came to a [decision](https://www.imy.se/tillsyner/moderata-samlingspartiet-riksorganisationen/) at the end of October 2025. On [public radio](https://www.sverigesradio.se/artikel/myndighet-fel-av-m-och-sd-att-skicka-reklam-pa-sms), its author explains that the decision should lead to a change of legal praxis in the way political parties target voters during elections with SMS and e-mails. Consent is now more or less required, especially if the content is adapted to the recipient using personal data.
In my opinion, this is a decision that has far-reaching consequences: the use of contact details from commercial actors buying addresses from Skatteverket, phone numbers from mobile carriers and hiding behind a so-called *utgivningsbevis* is widespread in advertisement. And although IMY is still very reluctant to attack these actors, their clients do not benefit from the same protection (like Moderaterna in this case) and will now hopefully be less willing to use this data without consent.
In politics, I hope that this can be a small brick in a bigger wall that protects voters against disinformation and manipulation. We have seen how very targeted advertisement in the USA in 2016 [could lead to polarisation and manipulation at massive scales](https://en.wikipedia.org/wiki/Project_Alamo) and the recent developments in dark patterns and AI will make it easier in the future. On the other hand, this raises the democratic question of how parties should be able to reach out to voters to inform about their ideas. In response to the EU's attempts to regulate political advertising (DSA and TTPA), most social media platforms [have decided to forbid it altogether](https://opentermsarchive.org/en/publications/political-advertising-after-the-ttpa-exit-adaptation-and-the-future-of-platform-governance/). We need to develop new and better democratic spaces to debate ideas and nurture enlightened citizens.
# Almedalsdata, diving into the data of the world’s biggest political festival
**THIS ARTICLE WAS ORIGINALLY RELEASED ON [MEDIUM](https://medium.com/civictechsweden/almedalsdata-diving-into-the-data-of-the-worlds-biggest-political-festival-267eb6865860).**
For those who don’t know about it, Almedalsveckan is a political festival taking place every year on the first week of July in Visby, a cute town on the island of Gotland. It was initiated around 50 years ago by then education minister Olof Palme and has now grown to be what I believe is the biggest political festival in the world. Thousands of events, tens of thousands of attendees and about every important Swedish political or economical stakeholder!
Ever since I arrived to Sweden, I’ve been fascinated by Almedalsveckan . I’ve been there every year up to the pandemic and it quickly became part of my Swedish summers. I can safely say it helped me a lot to understand the Swedish society, its political system and culture. I still remember how amazed I was the first year I attended. I could barely speak Swedish but I already managed to speak to two ministers and a few MPs (not sure they understood me 😁).
As my local network and my understanding of the system grew, I also learnt to see the dark side of this festival, how big corporate lobbying and money have become and how exclusive it is today. While there has been criticism about that worsening trend, I have yet to find some ambitious journalistic or research work about lobbying (completely possible that I missed it, please send your book/article tips!).
## The festival program
One object that always fascinated me is that thick program that you get handed when you exit the boat and enter Visby’s city center for the first time. A long list of all the events (actually, all the public events but it took me some time to understand it) that I had to religiously browse to make my own list.
Could that program be a good starting material to understand what interests have come and passed in the public debate? To evaluate the growing influence of private interests on Swedish politics?
Although the book itself is freely distributed and the [website](https://program.almedalsveckan.info) is public, all of these questions were hard to answer. As usual in Sweden, structured data of really good quality exists but the idea of making it available as open data is quite remote… That’s why I decided to have a closer look and today I present to you…
## Almedalsdata
Almedalsdata is a small civic tech project I worked on in my free time. The first step was to find as much of the programs as I could online. The second was to write a small script to fetch that data and convert it to a structured format.
The result is available on Github [here](https://github.com/civictechsweden/almedalsdata).
### What data is available?
As of now, I managed to gather all events from 2009 to 2021 .
For each event, the title, organiser, speakers, description, dates, address, category and a few more metadata are saved.
I’m probably going to add the years 2003 to 2008 soon but I only found event lists without much information about them. Nothing earlier than 2003. Don’t hesitate to contact me if you know about sources that I haven’t found.
### What can be found in the data?
A lot, but I haven’t had time to dig yet. This dataset could for example be used to trace the career of some politicians. Have a look at Nooshi Dadgostar for example:
- 2016: [Bostadspolitisk talesperson, (V)](https://program.almedalsveckan.info/event/user-view/42157)
- 2018: [Vice ordförande, Vänsterpartiet](https://program.almedalsveckan.info/event/user-view/51842)
- 2021: [Partiledare, Vänsterpartiet](https://program.almedalsveckan.info/event/user-view/60998)
In the same spirit, it could also help to identify some revolving doors:
- 2016: [sektionschef för Center för e-samhället, SKL](https://program.almedalsveckan.info/event/user-view/42241)
- 2018: [CDO of Sweden](https://program.almedalsveckan.info/event/user-view/50206)
- 2019: [Förbundsdirektör, IT&Telekomföretagen](https://program.almedalsveckan.info/event/user-view/59563)
For now, I’ve only gathered a few interesting statistics such as the amount of events per year:
After a stable growth over the past ten years, culminating in the 50th anniversary of the festival (and election year), Almedalsveckan was simply cancelled in 2020 before reappearing in a much humbler digital 2021 edition.
The 10 most active organisers and the 10 most popular event categories:
And finally, maybe the one piece of information you couldn’t live without!? The proportion of events offering food to attendees, available since 2011!
The amount of events attracting people with a free breakfast/fika seems to be steadily decreasing since the beginning of the 2010s. The first pandemic Almedalsveckan reaches a bottom of only 4 events with food in 2021.
## Now, your turn!
What will you find in this data? It’s all yours to [explore](https://flatgithub.com/civictechsweden/almedalsdata?filename=summary.csv) and [reuse](https://github.com/civictechsweden/almedalsdata)! Please mention the project if you do! And don’t hesitate to contribute if you know where to find more data!
# Jag vill ha vaccin!
{{< icon-button "jagvillhavaccin.mesu.re" "link" >}} jagvillhavaccin.nu {{< /icon-button >}}
{{< icon-button "github.com/civictechsweden/JagVillHaVaccin" "github" >}} Source code (scraper) {{< /icon-button >}}
{{< icon-button "github.com/civictechsweden/JagVillHaVaccin-front" "github" >}} Source code (frontend) {{< /icon-button >}}
## Finding a vaccine time in Sweden
Like other countries, Sweden began vaccinating its population in early 2021, starting with the most vulnerable and gradually expanding the campaign to younger age groups. However, Sweden's healthcare system is regional and decentralised, with surprisingly little collaboration between regions. In particular, the digital systems used in Swedish healthcare are often criticised for their lack of common data standards and APIs, and a tendency to either reinvent the wheel (at great expense) or fall into the hands of ineffective monopolies[^fn:1].
This was a poor foundation for a vaccination campaign where everyone expected to book appointments online. The result was exceptionally... varied, though I could not have anticipated quite how chaotic it would become when I first started this project.
A few regions used 1177, a robust service in terms of authentication and interface, but one that completely lacked a way to get an overview of available times and locations. The regions of Stockholm and Gotland chose their *Alltid Öppet* app, a proven solution for other healthcare bookings. Unfortunately, the app did not allow users to see available slots without logging in via BankID. This led to numerous crashes when hundreds of thousands of people tried to access the few hundred slots added periodically, which in turn hindered patients trying to access more critical care[^fn:2]. It was also reported that the Region of Stockholm had to invest several million crowns in server capacity just to meet this largely unsuccessful demand. One region simply abandoned online bookings entirely, choosing to contact all citizens individually by phone. Finally, the remaining regions opted for private solutions with surprisingly poor user experience. In Södermanland, users had to manually click on every vaccination centre to check for availability in the current week, then click through several times to explore subsequent weeks. The worst experience was for citizens in Västra Götaland[^fn:3] and Skåne[^fn:4] [^fn:5], where authorities delegated the booking infrastructure to each individual centre. Min Doktor, Din Doktor, Doktor.se... In cities like Gothenburg and Malmö, it was not uncommon to have to create 3–4 different accounts just to see the (un)available slots in the centres within a single district.
In comparison, across the Öresund, the Danish government had established a single portal. Right next to the login button, a clear sign deterred users from logging in if no appointments were available.
## The French example
France was initially even more chaotic; much like the worst Swedish examples, the government had delegated booking responsibility to each vaccination centre. These centres used 5–6 different e-healthcare platforms. While they showed available slots without requiring a login, it was still far from easy to navigate.
France has a strong culture of "hacktivism", individuals who create software and use their skills for the common good. Many solutions emerged during the pandemic, but one gained incredible traction: [Vite Ma Dose](https://vitemadose.covidtracker.fr). Started by Guillaume Rozier, the platform aimed to gather vaccine bookings into a single, simple interface for the entire country. The [code](https://github.com/CovidTrackerFr/vitemadose) was open-source, and the website scraped available slots from hundreds of vaccination centres every minute.
Quickly, over 100 people joined the project, maintaining data flows and improving the website's search functions and accessibility. Within days, mobile apps were available on app stores[^fn:6], and communication materials were created to reach all audiences. Traffic exploded. Recognising the benefits of this grass-roots solution, the French government quickly decided to support and promote the initiative. Vaccine platform providers also established APIs to make it easier for their data to be republished, seeing the new site as a welcome buffer that relieved their own servers from the millions of visits they had experienced early in the campaign. By the end of the year, over 100 million visitors had used the website.
Guillaume Rozier was eventually awarded the National Order of Merit, with the government praising his and others' initiatives for saving lives and helping millions of people[^fn:7].
## Creating "Jag vill ha vaccin!"
Like most French people, I heard of this initiative when the media began covering it in early April 2021. The source code was straightforward, and I decided to spend a few hours investigating whether it would be possible to reproduce the same service in Sweden. I looked into how vaccine bookings were made available across all regions and quickly realised that many could be scraped automatically. None provided open data, but a script simulating user behaviour could save the information in a structured format. I spent a full weekend setting up a prototype. By the end, I had a version based on the original *Vite Ma Dose* code, initially still in French, but displaying Swedish vaccination centres. I quickly translated the page and published it under the name "Jag vill ha vaccin!" (I want a vaccine!) with a short post on LinkedIn and Twitter. It took off almost immediately.
## The reception
Despite my limited reach on these platforms, the posts went viral, and over 10,000 people visited the website on the first day. Volunteers reached out to help with proofreading and data analysis, and several journalists contacted me[^fn:8] [^fn:9] [^fn:10]. I was taken aback, especially as their questions often framed my modest prototype as a solution to the failures of the regions and their massive budgets. At the time, I knew little about just how flawed some of their systems were; I had mostly started the project as a fun experiment.
What I also hadn't foreseen was that these same journalists would ask the regions why a single individual could deliver what they could not. Press secretaries, often without understanding the service, advised against using it[^fn:11], citing potential data theft and security risks. Some platform providers even modified their systems to make it harder for me to retrieve data. Since none of the regions reached out to me, I took the initiative to contact them myself.
I contacted every region and managed to secure meetings with about half of them[^fn:12]. Most were adamant that I should stop. Their reasoning was often confused and driven by fear, though one valid argument was that users might miss mandatory information by skipping the official landing pages. I addressed this by ensuring all users were redirected to the official information pages on the respective regional websites. Another concern was the hypothetical risk of the service distributing viruses. *Jag vill ha vaccin!* did not track users or collect any personal information, but regardless of who I spoke to, the regional employees' technical understanding was limited. I suggested they reuse the source code and host it themselves, but the culture of open source is unfortunately weak in the Swedish public sector, and no region accepted the offer[^fn:13].
I vividly remember a meeting with Daniel Forsberg, then a leading politician on the matter for Region Stockholm (now a lobbyist). I offered to display their vaccination slots on *Jag vill ha vaccin!* to relieve the heavy traffic on their app. He repeatedly stated how "pro-innovation" he was and how much the region valued open data, but explained that in this specific case, it was unfortunately not welcome[^fn:12].
In the end, only one region welcomed the collaboration: Västra Götaland. Regional executives had recognised the disastrous situation they were in, where over 15 private providers were releasing slots on proprietary platforms, making it harder for citizens to book and for them to get a clear overview of dose distribution. A small team, including Marcus Österberg (creator of [webperf.se](https://webperf.se)), was tasked with building an official aggregation page. A strong advocate for internet standards and openness, Marcus quickly published the region's data openly[^fn:14].
This allowed me to make the service more comprehensive. Ironically, *Jag vill ha vaccin!* still displayed more slots than the region's official platform, as their team had to ask for data from private providers[^fn:15]. I simply scraped it without seeking permission, knowing that the Freedom of Information Act and the public interest were on my side.
## The end
Unfortunately, the project never received the institutional backing I had hoped for. Despite popular support and media coverage, I could only aggregate about half of the country's vaccination slots, and several regions insisted I shut the service down. I was even personally threatened with lawsuits by the manager of one healthcare centre.
Although I never shut down the website[^fn:16] [^fn:17] [^fn:18], the number of visitors gradually decreased after the initial peak. *Jag vill ha vaccin!* eventually became irrelevant once the majority of the population had received their first dose and scarcity was no longer an issue.
Whenever journalists reached out, they were often seeking critical quotes to build an antagonistic narrative. I explained that the entire history of vaccination slots gathered by the site was available as open data and could be used to analyse the effectiveness of the rollout. I used it myself to report on centres with hundreds of available slots while other parts of the same region had none. Unfortunately, this kind of analysis required more time than most journalists were willing to invest, so the data remains mostly unused to this date (if you are a researcher, feel free to message me—I have archived it 😉).
## Moral of the story
Unlike its French counterpart, and to my initial disappointment, *Jag vill ha vaccin!* never became a government-backed success story. However, success was never the primary goal, and this project taught me a great deal. Not only technically but also about the challenges of innovating in the Swedish public sector.
## References
[^fn:1]: [Detta gick snett med Millennium i VGR (lakartidningen.se)](https://lakartidningen.se/nyheter/externa-granskningen-detta-gick-snett-med-millennium-i-vgr/)
[^fn:2]: [Alltid öppet var inte alltid öppet (dn.se)](https://www.dn.se/sverige/alltid-oppet-var-inte-alltid-oppen-darfor-kraschade-appen/)
[^fn:3]: [Svårt boka vaccintid när spärren lyftes: ”Det fungerar inte” (molndalsposten.se)](https://www.molndalsposten.se/nyheter/svart-boka-vaccintid-nar-sparren-lyftes-det-fungerar-inte.f68bd423-3165-4599-9bf1-839d6290ec03)
[^fn:4]: [Vaccinationernas fas 4 började med kaos (sydsvenskan.se)](https://www.sydsvenskan.se/skane/vaccinationernas-fas-4-borjade-med-kaos-sedan-tog-tiderna-snabbt-slut/)
[^fn:5]: [Vaccinstrulet tvingar regionen annonsera för att fylla tiderna (sydsvenskan.se)](https://www.sydsvenskan.se/naringsliv/thomas-frostberg-vaccinstrulet-tvingar-regionen-annonsera-for-att-fylla-tiderna/)
[^fn:6]: [Vite Ma Dose tops the App Store rankings (in French, leparisien.fr)](https://www.leparisien.fr/high-tech/la-vaccination-elargie-et-le-pass-sanitaire-dopent-les-telechargements-de-vitemadose-et-tousanticovid-21-05-2021-3H7Y6E2FCNHWXC264ZM6DRUE5Q.php)
[^fn:7]: [Guillaume Rozier receives the medal of national merit (in French, 20minutes.fr)](https://www.20minutes.fr/societe/3046971-20210522-guillaume-rozier-createur-vitemadose-recoit-ordre-national-merite)
[^fn:8]: [Nystartad privatsajt visar tusentals svenska vaccintider (svd.se)](https://www.svd.se/a/561W4E/privatsajt-visar-tusentals-svenska-vaccintider)
[^fn:9]: [Pierre skapade sajten som är lösningen på vaccin-kaoset (emanuelkarlsten.se)](https://emanuelkarlsten.se/pierre-skapade-sajten-som-ar-losningen-pa-vaccin-kaoset-sa-hittar-du-lediga-tider-i-din-region/)
[^fn:10]: [Han listar lediga vaccintider (svt.se)](https://www.svt.se/nyheter/inrikes/han-listar-lediga-vaccintider)
[^fn:11]: [Regionens uppmaning: Använd inte nya bokningssidan (svt.se)](https://www.svt.se/nyheter/lokalt/blekinge/regionen-avrader-anvandning-av-ny-bokningssida)
[^fn:12]: Presentations from my meetings with the regions are available [here](https://github.com/PierreMesure/pierre.mesu.re/tree/master/static/jagvillhavaccin/presentations)
[^fn:13]: [Regionen inte så intresserade av nya sajten för att boka vaccin: ”Vår egen fungerar ganska bra” (smp.se)](https://www.smp.se/vaxjo/regionen-inte-sa-intresserade-av-nya-sajten-for-att-boka-vaccin-var-egen-fungerar-ganska-bra/)
[^fn:14]: [Hjälp VGR testa vårt API med öppna vaccintider (vgrblogg.se)](https://vgrblogg.se/utveckling/2021/05/27/hjalp-vgr-testa-vart-api-med-oppna-vaccintider/)
[^fn:15]: [Sidan skulle samla lediga tider - ännu saknas manga (archived)](https://github.com/PierreMesure/pierre.mesu.re/tree/master/static/jagvillhavaccin)
[^fn:16]: [Egenutvecklad vaccinationssajt igång igen: Listar många lediga tider (nyteknik.se)](https://www.nyteknik.se/nyheter/egenutvecklad-vaccinationssajt-igang-igen-listar-manga-lediga-tider/352149)
[^fn:17]: [Regionerna motarbetade sajten som ville effektivisera vaccinbokning – nu är den tillbaka dubbelt så stor (emanuelkarlsten.se)](https://emanuelkarlsten.se/regionerna-motarbetade-sajten-som-ville-effektivisera-vaccinbokning-nu-ar-den-tillbaka-dubbelt-sa-stor/)
[^fn:18]: ['It's a democratic tool': The site that helps you find a Covid vaccine slot in Sweden (thelocal.se)](https://www.thelocal.se/20210615/its-a-democratic-tool-the-site-that-helps-you-find-a-covid-vaccine-slot-in-sweden)
# 10 tips before HackTheCrisis Sweden
{{< alert "circle-info" >}}
This text was written with **Anna Fahlgård**, PhD student at House of Innovation, Stockholm School of Economics.
{{< /alert >}}
**THIS ARTICLE WAS ORIGINALLY RELEASED ON [MEDIUM](https://medium.com/civictechsweden/10-tips-before-hackthecrisis-sweden-15799b8785d).**
You just registered for this weekend's event HackTheCrisis Sweden? Maybe it's even your first hackathon? Great! The more, the merrier!
Crowdsourcing can be an invaluable source of innovation when the right type of crowd tackles the right type of problem. Before you dive right in, we thought we'd give you a few tips and a few of our own "lessons learned" from taking part in a couple of hacks on **what not to do** in order to make a great contribution in the fight against the virus and its aftermath.
## 1. Don't reinvent the wheel
HackTheCrisis Sweden is but one of many (**46 so far!**) other HackTheCrisis initiatives in the world and a lot of them have already taken place. This makes for an excellent opportunity to refine and recombine what has already been done! Before you start working on your genius idea, make sure you check out the other hackathons' projects, many of them are in English and publish their code and other resources that you could build on!
Check out for instance **WirvsVirus**' [1500 projects](https://wirvsvirushackathon.devpost.com/submissions) or **HackTheCrisis Finland**'s [230 projects](https://app.hackjunction.com/projects/hack-the-crisis-finland) and see if you can build on some of the code and research that was already done there!
Many civic tech and tech4good organisations have already been busy setting up projects, tools and websites. Check out existing initiatives from the [Code for All](https://medium.com/u/ff8d09c2c0b4) network [here](https://docs.google.com/document/d/e/2PACX-1vTyatyMxWNslG0YacwTf381Dy089rqalSjqsZ6LAUe9qj12sYRQFizKCPpI1SpFzG-ey5ybVo7oqoxS/pub) and the Coronavirus Tech Handbook [here](https://coronavirustechhandbook.com/home).
Look at existing source code as well, thousands of developers have already gotten their hands on COVID-19 and many of them have shared their code on [Github](https://github.com/topics/covid-19?o=desc&s=stars). And don't forget to give back everything you write to the community!
## 2. Don't build yet another social platform
Is the success of your idea depending on the fact that a large number of people use it? Are you relying on user crowdsourcing for the content? In that case, think twice (thrice) before you move forward with it. The strength of such platforms usually resides in the scale effect and it's unlikely you will get it, especially if 20 similar initiatives see the light at the same time. A good example is the multitude of services appearing recently to connect people offering help and those needing it (nurses, elders…). Don't spend your weekend developing one more!
In France, the civic tech community came together to develop a [unique tool](https://web.archive.org/web/20211128113154/https://enpremiereligne.fr/) ([source code](https://github.com/tgalopin/enpremiereligne.fr)) and it has already been used by tens of thousands of people with a matching rate of almost 100% between helpseekers and givers.
## 3. Don't build a new mobile app
[Seriously, don't](https://medium.com/@dulitharw/dont-build-a-mobile-app-5077541e1415). Unless your goal is to sharpen your mobile development skills. But understand that it's unlikely anyone will be able to find it on an app store and bother installing it, especially among the most vulnerable communities. Instead, consider what type of channels they use in their everyday life in order to get information and help. It is likely not an app.
Try to think low tech or to reuse services and apps that already exist and might even be installed on people's devices. A good example is [dataprata.se](https://web.archive.org/web/20240102071445/https://www.dataprata.se/) the guide developed by Daresay to help people to get started with videocalls.
## 4. Don't create a new visualisation (before reading this article)
As data people, we like to make sleek and flashy visualisations. They can be an incredibly powerful way to convey information to people. But visualisations can also be a source of misinformation, especially when created and consumed in a state of emergency and information overload. So make sure to read this article listing [ten considerations before you create another chart about COVID-19](https://medium.com/nightingale/ten-considerations-before-you-create-another-chart-about-covid-19-27d3bd691be8) by [Amanda Makulec](https://medium.com/u/af3597c2cb6e).
## 5. Don't repeat discussion topics in the team
Facilitate discussion of varied topics by creating teams whose members have different life experiences, technical training and educational backgrounds. By drawing on a more eclectic set of topics you will likely design a better idea and solution.
## 6. Don't overestimate technology's capabilities to solve a really complex problem
Being coders, we sometimes get the impression that most problems in life could be solved with a few algorithms, a well-trained machine learning model and a chatbot.
, https://xkcd.com/1831/](xkcd-comic.webp "CC Randall Munroe, https://xkcd.com/1831/")
Be humble and be realistic about the problem you want to solve. In particular, the two following points can help you to find a use case where your solution can make an impact.
* **Don't solve a non-problem**
Spend the main part of your time on identifying the root cause of a real-life problem, rather than building a solution for a trivial or misunderstood problem.
* **Don't try to solve everyone's problem**
Focus on one aspect that is linked to one problem. Then choose the audience that will be helped by your solution — on individual, organizational or societal level. By breaking things down, your solution will be more concrete and useful than a hypothetical solution to everything.
## 7. Don't dig in someone else's backyard.
Use your own experiences and dig where you stand. Take the perspective of a parent, caregiver, child, family business owner or wherever you have real life experience. This will root your problematization in a reality that concerns people's everyday life that you have in-depth knowledge about.
One example of an insight brought forward in the media lately is that abused women and children will suffer from isolation together with perpetrators, that are either physically or virtually present in their homes. This perspective is brought forward by those who experience it, and point to an aspect of the crisis that to a large part has been overlooked, opening up a gap for contribution — what unique perspective can you bring to the table?
**OK, so let's say you have agreed on what project to proceed with — now what?**
## 8. Don't discuss everything together ALL the time
When we collaborate with people we've never met before, we run the risk of assuming that we have to discuss and agree about every detail. However, research on crowdsource innovation has shown that teams who alternate between periods of independent work followed by "communication bursts" of high activity perform better than teams that communicate in continuous streams of messages. Set discussion time!
## 9. Don't spend too much time building stuff
If you have a great idea, it is better to spend the time on developing the idea and focus on *telling a clear story of how it would be developed* into a tangible solution once you win. If your main competence is about building stuff, translate this into a competence of understanding how your solution would or could be designed,as well as the limitations to a technology that your peers might not be aware of.
## 10. Don't forget to have a good time
To make the most out of this hackathon make sure you keep expectations on yourself, your team and the event itself on a realistic (but committed) note. The reality is that the majority of the amazing ideas and prototypes will not continue after the event is over.
That being said, that doesn't make this weekend less important: No matter where you come from, we can assure you that this experience will bring you new friends, insights, perspectives and experiences that will make you grow an inch by monday. You won't regret it!
And most importantly — have fun hacking! Don't let this pandemic take away your happiness!
# 5 ideas for Sweden’s new open data portal
**THIS ARTICLE WAS ORIGINALLY RELEASED ON [MEDIUM](https://medium.com/civictechsweden/5-ideas-for-swedens-new-open-data-portal-d47beb65ec5f).**
A few weeks ago, I was in Sundsvall to discuss open data. I was invited by the newly formed open data team at DIGG (Sweden’s digitalisation agency) to help shape the new version of [oppnadata.se](http://oppnadata.se), Sweden’s open data portal.
The workshops went well and allowed us to express our needs and frustration as members of the small civic tech community in Sweden. I would even go as far as to express my hope as I saw a lot of new and motivated faces with what seems like a genuine interest in pushing for more openness in Sweden.
Not everyone seems to share my optimism. After seeing the government move open data around, from Vinnova to Riksarkivet and now DIGG, I’ve heard many wonder if this is actually the reboot they hope for. Time will tell if I was naive but I think most people will agree that Sweden can hardly lose three more years.
Unfortunately, I won’t be able to attend the last workshop. Civic Tech Sweden will be well represented by my friends Alessandro and Mattias but I thought I’d also put on paper some of the ideas that we got from these workshops so they can come to use, in Sweden or elsewhere. So here’s a list of five cool ideas for the future [oppnadata.se](http://oppnadata.se), with examples to illustrate! Read until the end, there’s a twist!
## User-friendly tools to manipulate the data
The Swedish open data portal is currently little more than a Wordpress website. It provides lists and a search engine to find the data you’re looking for but doesn’t help you further to visualise or process it.
Many countries and cities choose to lower the bar by providing interactive tools which make it simpler to display the data on a map or create basic graphs with it.
This is for example the case of the city of Helsingborg. Their portal, developed by OpenDataSoft, can display gorgeous maps with multiple datasets and let you share the visualisation you created or integrate it in another webpage.
These functions would be interesting in Sweden because the government has historically neglected one category of users even more than the others: citizens and civil society organisations. They are usually not data literate but provide a lot of value to the society in many ways that could benefit from data.
## Leveraging on Freedom of information requests (Offentlighetsprincipsbegäran)
When it comes to open data, there’s a mantra you hear all the time in Sweden:
> Government: We need to listen to people’s needs!
> Citizens: We’d like you to publish this dataset, can you do it?
> Government: No, sorry, we prefer to release what people actually need!
> Citizens: OK, here’s my need: this dataset. Can you please open it?
> Government: Later, we first want to focus on people’s needs!
> Citizens: …
The Swedish government has always been really selective in the needs it listens to. And even when they do listen, they can be really good at ignoring requests that they don’t wanna hear.
But theoretically, that makes sense, right? With limited resources, the government should prioritise the data to release.
What’s a **really good source of information** (the best, maybe) on which data people want?
> **Freedom of Information requests**
> (Offentlighetsprincipsbegäran in Swedish)
Every day, all public organisations receive requests for documents and information from citizens, journalists, businesses on all issues. We’re talking about hundreds of thousands of e-mails every year.
In the UK, all the requests that come to the central government and its agencies are analysed and they even publish [a quarterly report on it](https://www.gov.uk/government/statistics/freedom-of-information-statistics-january-to-march-2019). The USA have a website called [foia.gov](https://www.foia.gov) to help citizens send an FOI request which gathers valuable statistics. Both countries then use it to prioritise which datasets to release next.
In Sweden, Elenor Weijmar and Mattias Axell at Civic Tech Sweden deployed a platform called [handlingar.se](http://handlingar.se) (formerly Fråga staten) to do exactly that. You can use it to send a FOI request to any administration but the cool feature is that all the requests and their answer become public, enabling anyone to access the freed documents and any administration to keep track of what they get asked. There is even an API to automate the sending and fetching the data. Government bodies can also request documents and data from themselves and automate that process.
Now you maybe get what I’m getting at. What if the national open data portal enabled you with a few clicks and a simple form to request the data it doesn’t have to the relevant administration? What if the data was added to the list of available datasets as soon as the administration sends it back?
Better. What if DIGG used information about these requests and could send, once in a while, the following message to other agencies:
> Dear government agency, we see that you received XXX requests this year regarding xxxxxx.
>
> According to our estimates, your collaborators spent a total of YY hours answering these requests.
>
> At DIGG, we would like to encourage you to release this data in a proactive way so you and the citizens who sent these requests can save a lot of time and effort. We remain available if you need help to do so.
>
> Best,
>
> DIGG
Well, I’m not making anything up. In addition to the UK and the USA, several other countries have adopted this approach.
Canada is a good example. On their [open portal homepage](http://open.canada.ca), a link to a form to request information stands in the middle. It checks that the data isn’t already freed, helps people to find which institution probably has it and automatically publishes the data on the portal when the request is fulfilled. Since 2017, more than 30 000 documents have been released this way.
By saving thousands of hours to the civil servants answering these requests, developing this function on the future open data portal would easily pay for itself 1000 times.
But there isn’t actually much to develop. Tools like [Handlingar](http://handlingar.se) (based on the open source solution [Alaveteli](http://alaveteli.org)) could easily be interfaced with the future portal. At Civic Tech Sweden, we are always open to collaborate.
## Creating a community around the data
This is a path followed by many open data portals around the world but let’s take a Swedish example.
[Trafiklab](http://trafiklab.se) is one of the cool kids of open data in Sweden. Thanks to Elias and his team, data for collective transportation is easily available and used in many services that people use daily.
On their website, a key feature is the *Projekt* tab, which shows all the registered services using the APIs. In many cases, people link to a Github repository with the code they wrote. In some, to the app that they built.
This achieves two goals: it helps data reusers to see how others have used it and it gives precious information to the data providers.
This logic is pushed even further on the main French national open data portal [data.gouv.fr](http://data.gouv.fr)
Anyone can add a “community resource” which can be the same dataset in another format or one with corrected data.
People can also link their reuses in a similar way to Trafiklab. Finally, there are social features to comment, like and follow a dataset, and discuss with the data providers.
Etalab, the French agency in charge of open data and open innovation, developed its own portal called **udata** to build these functionalities. Like everything else they create, it is [open source](https://github.com/opendatateam/udata) and it is used by other countries like [Luxembourg](https://data.public.lu/en/) , [Portugal](https://dados.gov.pt/en/) or [Serbia](https://data.gov.rs/sr).
## One portal fits all?
DIGG has yet to build one good open data portal. But if we look a little further, few countries with a mature open data strategy satisfy themselves with a unique portal.
In France, Etalab opened 5 specialised portals for [addresses](https://adresse.data.gouv.fr), [transport](https://transport.data.gouv.fr), [geographic](https://geo.data.gouv.fr/fr/) and [corporate](https://entreprise.data.gouv.fr) data. And more are probably on their way.
In the Swedish context, where the effort has been considerably decentralised, that doesn’t mean that DIGG should start developing all these portals by itself. But it should try to coordinate the work done by Lantmäteriet for its [geodata portal](https://www.geodata.se/geodataportalen), Bolagsverket with its [APIs](https://bolagsverket.se/om/oss/api-1.11642) for corporate data and Trafiklab. Pushing them to open and reuse the same code would be a good start and would provide smaller agencies sitting on valuable data with tools to publish.
Moreover, DIGG should also aim at releasing and gathering data from private providers, positioning itself as a broker. In many countries, ridesharing services and free floating companies are forced to publish open data to prevent a lock-in and improve competition. I will come back to this point in another article.
## Any portal at all?
If you’ve managed to read this post until the end, I hope you’re hyped by all of these international examples and want Sweden to have the shiniest portal of all!
**But what if I ended this article by saying that the portal doesn’t matter that much?**
Let’s look at the UK. Uncontested European champion of open data since the term has been coined, its government digital services successfully managed to release a huge amount of datasets while creating a vibrant ecosystem around it.
Its national portal? Little more than a list with a basic search engine.
What does that tell us? That building an advanced portal is not necessary.
**More importantly, that it’s not enough!**
DIGG needs much more than a shiny new open data portal to catch up with the rest of the crowd.
It first needs to push for the release of more data (like way more). When the data is not used, it needs to go beyond the data literacy barrier and help the administration develop its own services to help citizens and organisations access it.
These two points are often linked. In Canada, in France, in the UK, an agency rarely releases a new service without making the underlying data open with an API.
Finally, it should also help to push open data in services that people already use.
Take these last two examples: In 2014, the Norwegian agency Kartverket released most of its geographical data as open data. But they didn’t stop there. The data was [subsequently added to OpenStreetMap](https://wiki.openstreetmap.org/wiki/No:Kartverket_import) thanks to the help of volunteers. In 2015, I remember preferring it to Google Maps because the latter didn’t have any of the ski tracks.
In Sweden, some of the national museums understood early the importance of putting their data where the public is. In 2016, Nationalmuseet [released](https://www.nationalmuseum.se/en/nationalmuseum-releases-3-000-images-on-wikimedia-commons) thousands of high-resolution artwork on Wikimedia Commons, allowing the material to be reused on Wikipedia, but also in ads or memes.
Most of the reusers might never learn that the original art came from the museum, but that is not what mattered to Karin Glasemann (digital coordinator at Nationalmuseet) and her team. What they wanted was to introduce their collections to a much larger public. In 2018, [these artworks were used in over 1800 Wikipedia articles in more than 50 languages, and seen about 18 million times](https://www.slideshare.net/ssuser37e15b/frn-open-access-till-ppet-museum). Not bad at a time where the museum itself was closed to the public.
As Karin Glasemann puts it, opening the data is only the first step. What matters is what you do after that: enabling reuse, creating partnerships and building communities.
Let’s hope that the future Swedish open data portal helps us to achieve all of this!
# Disrupting public procurement with open data in Sweden
In Sweden, public procurement was estimated to be worth a staggering **683 billion crowns** in 2017[^fn:1]. Over 18000 tenders are launched each year by public institutions.
{.dropcap}
Public procurement is a crucial part of a working government and a fair and effective process has positive consequences on the whole society, freeing needed resources and leaving less room for corruption.
It also faces constant challenges due to the ever-changing nature of the services and goods that public institutions need to buy and to the constant trend of privatisation and subcontracting that most governments have been going through for the past 40 years.
---
In this article, we're going to look at the **state of public procurement in Sweden**, list some of the problems it's facing and compare it to other countries from the perspective of openness. Finally, we'll introduce a project we are starting at Civic Tech Sweden to address these issues and raise awareness among the relevant authorities. Ready?
## Data all over the place
Let's start with an obvious statement: **public procurement is a data-intensive process**. From tender ads to signed contracts and invoices, the whole chain is documented with care, ideally in a standardised digital form.
Almost all of this data is public sector information and as such, it can usually be obtained sending a FOIA request to the corresponding institution. To make the process smoother, the law sometimes requires to publish part of this information proactively. It bears a critical importance for all the parties involved:
* **journalists**, who make FOIA requests regularly, need this data to investigate corruption and wrongdoing, leading to a healthier public sector;
* **private companies** need to be informed about tenders in an easy and timely way to be able to bid. They also need as much insight as possible into past tenders to be able to optimise their offer in terms of price and service;
* **public institutions** need as many companies as possible responding to their tenders to foster fair competition, leading to lower prices and a higher service level. They also need analytics on how they are performing with their purchasing. A good way to do so is to compare themselves to institutions of similar nature and size;
* **citizens** benefit from all of this by getting better services for a lower price.
## The current situation in Sweden 🇸🇪
In Sweden, a continuous effort has been made to digitalise the whole process, the latest episode being the obligation for public institutions to be able to receive electronic invoices[^fn:2] (with the european standard PEPPOL BIS Billing 3[^fn:3]). Apart from a few exceptions, digital data is now the norm all along this chain, and **Sweden could even be considered one of the most advanced countries in that regard**.
One might think that data is used as much as it could be to provide the aforementioned benefits and match offer and demand in the best of all worlds. As our first investigations revealed, this is absolutely not the case and that is exactly why we are setting up our project.
Right now, the situation is along these lines:
* **journalists** have to send FOIA requests to access the documents they need. In practice, it’s a tedious and unnecessarily complex process, the documents are rarely provided in a standardised format, and it costs both the demander and the responder some unneeded time. The best case scenario here is e-mails and PDFs, Excel spreadsheets at best. Sometimes, it’s a lot of paper and remember that the Swedish FOI act allows public entities to shamelessly charge 2kr per page without justifying of a related cost…
* **private companies** have to pay for a private service to see and get alerts about the tenders that interest them. The market for these services (Upphandlingsmyndigheten provides a list) is small, almost a monopoly due to the lock-in of the data by the providers of the procurement software used by the institutions. Visma, the biggest of them, keeps a dominant position by circulating the data between its own services. It’s hard to measure how much **this cost deters small and medium companies from even considering public sector institutions as a potential buyer** but one could hardly deny that it probably reduces the competition.
* **public institutions** get really little insight in their buying behaviour, much less than we assumed in the first place. Some of it is sold by the providers of procurement software (Visma benefiting from its dominant position to provide better analytics) but in general, municipalities and government agencies have to resort to time-consuming audits and the smallest organisations do not have the resources for anything.
This is not to deny the fact that the procurement system in Sweden might be one of the most effective and least corrupted in the world. But the above should give you an idea of the **huge potential for improvement**.
## Do they do it differently elsewhere?
Spoiler: yes, they do.
In developed countries, a free public tender ad database is usually considered an essential piece of a fair and effective procurement system. According to research by Digiwhist and Open Knowledge Germany[^fn:4], almost all European countries (31 of 34) had one in 2017 (EDIT in 2022: Sweden is now the only one who doesn't have one). In Scandinavia, these services have cute names such as [udbud.dk](https://udbud.dk) in Denmark, [doffin.no](https://doffin.no) in Norway, [hilma.fi](https://www.hankintailmoitukset.fi/sv/) in Finland.
It’s plain logic: the more you restrict access to information about the tenders, the less competition you’ll have. A paywall is probably the most effective way to deter companies from trying to address your need as a government.
---
Many countries and municipalities have also started publishing open data for other steps of the process to improve transparency or efficiency.
The most famous case is Ukraine, where the platform **ProZorro** was created by the NGO Transparency International to monitor, streamline and open procurement, with great success[^fn:5].
Maybe closer to Sweden, countries like Canada[^fn:6], the UK[^fn:7] and France[^fn:8], have taken on the task of publishing as much data as possible since the beginning of the 2010s.
Finland has gone furthest in the Nordics and publishes all the invoices of its government agencies as open data and on an interactive platform called [granskaupphandlingar.fi](https://granskaupphandlingar.fi). Earlier this year, the decision was also taken to add the documents of all local institutions.
## Fighting against the status quo
Why is Sweden hardly doing anything in the area?
Hard to tell but the situation doesn’t seem to improve. Recently, the government gave Upphandlingsmyndigheten the responsibility to gather statistical data about procurement and not once was the possibility of opening the raw data mentioned, even though their report was mentioning Finland as an example to follow[^fn:9]...
When asking an executive of the agency about it this Summer, we got this unfortunate answer:
> We have the offentlighetsprincip and this is something that very few countries in the world have. We should already be satisfied with it.
~~Incompetence~~ Ignorance is bliss.
---
In a country convinced of its superiority in a certain number of fields like transparency, the misconception that Sweden’s FOI act would be rare (most democracies in the world have one) or even superior (many countries have modernised theirs to adapt to the data era and take advantage of the benefits of open data, unlike Sweden) is not uncommon.
This explains that after asking all the relevant government agencies (Upphandlingsmyndigheten, Konkurrensverket, DIGG, SKL and its purchase company Kommentus), not a single one could point to an internal contact person working on opening this data…
A quick review of the local administrations found that only 3 municipalities have taken the initiative to open their invoices ([Göteborg](https://catalog.goteborg.se/catalog/6/datasets/75), [Örebro](https://www.orebro.se/fordjupning/fordjupning/fakta-statistik-priser--utmarkelser/information-tillganglig-for-ateranvandning/inkomna-leverantorsfakturor-reskontra--kontoklasser.html) and [Lidingö](https://lidingo.dataplattform.se/#/data/leverantorsfakturor)). This is always the result of pioneers who saw early the potential of releasing them, such as Kim Lantto or Björn Hagström who already wrote about it on his city blog in 2015[^fn:10]. These initiatives should be praised but they hardly show a positive trend and they can’t benefit from the potential of doing this on a bigger scale.
At this pace, one might expect SKL to catch up in 2050 and all the municipalities to release the data by 2070. This is not satisfactory!
## Our project
Civic Tech Sweden has thus decided to start something on this topic with the hope of:
* raising awareness!
* assessing what the possibilities are for opening this data and what some of the blocking points could be;
* writing guides and setting standards to help the public actors to release the data in a painless and effective way;
* prototyping a selection of services around the data to showcase the possibilities;
* eventually building an ecosystem for innovation around procurement data that benefits all involved actors and saves public money!
The project is currently called *Öppen Upphandling* for lack of a better name, feel free to suggest a catchier one!
Sounds exciting? Well what are you waiting for? Our goal is to get something started and everyone is welcome to join our efforts or build something on top of what we do, as long as you embrace the principles of open innovation and open data!
If you are a public institution, a private company or a journalist and you are interested in taking part in this project in one way or another, we are setting up a working group and are really interested to hear about your needs!
Contact us by [email](pierre@mesu.re) or directly on our [Matrix.org chat](https://app.element.io/#/room/#civictechsweden:matrix.org).
---
**UPDATE**: We received funding from Vinnova to run this project with Open Knowledge Sweden, DIGG and partners like FGJ and Transparency International Ukraine. The project is now called OpenUp! and has [its own website](https://openup.okfn.se).
**UPDATE 2025**: Unfortunately, our project has not led to the major changes we hoped for. I (Pierre) am still fighting in every possible way for open procurement data and you are more than welcome to contact me if you want to talk or collaborate on it. 🙂
## References
[^fn:1]: [Statistics from Konkurrensverket's 2018 report on public procurement (konkurrensverket.se)](https://www.konkurrensverket.se/globalassets/publikationer/rapporter/rapport_2018-9_statistikrapport_2018_webb.pdf)
[^fn:2]: [DIGG on compulsory e-invoice (Archived, digg.se)](https://web.archive.org/web/20200103195603/https://www.digg.se/nationella-digitala-tjanster/e-handel-och-e-faktura/obligatorisk-e-faktura-till-offentlig-sektor)
[^fn:3]: [SFTI on PEPPOL (sfti.se)](https://sfti.se/sfti/standarder/peppolbisehandel/peppolbisbilling3.49021.html)
[^fn:4]: [Recommendations for the Implementation of Open Public Procurement Data (Archived, opentender.eu)](https://web.archive.org/web/20220331045834/https://opentender.eu/blog/2017-03-recommendations-for-implementation/)
[^fn:5]: [From the fires of revolution, Ukraine is reinventing government (wired.com)](https://www.wired.com/story/ukraine-revolution-government-procurement/)
[^fn:6]: [Jaimie Boyd on transparency in Canadian procurement (medium.com)](https://jaimieboyd.medium.com/canadas-open-by-default-procurement-pilot-an-experiment-in-agility-e10c9acd5806)
[^fn:7]: [Improving and opening up procurement and contract data (gds.blog.gov.uk)](https://gds.blog.gov.uk/2015/11/19/improving-and-opening-up-procurement-and-contract-data/)
[^fn:8]: [Open Data Laws in France Increase Competition for Public Contracts (ogpstories.org)](https://www.ogpstories.org/open-data-laws-in-france-increase-competition-for-public-contracts/)
[^fn:9]: [Förberedande studie om vissa inköpsvärden (upphandlingsmyndigheten.se)](https://www.upphandlingsmyndigheten.se/kunskapsbank-for-offentliga-affarer/publikationer/slutrapport-vissa-inkopsvarden/)
[^fn:10]: [From paper archive to open data (Archived and in Swedish, blogg.orebro.se)](https://web.archive.org/web/20231206053715/https://blogg.orebro.se/enklarevardag/2015/09/10/fran-pappersarkiv-till-oppna-data/)
# Din Riksdag
{{< icon-button "github.com/DinRiksdag/dinriksdag" "github" >}} Source code (platform) {{< /icon-button >}}
{{< icon-button "github.com/DinRiksdag/decidim-module-riksdagen" "github" >}} Source code (scraping module) {{< /icon-button >}}
{{< icon-button "github.com/PierreMesure/pierre.mesu.re/tree/master/static/dinriksdag/presentations" "link" >}} Presentations {{< /icon-button >}}
## Learning that Swedish democracy is far from perfect
I came to Sweden in 2016 and as long as I can remember, I have always been fascinated by its democracy. During my first years in Stockholm, I often visited the parliament, sitting above the main chamber and listening to the debates. Swedish parliamentary democracy had something I didn't grow up with: a proportional voting system, high levels of trust, less polarisation, and a great deal of consensus and respect for institutions and processes.
In my home country, distrust and anger were at record highs. But because of that, France was also bubbling with an enormous drive to hold politicians under scrutiny and accountable, and many citizens had a strong will to get involved in the decisions that affected them. To meet this demand, countless innovative projects were started in the 2010s, and institutions were reformed (slightly) to allow these new forms of democracy to have an impact.
When I arrived, I regarded Sweden as the pinnacle of democracy and was certain none of this was new here. I was wrong. Perhaps because of this high trust, information on elected representatives' private interests is not available as open data. Until recently, it had to be requested and collected physically for a hefty fee. Sweden is the only democracy around the Baltic Sea that doesn't give citizens the possibility to write petitions to start debates in its parliament. The official process to give feedback on future legislation (called *remissprocessen*) is tailored to corporate interests and organised civil society but leaves very little room for grass-roots engagement and significantly lacks transparency.
## Building Din Riksdag, one workshop at a time
Naively, but with a genuine drive to improve things, I started a project called **Din Riksdag**. Din Riksdag's goal was to become a platform where citizens and grass-roots groups could get a better understanding of the legislative process and have an impact on it.
I quickly realised how little interest there was from institutions and political forces to reform the current system, so I decided to build something that would "hack" it rather than await change from within. I drafted two potential mechanisms:
- ***citizen referral responses*** (*medborgarremissvar*) to be sent to the government before they start drafting the bill.
- ***citizen motions*** to be sent when the bill is discussed in parliament.
Both could be crafted collaboratively and gather support through votes, giving increased legitimacy to popular proposals and making it easy for both the Government Offices and parliament committees to include them in the lawmaking process when creating the bill (*proposition*) and the committee report (*betänkande*), respectively.
These ideas came from personal research but also through feedback and ideas from various workshops I conducted throughout 2017 and 2018.
## Swedish civic tech?
To support this new process, I also set up a digital platform where the entire legislative process was replicated and interactive, a website where citizens could browse through public inquiries (*utredningar*), referrals (*remisser*), bills (*propositioner*), and parliament activity, as well as submit or support citizen feedback and motions. I used the leading civic tech tools of the time: first Consul, then Decidim, and developed modules to integrate Swedish legislative data.
Unfortunately, despite the parliamentary administration's best efforts, a lot of that data remains unavailable in a structured format. That is why I started projects like [OpenRemiss](../openremiss) and ultimately [g0v.se](../g0vse) to make it easier to reuse data from regeringen.se.
## Turning the page
Although I never officially ended Din Riksdag, I stopped investing my time in it around mid-2018. At that time, I met the founders of Digidem Lab (Sanna, Anna, and Petter) and decided to join them to apply my newly acquired skills to other participatory projects in Sweden. Together, we used Decidim in many municipalities here and abroad and introduced methods such as participatory budgeting and citizens' assemblies.
My hope was that if I gave the national institutions a few years, they would mature and become more interested in letting citizens have their say. As I said earlier, I was very naive.
# OpenRemiss
{{< icon-button "github.com/DinRiksdag/OpenRemiss" "github" >}} Source code {{< /icon-button >}}
## What is a "remiss"?
The referral process (*remiss*) is a critical step in the Swedish legislative process. Before the Government Offices draft a formal bill (*proposition*) for Parliament, the public inquiries (*utredningar*) used to support the work are released and opened for a public round of feedback.
This process is sometimes criticised for being slow or because consulted public bodies do not always speak clearly or independently. Furthermore, the government has been accused of ignoring feedback or skipping the referral stage altogether. In my view, the primary issues are a lack of transparency and that the process is largely adapted for lobbying by corporate interests and organised civil society, often overlooking the voices and opinions of a large segment of the population, particularly marginalised groups.
To open up this process, I created *Din Riksdag* and designed a system for collaboratively written petitions called *medborgarremissvar* (citizen feedback). However, I needed access to all existing referral responses (*remissvar*), which led to the creation of OpenRemiss.
## Just PDFs on a plain website
Despite many requests from other government bodies, journalists, and researchers, the Government Offices (*Regeringskansliet*) refuse to provide their publications as open data. When it comes to the referral process, the essential data is structured as follows:
- The list of referral processes must be scraped from HTML pages, alongside metadata about the related public inquiries (*utredningar*).
- The list of files for each specific process also requires HTML scraping.
- The list of organisations invited to provide feedback (*remisslista*) is published as a PDF file. Furthermore, every department uses its own template for this list, so standardised extraction is not straightforward. Organisations are often listed only by name, with a surprising number of typos and variations.
- The referral responses (*remissvar*) themselves are also PDFs, the format requested by the government. Each organisation uses its own template, making content extraction difficult without losing formatting or context.
- Feedback submitted by organisations not included in the original *remisslista* is not even published. These must be requested under the Freedom of Information Act and are often provided as physical paper for a fee.
- Similarly, the government's own analysis of the feedback (*remissammanställning*) is not public, though a brief summary is usually included at the end of the subsequent bill.
**OpenRemiss** is a script designed to automatically download this information (except for the last two items) and convert it into structured formats suitable for analysis, research, and innovative services. It downloads lists of referral processes with their metadata, fetches the files for each process, and converts *remisslistor* into structured lists while cleaning up organisation names.
Interestingly, this project led to my current role at the data lab of the [Swedish Agency for Financial and Public Management](https://www.statskontoret.se) (*Statskontoret*). I connected with them while they were building *Hitta remissvar* ("Find referral responses") and a similar scraper to collect the necessary data. The interface built before my time there clearly demonstrates the potential of open, structured referral data, as do subsequent projects we have done with researchers and the Government Offices.
Since then, I have also built [g0vse](../g0vse), a scraper that can fetch any information from regeringen.se, though it does not handle *remisslistor* or name cleaning.
Don't hesitate to reach out if you're interested in using the code or data from this project! I’m happy to help you get started and to collaborate on making this data more accessible.
# About me
{style="width:50%;"}
Currently in charge of the **"datalabb"** at a government agency under the Swedish finance department.
I have been working in Sweden since 2016, where I have led numerous projects in the fields of **civic tech, transparency through open data, and citizen participation**.
For 6 years, I co-directed [Digidem Lab](https://digidemlab.org), a nonprofit citizen participation cooperative and a pioneering organisation in orchestrating Sweden's first citizens' assembly and many other initiatives (participatory budgets...).
In parallel, I have worked to improve access to public interest data in Sweden, notably in areas like public procurement or lobbying in legislation-making. Through the [Civic Tech Sweden](https://civictech.se) network, I organised many civic tech meetups in 2018-2021.
This led me to become a civil servant in October 2023 to continue the fight from within.
# Privacy
This website is hosted on [statichost.eu](https://www.statichost.eu), a service founded by Eric Selin in Sweden with a deep commitment to privacy and digital sovereignty. Therefore, no personal data is processed nor stored when visiting and the website is hosted in Europe. 🇪🇺
In order to get basic statistics on the number of visitors, I am using [GoatCounter](https://nlnet.nl/project/GoatCounter/), a service funded by the NLNet foundation that also guarantees that none of the collected data can be used to identify a visitor. This is why I didn't add a cookie banner to the website.