Inherent Politics in Public Data
By Benjamin Hervit
This blog post is an extract from a paper I wrote on the Financial Act Database after I read about Jesper’s experience. In the paper, I try tried to answer:
“How are politics performed through the Finance Act Database affecting the Danish citizens use of the database?”.
The network surrounding the Financial Act Database is dominated by ministries and agencies that is concerned with supporting the government in economic politics and create a more efficient public sector. The result of this is, that the Financial Act Database is it near to impossible unless you have a degree in law, politics, finance or programming. This raises the question if the database then can be defined as public and what other options there is.
Jesper Henrichsen, a software developer student at the IT University of Copenhagen (Boye, 2017) made a visualisation of the Danish Finance Act for 2017 by using data from the official online Finance Act Database http://www.oes-cs.dk/olapdatabase/finanslov/index.cgi. His visualisation can be seen here. http://188.8.131.52:10000/ . Long story short, is it not without complications to collect all the data from the Finance Act Database, which Version2’s article about Jesper’s experience tells way better than I can: https://www.version2.dk/artikel/itu-studerende-scrapede-aabne-data-med-python-script-fik-sin-ip-adresse-blokeret-1071184.
Figure 1: A compressed example of Jesper Henrichsen’s Visualisation of the Finance Act Database
So, the first question that got to me after reading the article was: Is this an issue? There are other ways to get information about the Finance Act Database such as PDF’s, electronic books, contacting them and asking politely, etc. Who can even benefit from this database besides those who are working with it? Data, that being big, small, thick, smart, structured, unstructured, apache helicopter or however you or the data want to identify – is one of the big buzzez at the moment. The government wants to share data and encourage public debates of Denmark’s future (The Danish Government, u.d.), and AMMF invites everyone who is interested to use the Finance Act Database and many other of the state’s corporate systems (The Agency for Modernisation Ministry of Finance, 2017) . Denmark is a part of the Open Governmental Partnerships, a commitment to, among other things, to create transparency, democratisation and use digital technology to improve society (Digitalisation Agency, 2016). There does not seem to be any actors or arguments that is against democratising more public data from the Danish government and public sector. But the case with Jesper Henrichsen shows how sharing and democratizing data takes more than simply sharing a link.
The purpose was to look at the network that surrounded the Finance Act Database and, maybe through this being able to answer the question. For the sake of your patience and the length of this post will you only get one part of the discussion and the methodology that was used for this part. This part of the discussion is concerned with the inner actor-network of Finance Act Database and how they are affecting the database.
I don’t think there is any reason for me to throw more Harraway at you. If my introduction and this article being posted on ETHOSlab’s MetaData blog havsn’t situated me enough… I could mention that I rarely engage myself with politics – so … that’s something.
Methodology and Theory
For my analysis, I mainly used Winner’s concepts from “Do Artefacts Have Politics?” (1980) and Latour’s Actor-Network Theory (1988) as my methodological approach. From Winner, I used the inner social system to identify how its actors could be reflected in the Finance Act Database, its design and the politics it performs. Also, the question of an artefacts political flexibility and whether the performed politics are intended or unintended was used in the discussion.
I used ANT with the purpose of identifying how tasks are delegated between the actors in the Financial Act Database’s network. What prescriptions does it take to be a part of this network and what transcriptions may be hidden in the embodied knowledge of the actors?
My empirical data started with Version2’s article about Jesper, investigating the Finance Act Database and visiting the websites of AMMF and the Ministry of Finance. Afterwards would my empirical data grow as I started to follow the actor and unravel the network.
Since the database is an artefact of the internet I decided to stay in the digital sphere of websites and let them play their role as non-human actors, representing companies, institutions, civilians, etc. I mainly looked for political indications and relations expressed by the actors on their “About Us” pages which I treated as the actors’ descriptions. My reasoning was that another actor who are mentioned directly by name on an “About Us” page, must be important to the author of the webpage and therefor worth following when trying to unravel the politics in the network. By keep asking questions to these descriptions of what they contain and what they are missing, I tried to build an understanding of what politics are enacted through the delegations and references between actors and seeing the transcriptions that are hidden in the network. I tried to keep my information to come from here, but when I found that more information was necessary I went further than just the About Us pages.
I am aware that it is not a complete solution and many other actors, hidden between the lines, on other pages or through transcriptions, was left out in this process, but I found it necessary to make come constraints because of the complexity of the actors in the network.
The Finance Act Database’s description
The Finance Act Database is accessed through oes-cs.dk which is where the AMMF holds their corporate systems (The Agency for Modernisation Ministry of Finance, 2017). The analysis will only focus on the web page that is the F Act Database.
Figure 2: Structuring a report in the Finance Act Database
The Finance Act Database declares its ownership already in the introduction in its manual as it is described as “Finansministeriets Finanslovsdatabase” which translates into “The Ministry of Finance’s Finance Act Database” (The Finance Act Database, u.d.). It informs that the user can define its own tables that contains “[…] withdrawal of funds from the State Budget and Supplementary Appropriations Act for one fiscal year.” (ibid). There is a short introduction to what the finance act is and some of the primary actors involved in making the finance act: the Danish parliament, the finance minister, on behalf of the government, and The Danish parliament’s Finance Committee (ibid). It is not possible for to search by words or numbers in the database. It can only be done by structuring and filtering variables or exporting standard reports (ibid). The description also informs that the database consists of eleven variables regarding appropriation, paragraphs, main area, activity area, main account, sub-account, standard account, top frame, sub-frame, type appropriation and expenses and revenue budget (Appendix 5) (The Finance Act Database, u.d.). These variables can be used for both structuring and filtering the data in various ways to deliver different reports (ibid). Figure 1 is an example of structuring given in the Finance Act Database manual. This structure will give a CSV-file with 119 rows and 7 columns, but only for the main areas. Individual CSV-files will have to be downloaded for each sub-area. The complete dataset which Jesper Henrichsen collected for the Finance Act 2017 was about 9500 rows and 14 columns.
If any my observations on how the Finance Act Database should be flawed or misunderstood, I will argue that it only support my conclusion in the end.
Unreasonably simplified analysis
Figure 3: The actor-network surrounding the Finance Act Database. Node size is in-degree; the grey arrows are when one actor have directly referenced the other; red arrows are those who are linking directly to a database with the Finance Act data; blue nodes are governmental actors; green nodes are non-governmental actors; yellow nodes are referenced actors that has not been analysed; red nodes are databases that contain the Finance Act data. Tools: Gephi, Python, NetworkX library.
I promised to spare you for the analysis so all I will present here is the network I created.
In this network, I also tried to find new actors which wasn’t referenced in the original network itself, by e.g. using Google queries. This network is only for the curious as I will only give you one half of the discussion: the network of direct references by starting at the Finance Act Database a move outwards.
Finance Act Database and its inherent politics
I identified the Ministry of Finance and AMMF as the inner social system of the Finance Act Database. I chose not include the government and the Danish parliament as a part of this though many of their politics are passed down to the Ministry of Finance and AMMF. My reason for this was that neither the government nor the parliament is social systems that had to be created and maintained (Winner, 1980) to make the database. The Ministry of Finance has been given those tasks which then have been delegated to AMMF (The Agency for Modernisation Ministry of Finance, 2014). The actors for this part of the discussion have been visualised below. The visualisation below includes the inner social system, but also the actors affecting the inner social system: The government, the Danish parliament, the public sector and the Danish citizens.
Figure 4: The most prominent actors for the inner social system of the Finance Act Database. The size of the nodes is defined by in-degree. The red node is the Finance Act Database; the blue nodes are governmental actors that have been analysed; yellow nodes are actors which have been mentioned but not analysed in this paper. The grey arrows indicate when one actor has referenced another directly; red arrows are created when an actor is linking directly to the Finance Act Database.
The Ministry of Finance and the AMMF identify themselves as analytical powercenters where “[…] leaders of tomorrow” can develop and realise their full potential (The Agency for Modernisation Ministry of Finance, 2014). I will argue this is an expression for centralised politics and a self-perception of being the expert in a complex sector. Both actors leave no doubt that their primary goals are to help the government and the public sector. This could indicate that the government and the Danish parliament are delegating tasks to the Ministry of Finance and AMMF which then enforces a set of prescriptions on those who work in the Ministry of Finance and AMMF. These tasks may be centralised and highly concerned with the inner social system of the Danish government, rather than the inner social system of Denmark. This is emphasised by no or few mentions of actors that goes beyond the public sector of Denmark.
Some of these politics can be seen in the Finance Act Database. It proposes that the user understands the financial budget account systems prior to using the database and an above average understanding of the hierarchal structure of the Danish Finance Act Database. When looking for actors it mentions big political actors in Denmark; The Danish parliament and the Ministry of Finance. The actors are the same that the actors of the inner system of the Finance Act Database state their interest in. The database acting as a supporting role for the actor shown in the figure 3, where it is isolated from the Danish citizens by the other actors. This is not an isolation that is withholding citizens to enter the database, but an expression of the interests and politics in the inner social system having other objectives, which is why no one is considering this interaction. The visualisation figure 2 also shows that most of the actors that has been included in this analysis are pointing towards the government, again emphasising a gap between the citizens and the database.
The transcribed knowledge in the network.
When looking at the network in general are the same patterns emerging.
The Danish citizens are free to enter the database through several websites, but it is, understandably, not something that pops up on the front page – except when searching very specific on Google. None of the actors which is linking to the database have been given the responsibility to prepare the citizen users, most likely because the database initially was designed for internal governmental purposes. The most obvious path then, to be the skilled user the database needs is to study law, political science, finance or alike. The knowledge from these studies could be so embodied in the actors in the network that no one no longer are aware of other possible use-purposes of the database.
I identified the inherent politics of the Finance Act Database. These inherent politics have been identified as centralised, authoritarian and excluding towards the common Danish citizens and comes from the inner social system of the database: The Ministry of Finance and AMMF. Also, I could not find a governmental actor that has the Danish citizens in mind when it comes to the Finance Act Database. The search design and even the manual for the database prescribe some level of knowledge that cannot be expected by the common Danish citizen. I do not believe these politics are intended but instead a reflection of the embedded knowledge in the actors and therefor their expectations to how the Finance Act Database should function.
Can this be altered?
Does the nature of the database make it impossible to perform other politics than those that have been presented so far? My own conclusion is that unnecessarily centralised and excluding politics are being performed both through the Finance Act Database and its network. These politics can and should be changed to improve transparency and democratize user-friendliness.
Jesper Henrichsen’s Github project may not be the perfect alternative, but it does show a more democratizing approach. The complete dataset can be downloaded without having to build a scraper and merge thousands of rows from two different CSV-files. And the reactions on the article on Version2 suggests that the visual representation is fostering curiosity, questions and discussions about Danish politics (Boye, 2017) – just like the government itself invites to on its website (The Danish Government, u.d.).
The nature of the database does not necessarily enact authoritarian and centralised politics but as the case has shown it does not naturally perform democratic politics either.
Winner writes technology is “[…] influenc[ing] how people are going to work, communicate, travel, consume, and so forth… “ (Winner, 1980, s. 28). Therefore, the government must consider carefully how they wish technology such as public data should be implemented into the Danish society so that it performs democratic politics rather than authoritarian politics. They must do this sooner than they may think themselves, before the original flexibility of the database’s politics vanishes into social habits and economic investments (Winner, 1980, s. 29)
In this paper, I tried to answer the following question:
“How are the politics performed through the Finance Act Database affecting the Danish citizens use of the database?”.
To answer this question, I used Latour’s actor-network theory to identify how actors are referencing each other and linking to the database and Winner’s theory of inherent politics in artefacts. I have argued that the Finance Act Database is performing authoritarian and excluding politics towards the Danish citizens. It has been built as a reflection of its own inner social system and is surrounded by a network of actors that is indirectly acting in the best interests of the Danish citizens, but neglects to prepare them for interacting with a public database that seems built for internal use and highly knowledgeable users within law, finance and politics. I do not believe it is intended. It is the job of the Ministry of Finance to support the government with improving efficiency, economic decisions and politics. This may be a part of their own inner social system and the database, to them, is a way to succeed in this. I argue that the politics of the database is in fact flexible and that it is possible to create a more democratic alternative, which also includes some of the authoritarian politics to secure the data itself from becoming corrupted. If we wish to make this change, the government must act before the inherent politics of the Financial Act Database is being fixed in our social habits.
It is unlikely that all Danish citizens have the same interest in raw data as Jesper and the people in labs such as ETHOS Lab. But there are plenty of students, researchers and enthusiasts whose curiosity may be set in motion simply through accessibility.