On 20 May 2017 Twitter sent an email to its users announcing that new terms and conditions apply. These changes largely apply to data usage, personalization and targeted advertising. One can’t really say it’s a revolution, but at least it’s a first step towards more transparency (something much needed in the field of Big Data). And the new privacy parameters give you an idea of how the personalization algorithms of twitter work and which kind of data is used for this purpose.
The classical data collection processes : devices and places
The two variables you can now disable are “devices” and “locations”. They are rather classical : nothing really freaky here at first sight but it might be interesting to go deeper.
Let’s start with places : depending on where you go or where your account was created, personalized content will appear. This makes perfect sense and is actually highly useful. The only thing you need to realize is that Twitter collects location data wherever you are and disabling location-based personalization will not (this is our interpretation at least) prevent from having that data collected.
Device-based personalization is more tricky. Once again it makes perfect sense from a user experience perspective. You are not consuming the same type of content accross all devices. Certain devices may be used more often to access sport content. This would lead Twitter to display different type of content depending on the device. Where it becomes more tricky is when it comes to the data collection process. Here’s what Twitter says about it :
“When you log in to Twitter on a device, we associate that device with your Twitter account. Whether or not you are logged in to Twitter, we may also receive information about your devices”
The less classical data collected : applications, pages visited outside of Twitter
Those two types of data seem to me to be much more invasive.
Twitter apparently collects the list of applications installed on your Android smartphone and personalizes content accordingly (probably using a type of collaborative filtering). We don’t see at first sight what kind of precision can be leveraged through app-based personalization, but if Twitter does we guess there is some value behind it.
Twitter also collects browsing data outside of Twitter itself. As they write
” we may consider your visits to other websites that integrate Twitter content (such as embedded timelines).” Facebook used the same kind of trick with the embedded “Like” button to track users outside of Facebook and this was highly criticized.
Some good news : removing consent is made easy
One good news though. Removing consent for personalization and (part of) the data collection processes is made easy. The general conditions are also very clearly written and in some reassuring of the user’s right to privacy. Consider for instance this sentence that appears in the section on data collected outside of Twitter :
” To protect your privacy, we never associate this web browsing history with your name, email address, phone number, or Twitter handle, and we delete, obfuscate, or aggregate it after no longer than 30 days”.
This is clear, to-the-point, and kept at the lowest level of technicality so that everyone can understand.
Posted in big data, Innovation.