... and does it seem like it is getting hotter to you?
While doing a bit of cleaning in the kitchen I found a surprise a friend had hidden nearly in plain sight. I was on the step stool storing some things that are rarely used high in the cabinets that are rarely reached. Looking around I saw the top of the refrigerator was very dusty. I pushed the stool over and found a smiley face that had been drawn with a finger looking back at me. A time capsule from the past.
This past week a kerfuffle erupted over a published Facebook experiment and I found myself thinking about social computing and the importance of filtering. I tend to think of three issues central to social computing - security, privacy and filtering. The first two are often confused and all three can be related. Issues of trust are tied to all of them. I'll approach it using filtering.
I didn't see the smiley face because my vantage point for seeing the world wasn't high enough. The person who drew it a few years ago is about a half foot taller than me and seeing the top of refrigerators must be a common experience for her. Making use of this she left a little time capsule in her plain sight using my comparative lack of height as a filter. Perfectly played:-)
Our senses only give a rough approximation of the world around us. We live slightly in the past due to neural signaling and synchronization speeds and some aspects of vision now appear to be averaged over very long periods (10 to 15 seconds) to beat down noise in the environment and our nervous system and to allow us to synthesize a stable image. Our vision is sensitive to a small part of the electromagnetic spectrum selected by evolution to correspond to the majority of sunlight that makes it through the atmosphere and that is energetic enough to cause chemical reactions without destroying chemical bonds. The signals from retina to to visual cortex undergo a good deal of processing and the effective data rate isn’t impressive — about the same as a twenty year old ethernet cable. We don't respond to fast or slow events, our sense of smell isn't up to that of our dog, and the list goes on and on. But from this highly filtered information we construct our own reality of sorts and we, at least I, find the result rich and compelling. We can and have managed through our wits to augment our senses and to quantify the result allowing for an empirical study of nature. And we have a brain that is powerful enough that we often can make empirical measurements and extend the range and variety of our senses.
One type of reality we construct is in the form of story. We experience the stories of others, construct our own and sometimes mix them. We are wired for patterns which can inspire them in our minds and fix some degree of trust to stories from others. Story telling is important to sociology, anthropology and the like is extremely complex and difficult to study. Complexity is tough.
An impressive amount of work goes can go into the storytelling in a movie. The music is carefully selected to enrich your experience — a filter. Consider this analysis of the main theme for ‘Frozen'
Facebook trades in stories. It wants to be the medium that people can share their own stories with friends and perhaps act on the stories advertisers are paying them to find an audience for. In my mind Facebook has two goals: first to keep its audience engaged in their conversations and second to serve their customers - the advertisers. Without the first the second is impossible. They need to understand how to keep and grow their audience and a firehose of data - very rich data.
Facebook needs people who can understand it and how it fits with their business. They employ a lot of data scientists, but data science is far from being a science. It is just a tool for addressing questions. It is important to ask questions that can be meaningfully answered.
The best people for understanding social systems are seasoned social science types. The good ones are driven more by research and the thrill of the hunt than by money. They often see the business world as limiting, but places like Google and Facebook offer treasure troves of rich data to play with.1
There was a failure in Facebook’s PR and legal departments. This type of internal research is very common and important for Google, Facebook and others. We did it in the data with call detail records at AT&T, dating sites and others do it all the time. Controlled A/B tests (and many other techniques) are how you figure out how to turn the knobs on your filters.
My hope is that there might be a conversation on filters. What they are, how they work, who controls them and so on. We’ve been dealing with filters for centuries. They impact politics, consumerism and even belief. … print, electronic and face to face: newspapers, radio, tv, used car salesmen, religion. Now they can be personalized (face to face has been able to do that for a long time, but it doesn’t scale well). A problems is filters can create signals that seem wrong or may have an unintended consequence. If FB tries to make people slightly happier in order to keep them it may be a mistake. For example some recent work shows that some people, when they see everyone else is happy, tend to get more depressed. The whole notion of what an emotion is turns out to be very fluffy.
I think the firestorm may be partly driven by an unease with 'big data' — few know what it is, how it is used and how it impacts them. They see ties into security and privacy - and now this may be illuminating filter manipulation for many.
Deeper issues why people are upset are largely overlooked. We have any number of signaling and out of band communication that are missed by the measurements of social media. They would completely miss the time capsule joke Colleen drew on the top of my refrigerator and millions of other subtle and not so subtle interactions.
We rely on computer aided networked communications to send fragments of larger stories, communication exchanges and other forms of signaling to others. It is a mistake to think this is sufficient and a greater mistake to attempt to curate it.
That may be a central point. In the past we have relied on media companies to give us books, movies, television and other types of media to consume. It is filtered culture, but we are mostly aware of that. We also rely on others to carry our remote communications - the post office and telcos. We have trusted them not to filter meaning. The phone company doesn't decide what is the the conversation we're having with our sister.
Media companies regularly do A/B testing to see what we will buy. Communication companies do testing of the quality of signal to optimize their networks. What interests me about Facebook is not that they have done A/B testing as organizations that have filtered information often do such things, but rather that they are filtering our personal culture - they have put themselves in the position of making some rather delicate choices that impact personal signaling and communication. We only have sketchy ideas of how they are doing this. I'm guessing their attempt to curate our personal culture may be at the heart of what is bothering so many.
I sent this image to a few of you the other day. A version of the famous duck/rabbit illusion, it shows how our brain can be on the fence making an image conclusion. Some practics data science poorly - a lot of not understanding the data, its errors or the questions. Pretty patterns are found - sometimes great effort goes into sorting out is it a duck or a rabbit, when in reality is is neither and had nothing to do with animals in the first place. It happens all to easily when there isn't enough information about the situation in the first place (it can also happen when there is too much information or a lot of noise). Social signaling and communication is much richer than what social media can deal with. Perhaps we have found a valley of social uncanniness with Facebook.2
We have a long way to go in our understanding of human interactions - I suspect current social media will be footnotes in the history of the subject a few decades from now. Hopefully we will begin to ask questions about filters and what they mean.3
1 It is very difficult to find this in the academic world and people will go into these companies with the intention of staying a few years and learning and then moving on. To keep them you can allow them to publish. Google mostly doesn’t and it shows. Facebook has tried a few times. My guess is this was part of a recruiting effort. The paper is ok - not brilliant or very descriptive. The problem it is addressing can’t easily be answered with the technique they used. They can get a sense of the windage and make adjustments. The nature of how the review board acted is under question, but that points to issues with academic review boards. I can’t see anything really wrong with this. The tech press went crazy, but it seems they weren’t parsing the meaning of some words that have very different definitions when used in academic research (a problem all researchers have when trying to communicate with the general public - I have failed terribly at times). It also strikes me that many people writing about this didn’t read the paper and certainly didn’t understand the context.
2 For the record I have a Facebook account, but only use it around my birthday and Christmas as something of an updated holiday card. I try to have deeper connections with those who are close to me. I also gave up on Google except for YouTube.
3 This will probably blow over like so many other issues do. It is important for us to have discussions and hold filtering companies accountable - or at least realize what is going on and make intelligent choices as users. Sadly the track record for society is not very good. Failure could be dangerous - we need to be asking ethical questions of filtering in everything from MOOCs to democracy. What is the public square anyway? What is the peronal discussion?
BBQed Sweet Potato Fries
° 2 large sweet potatoes, cut in french fry like strips
° 2 tbl olive oil
° 2 tsp sea salt
° 1 tsp ground cumin
° 1 tsp chile powder
° 1 tsp paprika
° 1 tsp freshly ground black pepper
° 1/2 tsp cayenne - more if you like hotter, less if milder
° toss sweet potatoes and oil in a large bowl
° combine spices a small bowl and mix
° add the spice mixture to the sweet potatoes and toss
° cook in a vegetable holder on a grill until the sweet potatoes brown nicely - alternatively cook them in a fairly hot (425°F or so) oven for about 25 minutes turning them halfway through.
° these can be very good with a yogurt or sour cream sauce. I made a yogurt sauce with plain yogurt, chili sauce, chopped cilantro, garlic, lime, pepper and salt .. what was around.