The European Commission has launched an open database where the big platforms must publish their reasons for moderating content. The Digital Services Act (DSA) stipulates that providers of hosting services must inform their users of the content moderation decisions they make and explain the reasons behind those decisions. The DSA Transparency Database is a collection of all the moderation decisions, presented in a publicly accessible and machine-readable format.
What does the data show?
The database collects data from 16 different platforms: AliExpress, Amazon, Apple AppStore, Booking.com, Facebook, Google Maps, Google Play, Google Shopping, Instagram, LinkedIn, Pinterest, Snapchat, TikTok, X and YouTube. Users can search through a series of criteria such as the type of restriction imposed but the most interesting are the categories for grounds for removal. The data can be filtered by several fields, e.g. the type of restriction(s) imposed, categories such as animal welfare, IP infringements, and keywords, or the type or language of the content.
A quick search shows us that in the month of September, so far, Facebook has taken 474.751 actions. These include 12794 (2.6%) cases referring to violence and only 153 (0.03%) for "negative effects on civic discourse or elections". The biggest reason for taking action was data protection and privacy violations which accounted for 127131 (26%) actions.
TikTok took 3,645.332 actions, these include 12794 (0.3%) cases referring to violence, 5065 (0.1%) for "negative effects on civic discourse or elections" and 1,663.763 (45%) cases of "illegal or harmful speech".
X had taken 23087 actions in the same timeframe; 23086 for "protection of minors" and 1 for pornography. Something is obviously not right with X's reporting; its overall number of statements of reasons is much smaller than comparative platforms and it obviously doesn't take the quality of the reporting seriously.
Limitations
Whereas I fully welcome the launch of the DSA transparency database, it has some limitations. First, technical limitations only allow users to export the first 1000 statements of reasons at a given time. This severely restricts the amount of data that can be used for research purposes and I hope that this will be addressed soon. 10,200.82 statements of reasons are submitted every hour so the export limit of 1000 is ridiculous.
Secondly, the quality of reporting will have to improve. This might be the result of platforms getting used to this mechanism but the data submitted X for example is inadequate. I wonder what Thierry Bretton will have to say about this...
Time will tell if this database becomes more useful - the European Commission might have to flex its muscle to require better reporting, but it should also make the data more useable for researchers.