Image may be NSFW.
Clik here to view.
This is Rhizome Today for Thursday, November 13, 2014.
Rhizome Today is an experiment in ephemeral blogging: a series of posts that are written hastily in response to current events, and taken offline within a day or so. The latest post can always be found at http://www.rhizome.org/today.
At the will of market pressure and general user trends, commenting culture is in a forever state of flux. Michael Connor has lamented its demise. Sensing its migration to social media, Rhizome's conservation program has taken an interest in important discussions occurring on Facebook, which aside from screen shotting are not really easily or intuitively documented. eflux, in response to the user shift to big social, just launched its own bespoke social platform, Conversations.
The other day I saw this article about Buzzfeed striking a deal with Facebook which would allow the publisher to gain access to the social giant's data. The aim: measuring sentiment during political elections, since social media is where lots of people express their political opinions these days. The article was about a huge problem in maintaining a neutral sample since Facebook algorithms are set up so that each person's or group of persons' feed is tailored, affecting how people respond or react.
This reminded me of a presentation I saw several months prior about a tool called SentiStrength, and how it measures sentiment within language on social networks. Basically the whole presentation started to sound like one big problem because people misspell words and people are sarcastic. So saying something like "i am sooooooo happy that my computer broke" is maybe registered as -2 negative for the word "broke", but +3 positive for "happy", and then maybe +2 for "sooooooo" before the "happy". Doing the math, SentiStrength would argue +5 –2 = +3 positive sentiment. (I am somewhat bastardizing the explaination.) Clearly, the whole process for automating the measurement of sentiment within natural speech is soooooo limited. Not to mention the tool is most adept to analyze and account for English speech, which makes little sense when measuring political sentiment in a diverse country with no official language.
A sentence or a paragraph is made up of a range of sentiments, not to mention levels of lyricism and vernacular, all of which yield a mathematical mess if you try to automate determining an ultimate position. Data gathered from particular platforms, be it Twitter or Facebook, say a lot more about the platform itself. That is, the mode in which people speak because of how the platform is structured: character count, pace, form of replying, retweeting, liking, faving, and the types of communities that tend to gather on Facebook or Twitter rather than, say, 4chan or Reddit.
Given Facebook's ultra-secret ways, their method for sorting sentiment within phrases is probably much more advanced than what is available to general researchers in the field of political science or media studies—those using tools similar to SentiStrength. Whoever has the ability to quantify and properly sort our daily feels-sharing is constructing the census of the populace, and providing evidence of affectivity, knowledge of which, it has been argued, can influence not just purchasing, but voting. In retail and politics, our feels are very valuable, and, it seems, pliable.
I suppose the most successful ads are ones you don't know are ads at all. It's like walking into a froyo shop in the summer of 2013—it feels artificially natural. The general argument to avoid this algo-subconscious influence is to decentralize across platforms. Unfortunately, many alternative platforms are really just alternative compared to mainstream media like major television networks, and while they may perhaps decentralize the value and influence of those mainstream channels for a bit, the popularity of their alternative voice is most often reabsorbed by the mainstream. After all, the Buzzfeed deal with Facebook also includes ABC.
I'm trying to think of a listicle joke here to correspond with the Buzzfeed style, but failing...maybe we are all just listicled out. A wise tweeter, handle @pourmecoffee, once tweeted (on October 9th, 2014):
Obama just posted a note to millenials on @medium. Also, check your Moleskine diary, he wrote a special note in there for you.
— pourmecoffee (@pourmecoffee) October 9, 2014It has 40 retweets and 82 favs. It must mean that people really like Obama. "Sentiment is up; young people on Twitter love Obama and his innovative demographic-specific platform penetration and awareness," study finds.