The Facebook Experiments

There has been numerous reports about the Facebook study of manipulating people emotions in the media and I even shared the story on my feed as I thought it was quite interesting. However very few people have looked at the actual research, and given the media habit to make a mountain out of a pebble. I thought I’d actually find the paper and read what they had done, and are potentially still doing.

Most of the outrage came around the fact that this sort of experiment on human subjects needs proper ethical consideration and informed consent. However it seems more that Facebook has used the ‘assumed consent’, due to a condition in the terms of service, which says that your data can and will be used for research. I have not read the full terms and conditions, but I am sure they will also sell your data to the highest bidder.

The fact is Facebook is optional and whilst I admit that having someone ‘manipulate your emotions’ as has been reported in the media made me like many sit up and think, ‘hang on a minute, thats wrong’. Even more ‘outrageous’ was that there were 689003 people chosen at random for this experiment, yet they are probably selling data and access to market niches based on all the users of Facebook. You should have a look at the specific niches you can market to on Facebook. (25-40, Interested in Climbing living in the UK, there are around 80000 people in that market segment)

However reading the paper I see that that manipulation was at best minimal. This is where the devil is in the detail, the experiment last 7 days. The manipulation was to alter your newsfeed positive or negative comments by 10%. Or lets say in those 7 days there were 100 comments that were +ive and 100 that were -ive, what Facebook did was show you either 10 more or 10 less posts that were either positive or negative.

The design of the experiment was interesting if you are a social scientist as lets say they want to increase +ive side. They could illicit a similar response by both increasing you +ive feed by 10% or decreasing the -ive feed by 10%. Two very subtle differences but they still had the same result.

Most interesting of all for a marketing perspective, was that hidden in the result was a golden nugget. Those emotional posts that can make you think you or someone else has shared a little too much of themselves online be they positive or negative got more engagement than an emotion free post.

I am not saying what Facebook did was right or wrong, but I think it important to highlight what the ‘manipulation’s’ were and that the experiment was actually into ‘social contagions within networks’. What is really incredible was that that small change in post with emotional content in your feed had an effect on people at all, despite them living most of the lives in the real world with all the ups and down associated with it, they were still changed by a 10% change in emotional feed.

This brings up even bigger issues, Facebook had to deny government backing for the experiment. If it had it would very much be big brother is watching you territory. It is interesting though as arguably Facebook could be used to influence us in more creepy ways. Google was accused of influencing the last Indian election through displaying skewed results, something that it vigorously denied. Whilst in the California election Facebook believed that they increased voter turnout by 2% by having a ‘I have voted’ badge appear on people timelines, whilst not as effective as face to face canvassing, it is much cheaper.

I guess the fact is that we are living in a world were massive amounts of data is created on us all, much of the time by us. Just look at the NSA and GCHQ mass spying allegations. Yet we still use Facebook, email and other methods to communicate where the service maybe free but its true cost is that the data is sold to the highest bidder.

The question is do you want the power to be in the hands of the governments or commercial entities like Facebook. Another of Facebook’s experiments discovered people who are lonely or depressed use the social network in a different way to other people. You wonder with all the data they have whether they could put that to ‘good use’ and see if there are precursors or warning signs to people becoming suicidal. Facebook, have already caught a child sex offender through one of its data mining algorithms, preventing them meeting up with a young child and helped leading to a conviction.

What the world is referring to as Big Data is potentially an awesome thing. The debate is whether it should be used to make the world a safer or nicer place. In one of my favourite books freakonmics talks about a forensic accountant that created a search criteria that identified the 7/7 bombers and a handful of other people, who instantly became a target for further investigation, all possible because of data mining.

So how can we used big data for climbing? Well there are many things we can do. I once for the fun of it created a map that auto populated from a twitter search for #climbing. I never made it public as I thought is was a little freaky because most people inadvertently have location services switched on, and when they tweet their location is populated. You could see where many people were climbing and I felt the idea was a little too ‘stalker friendly’, as I suddenly realised that I knew what crag Tom Randall was at that day, and so could anyone else it they used the search API.

UKC also use biggish data by getting thousands of climbers to vote on the grades of climbs. This data was then used in North Wales Climbs and other Rockfax guidebooks. I do wonder how the massive online collaboration will effect guidebooks in the new era of apps.

Anyway sorry for the techno blog, climbing service will return shortly.

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *