11 Dec 2013 Facebook’s Year in Review – can social data choose your most important memories
Facebook’s Year in Review feature is back for 2013. Not so much their year in review, but yours – the feature brings together your most important ‘moments’ on Facebook in 2013. The photos and status updates that attracted the most comments, the friends you interacted with most collected together in one space to tell the story of your year. Dig a little deeper and you can find your friends’ year in review too – the most memorable moments from all your friends. But can data really tell this story for you; can memories be constructed based on data analytics?
December is often a time when we reflect on the year – the things we have done, achieved and learned; the places we have been and the friends we have made. It’s often also the time we make resolutions for things we will change – things we did this year that we will choose not to do next year. Which memories we choose to recall will be different for different people and will depend on how we reflect on the year as a whole. They will reflect how we think about our year when we look back on it, what is important to us with the benefit of hindsight and what we want to reflect on.
What Facebook is offering us is very different. There is no reflection, no interpretation and no benefit of hindsight. The memories they choose to construct our year are the ones that were most popular at the time they were shared.
On one hand this might look like a perfect objective data-set – no interpretation is applied, we are not covering up things we would like to forget. However, building memories and telling stories in this algorithmic way also brings problems and biases. And this raises interesting challenges to anybody using similar social media data, especially as we consider the future of market research.
We have written before about the nature of memories and social media data and what this means for privacy concerns and the value of the data we can get from social media data. But we should also raise a concern about using this kind of data alone to select what is important.
When viewing the ‘most memorable moments of your friends’, you see a lot of what you might imagine. When forced to choose just one event per friend per year, it is unsurprising that the big events come forward. Births, deaths, marriages, new jobs and relationships. This is probably a fair reflection of what was most important across your friendship group in the year.
It is when you look at an individual’s own memories that things become less clear. Facebook chooses the memories that attract the most interactions and this depends very much on the time of day and year you post, what your friends are doing at the time and many other criteria. For example, is a discussion of a sandwich I ate on 4 January 2013 really one of my most important memories of the year? Facebook data would suggest it is as lots of my friends (perhaps bored on their final day off work for New Year) commented on this post. I can honestly say I had forgotten I’d even eaten a sandwich on that day, let alone what was in it.
Just because a post cause reaction at the time it was posted, and just because this might have been more reaction that we usually get, does not make that post important as a memory. To say so tells an incorrect story of my year.
Using algorithmic analysis of memories in this way is perhaps useful for spotting the most extreme outliers (the birth or marriage that will attract a lot of comments) but is less useful at telling a consistent story of an individual’s life and what matters to them or happened to them.