How aware are you of the algorithms that rule your life? I have a vested interest in your answer. I make my living building vibrant online communities from participants on social media platforms that lead to increases in revenue for clients. Algorithms are a perplexing problem in my decision making.
Devious mediation by algorithms imprint the data received from social media platforms. Much of my time becomes invested in cleverly reverse engineering them. This muddles the insights that data from these sources provide.
We have to ask ourselves, what information is reliable when we work with data imported from social media channels?
The current reference frame that marketers use to make sense of social data is incorrect. We know that follower counts, likes, and comments will not necessarily generate leads and are poor indicators of performance. The uncertainty that algorithms bring also introduces reliability issues when we measure this type of data.
We need a new model that corrects for this. Research shows that creating detailed and accurate psychological profiles of users from social data is a better indicator of their intent.
This article will provide profound new insights from the field of human-computer interaction which will clarify the following points:
- How actions on social platforms impact algorithms and their devastating consequences
- How to frame reliability issues in social metrics for better client reporting
- What psychological metrics to consider for reporting conversions
Do people mess with the Facebook Algorithm?
There was a heated debate on the ArCompany Millennial Think Tank revolving around issues of online privacy. During that discussion Amy Tobin mentioned the sometimes ridiculous ads she sees in her Facebook feed. Our panelists admitted to occasionally clicking on things to purposefully play with the algorithm that curates what appears in their timelines.
Although you might have experimented yourself, you are part of about only 25% of users who are algorithmically aware.
That is according to a paper for the ACM CHI 2014 conference in Toronto authored by Kevin Hamilton, Christian Sandvig, Karrie Karahalios and Motahhare Eslami.
They conclude their paper, A Path to Understanding the Effects of Algorithm Awareness, with the statement:
Our ongoing work to this end suggests, for example that less than 25% of regular Facebook users are aware that their feeds are curated or filtered, and even less know how to affect that process.
According to Facebook’s Quarterly Earnings Report for Q2 2014, they currently have 829 Million active daily users.This means about 600 million people worldwide have little to no awareness of how to impact the algorithm or that it even exists.
This raised some red flags for me about the reliability and dependability of the data which I use when consulting with clients.
There is an assumption in algorithms that goes unspoken, free will on the part of the user. Facebook assumes that you are an active member of their platform making decisions that relate to personal preferences.
But 600 million individuals are doing nothing to consciously impact the decisions the Facebook algorithm makes daily. Who would have the time?
It took me three years to create a Lo-fi music station on Pandora that only plays music that I like. That required a certain amount of loyalty, dedication and consistency on my part.
That is why I question that the data any platform produces from an individual’s actions is an accurate reflection of their preferences. Would they become more proactive curators through their individual action if they were aware that it would change their experiences?
Here are two people who took this question to the extreme!
- “I Liked Everything I Saw on Facebook for Two Days. Here’s What It Did to Me” – Mat Honan for Wired
- “I Quit Liking Things On Facebook for Two Weeks. Here’s How It Changed My View of Humanity.” – Elan Morgan on Medium
It is only because of these absurd lengths on the part of the authors that we see how simple actions aggregate and change our experience.
Analyzing big data in regards to social requires making assumptions which we are not yet qualified to make. Choices by users may not represent their true preferences. The results that algorithms produce from those actions may not represent what users want. These are two halves of an important whole that governs the choices that marketers make on spend.
Having more data on its own won’t resolve the problems that we face.
Reliability Issues in Big and Small Data
Working with social data isn’t the same as working with the data which comes from observations of nature. I can track the PH balance of the soil in my front yard over time. This is a definitive quantity that has a direct impact on the way plants grow and their ability to thrive. I can correlate changes with temperature or time and be certain of the correctness of results because I am measuring physical properties.
We are lacking such precision when it comes to tracking social data.
How do we isolate an individual when we look at data? We have multiple usernames, accounts, and purposes when we interact online. You might be clever and think that IP address is a clear identifier. Not all users have static IPs. Users can also use proxies to hide their information. An IP address can also be spoofed for purposes of concealment or impersonation.
Let’s face it, the API of most social networks are restrictive on purpose. It has more to do with profit margins than security or privacy of users. Restricting the API has not prevented platforms from selling user information. Everyone loses this way as it also prevents individual users from freely accessing their own data.
Smart devices are everywhere! It impacts the way traffic sources appear in your website analytics. There is a multitude of ways that analytics source traffic from social media. Take Twitter as an example. In Google Analytics it will appear as various sources; twitter.com, m.twitter.com, or t.co (Twitter’s own link shortner). But it doesn’t end there. It may also be from Bit.ly or Hootsuite or Ow.ly or any other 3rd party way that can post to Twitter on the behalf of users.
We need to pivot the metrics we use as marketers to understand performance. A way to tie this behavioral data to something more consistent and reliable. Big data can still help us, but I would like us to move away from the idea that large data sets will lead to the solutions. It helps in a more important way by shining light onto previously dark corners of the social landscape.
Humanity in Data
I speak frequently to my coworkers at ArCompany about a personal philosophy of my own. We need to look for the humanity in numbers.
When it comes to social you cannot separate behavioral data from the mindset that creates it.
There is a shift in marketing that still needs to take place. This means our methodologies for understanding performance and optimization also has to adapt. We must stop thinking in demographics like location or age and opt for a focus on psychographics like extroversion.
There is research that supports this fact.
Modern psychology recognizes five dimensions of personality: extroversion, agreeableness, conscientiousness, neuroticism and openness to experience. Previous research has shown that people’s scores on these traits can, indeed, predict what they purchase. Extroverts are more likely to respond to an advert for a mobile phone that promises excitement than one that promises convenience or security. They also prefer Coca-Cola to Pepsi and Maybelline cosmetics to Max Factor. Agreeable people, though, tend to prefer Pepsi, and those open to experience prefer Max Factor. –No hiding place, The Economist
Big data plays an important role in linking sales with psychology. A research team discovered that they could accurately profile a user’s personality with as little as 50 Tweets!
We can use big data to create accurate profiles of individual’s values, views, self-expression, identity and other psychological factors of interest.
There are so many points which we could debate and I hope you will do so in the comments. The looming question is how intrusive is it to use social data to build an eerily accurate and detailed psychological profile of a user?Does a TOS agreement serve the same purpose as informed consent (i.e Facebook’s recent experiment)? In what ways do you understand the algorithms of your life? Do you feel that you are able to influence them in a way that improves your experience of using social platforms?
Originally published on SteamFeed