Monday, June 30, 2014

Facebook Psychology Study Scandal: BS Science to Prop BS Technology

I've got better things to do than write about Facebook's questionable practices (or Facebook in general, or any other social networking site for that matter). So, this is going to be a short rant.

The Ludovico Technique is performed on Alex in Stanley Kubrick's "A Clockwork Orange". Image credit: Warner Bros.

Like many of those in the habit of checking the news in the morning, today I couldn't help noticing the media outcry about the study Experimental Evidence of Massive-Scale Emotional Contagion Through Social Networks published in the Proceedings of the National Academy of Sciences.

If you followed the link above and read the article, you may have noticed that, according to the researchers, "[p]osts were determined to be positive or negative if they contained at least one positive or negative word, as defined by Linguistic Inquiry and Word Count software (LIWC2007)". LIWC itself is beyond the scope of this post, but -- even assuming that it is a very reliable text analysis tool, which I doubt, especially when it comes to informal language (just think for a second about the positive and negative meanings of the word "shit", as in "He knows his shit" and "He knows shit", and you will know what I mean) -- the way it was applied by the researchers is problematic. This is not the only thing that raises questions about how kosher this study is. I have neither time nor desire to further elaborate on those. Read the article. They are hard to miss.

As I have written before, Google, having been sifting through thousands of my e-mails for eight or nine years now, still "thinks" I am a plus-size flirty female who likes knitwear, has anger issues, and may be a little slow in the head:
Above is the ad that recently popped up (and kept consistently doing so) in my GMail. It must have been based on thousands of words I used in my e-mails over the years (and, most likely, on my on-line behavior patterns as well). The result is definitely impressive, isn't it? :-P Yes, the ad did get my attention, but only because of how absurdly irrelevant it was. I have no reason to believe that Facebook is any "smarter" than Google.

We, humans in the IT industry, have figured out how to collect and save petabytes of data about other humans. We just have no idea how to make sense of it (especially when it comes to the ability of machines to "understand" human language). However, because there is so much hype about "big data" and so much money in it, nobody will ever admit that. So, the "big data industry" needs some quasi-scientific props to keep the bubble growing. If there is some kind of scandal involved - even better: it will go viral, yet very few will pay attention to how treif (the opposite of "kosher" in Yiddish) the actual research is.

To me this whole fuss about Facebook messing with its users' heads looks like just a clever campaign orchestrated by... Facebook. Wait a week or so. Now that there is a scientific study published in a reputable journal implying that Facebook can manipulate its users' emotions, Facebook will raise its advertising rates...

Aside from what many find scandalously unethical (by accident or by design) about this study, it isn't any different from those that suggest that coffee prevents Alzheimer's disease or eating nuts makes you near-immortal. Coincidentally :-P such studies seem to be always paid for by coffee retailers or nut growers. The few of us who bother to read the publications usually find out that even the researchers themselves admit that the evidence is inconclusive and further research is needed... The rest see the sensationally-sounding headlines in their Google/Yahoo/whatever news feeds, and that is exactly what the businesses backing this kind of research want. Don't get me wrong -- I like coffee, and nuts, and even big data. I just can't stand hype and BS.

Further reading (loosely) on the topic:

No comments: