“It's impossible to have a meaningful debate if you've not got an idea about what's being de-listed.”
— David Jordan, head of editorial policy, BBC
In May 2014, the European Court of Justice made a historic decision that caused many people to think differently about the internet and who we are online.
Mario Costeja González, a Spaniard, had long-since paid back his debts from the 90s, but two short newspaper notices from that period about the auction for his foreclosed home still showed up prominently when someone searched for his name online. He challenged Google to remove the respective articles from its search results — and won.
Subsequently, Google, and all other search engines, had to provide a mechanism by which people could request removal of a specific search result if the content it linked to is “inadequate, irrelevant or excessive”.
Critics argued that shady politicians and criminals were likely to abuse the ruling to white-wash their online presence, but actual data on the nature of requests was missing. Despite calls from numerous academics for more transparency, Google did not share statistics on the requests it received. Google is taking decisions that are publicly relevant. As such, it is becoming almost like a court or government, but without the fundamental checks on its power.
The numbers graphed below come from Google's own Transparency Report. The source code of the website actually contained more data than the report visualised in its front-end. About three months ago, the extra data disappeared but snapshots on WayBack Machine, the ‘Archive of the Internet’, are a testimony to its prior existence.
Obviously, the number of delinking requests sent to Google increases with the number of citizens in a country. For example, Germany and France, the most populous European countries, contribute the largest number of requests. However, this is less interesting than the question:
What proportion of a country's population has sent a request to Google?One would expect this to be similar across countries. That is not what we see. Normalising the request rate by the population size of the associated country shows a high request rate in smaller countries, in particular Estonia.
At this point, it's important to note that the relationship between person making the request and the associated country is not neccessarily citizenship.
According to a letter from Google to the Article 29 Working Party:
“The data subject will need some connection to that country, which will normally, but not always, mean that he or she is a resident in it.”Some food for thought, maybe.
Numbers accurate as of March 23rd 2015
The source code underlying Google's Transparency Report contained a breakdown of requests into five issue categories: ‘private_personal_info’, ‘serious_crime’, ‘political’, ‘public_figure’ and ‘cp’. These labels came with no explanation as to their meaning. The key insight is that Google has labeled 95% of requests made under the Right to be Forgotten as ‘private personal info’. This suggests that these are mainly issues that you and your neighbour could have, not a criminal or a public figure. Furthermore, this is in sharp contrast to the examples of requests that Google shows on its Transparency Report website.
Although requests concerning private information dominate the statistics throughout Europe, there are some curious differences between countries. Italy's proportion of ‘serious crime‘ requests, for example, is significantly larger than any other country's. This country-to-country variation has many potential origins. For example:
These are interesting questions, which I hope local journalists can help to uncover.
Numbers accurate as of March 23rd 2015
So, what about the concerns raised by the critics of the Right to be Forgotten — is it giving criminals and shady politicans a shortcut to clean up their online curriculum vitae? We believe that it is probably not. While around half of request labeled by Google as ‘private_personal_info’ are granted, the great majority of requests in the four other categories are rejected.
While 1,892 request have been granted in connection with crime and politics, the Right to be Forgotten led to nearly 100,000 successful requests for ‘private_personal_info’ to be removed. These are, as far as we know, just normal people concerned with their digital privacy.
Paul Bernal, technology law lecturer at UEA Law School, argues that this shows that the Right to be Forgotten is a legitimate piece of law. “If most of the requests are private and personal ones, then it’s a good law for the individuals concerned,” he says. “It seems there is a need for this — and people go for it for genuine reasons”.
cp
serious_crime
political
public_figure
private_personal_info
Numbers accurate as of March 23rd 2015
Given a universal set of criteria to give a thumbs-up or thumbs-down, a request should be decided equally in each country. Ok, we know that different countries have slightly different request issues, so it is only fair to compare within one group. Here, we are looking at Google's decisions for the issue type ‘private_personal_info’, the largest category by far, covering 96% of requests. Google's compliance with the request varies significantly between the countries. While 50% of the requests from Germany and France in this category were successful, only about 30% of Bulgarian requests were.
Again, I am hoping these data stimulate questions, and journalists who have more local insight will be able to help to uncover why we see certain country-specific trends.
If you want to help and extend the insights of this page to your country, please fork the HTML on Github and add a version in your language!
Numbers accurate as of March 23rd 2015
