Social Media Distancing 3: Technological Innovation
5th of7-part series, each standing on its own (see list at end)
I nerded out for months over experimental studies aimed at understanding the role of algorithms in political polarization. But now, in the midst of the COVID-19 pandemic, I find myself tempted to unsubscribe from newsfeeds from the opposite end of the political spectrum. Theory is no match for practical emotions during times of crisis. It’s getting personal.
In this fifths part of our seven-part series on the search for a more healthy relationship with our digital algorithms, we look into the intersection of our ‘Algorithmic Exist Strategy Matrix’ that explores how we could technologize our way out of the current challenges. In agreement with the 8,000 prominent signatories of the open letter on research priorities for robust and beneficial Artificial Intelligence, the goal is to assure that “our AI systems must do what we want them to do.” This starts with rather ad-hoc attempts to curtail the symptoms of an unhealthy system and ends with inventing the yet unimaginable that replaces the existing industry giants with something new. What will replace Google, Facebook, Amazon, Apple, and Microsoft?
Curing the Symptoms of the Current Solutions
The global pandemic shot screentime through the roof and it quickly became mainstream advice to “limit how much information you consume about the coronavirus outbreak.” The technological response to this call is so-called ‘time well-spent apps.’ It was shaping up to be tech’s next big debate in 2018. But the debate was effectively over before it really began, as Apple, Google, and Facebook swiftly added features designed to help users to monitor and manage their digital consumption. A new industry of technological well-being and related coaching services started to boom around screen time track and limiting. By now, one in four 16–34-year-old has used them.
Critics counter the industry’s enthusiastic endorsement of these new tools with an analogy of a tobacco company offering you an app to track your daily cigarette consumption. Persuasive technologies are designed for addiction. Just as it sounds like a bad plan to ask the hungover brain of an addict about the desired consumption the following night, it is questionable if these add-ons are a match for our dopamine-craving brains.
Going one step further, technology ethicists argue for including safeguards against the exploitation of human vulnerabilities right into the very design of a new generation of technologies. Those could be used to nudge people into socially desirable behavior during the pandemic. Nudging is a Nobel Prize-winning concept that proposes positive reinforcements and indirect suggestions to influence human behavior. During the pandemic, this has resulted in a variety of gamification tools in support of lifesaving social distancing measures and the fostering of personal responsibility through the quantification of your pandemic footprint. It could certainly be scaled up. A 61-million-person experiment during the 2010 U.S. congressional elections showed that a single go-out-to-vote message on Facebook increased voter turnout by over 340,000 people. The spread of a virus is a collective action problem, and nudging can be useful. It uses algorithmic manipulation for the good.
Other proposals for new technologies advocate for repurposing existing technologies. It has long been proposed that digital surveillance on citizens should be accompanied by a similar increase in surveillance on public servants. The technology is here and ready to go. It is just focused on the citizen and consumer, not on the politician and CEO. If surveillance systems would be set up, the trillion dollars spent by COVID-19 relief packages across the world could be accompanied by a high-tech real-time monitoring system. We would closely track anyone getting their hands on the money, just like we do when watching a big-brother reality show in a glasshouse.
Hard Problems
Some aspects of the tech puzzle are genuinely hard to solve, mathematically. For example, how to choose content that — at the same time — reduces polarization and keeps you profitably engaged with a social media platform? Today’s algorithms maximize profits at the expense of growing polarization. Showing you content you agree with, keeps you engaged and happy (by feeding your innate ‘confirmation bias’). But it also makes you an extremist. Think about bringing together a group of young men who love fast cars, and another group of mothers who lost a child in a car accident. But before you have them talking about speed limits, feed them with social media content that reinforces what they previously engaged with. Good luck trying to sit them on the same table. Ideological affect for political parties has increased by 36% between 1996 and 2016, while perceived polarization increased by 32%.
The alternative is to show you content diametrically opposed to your current beliefs and interests. Since the 1960s we’ve already known that this will also strengthen your existing beliefs and also lead to polarization. Mathematically speaking, the identification of these two extremes is relatively easy in a multidimensional vector space. There are few ways to agree and to disagree with you. Follow the direction of the vector in your AI vector space, or reverse it. But there are infinite ways to expand your perspectives. What other directions could the vector point to? That’s a hard problem.
Instead of categorizing social media posts according to their political left-right ideology, we classified YouTube videos according to their emotional content the week after the 2016 Presidential election. We found that joy is prevalent in content that leads to emotional polarization, while sadness plays a significant role in emotional convergence. What brings us joy, unfortunately, drives us apart. When watching sad videos, our compassion seems to bring us together. Social media companies might not be excited about the prospect of trying to keep you lucratively engaged by flooding you with depressingly sad content. But emotions might be one piece of the puzzle of designing recommender algorithms that produce both profit and social cohesion.
There are countless variables to classify content and exponentially more possible combinations of those. Any solution to this problem is unlikely to find a definite optimum, ever. This opens up a long-term race to find algorithms that help to build an ever more sustainable network of democratic opinions.
You’ve been waiting for it: blockchain!
Finally, no discussion of future technological promises in 2020 would be complete without a shout out to the blockchain. It has become a running gag among technologists at the latest tech event to bet on the timing of the next exaggerated overpromise from a blockchain enthusiast. However, it is not only the opinion of tech forecasting gurus like George Gilder that “Life after Google” consists in “the Rise of the Blockchain Economy”. The decentralized, distributed, and oftentimes public, digital ledger can be used to record transactions in a way that any involved record cannot be altered retroactively without notice. Additionally, as evidenced by worries about the intractability of some cryptocurrency transactions, it also offers solutions for anonymization. Recently, the World Health Organization (WHO) has partnered with major blockchain and tech companies, including IBM, Oracle, and Microsoft, to launch a distributed ledger platform called MiPasa in the fight against the coronavirus.
Innovation’s First and Last Name: Creative Destruction
According to the legendary science fiction writer Arthur C. Clarke, “any sufficiently advanced technology is indistinguishable from magic.” In 1945, he proposed the magical idea of a satellite communication system using geostationary orbits to give world-wide radio coverage. It was first implemented 20 years later. We cannot predict what will be the next big thing for sure. But we do know for sure that also Google, Facebook, TikTok, Apple, Microsoft, and Amazon will sooner or later be replaced by technological companies that satisfy a more extensive variety of consumer demands in a pleasing manner.
This process creates new things and destroys the existing ones. For Schumpeter, the prophet of innovation, the “perennial gale of creative destruction… [drives the] process of industrial mutation… that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one.” Schumpeter’s students in the field of innovation theory have long pointed out that new technological breakthroughs might not be entirely predictable, but they can surely be fostered and guided into specific directions.
In what has been called the ‘best presentation of all time’, the legendary Steve Jobs (chief innovator of Silicon Valley, 1974–2011), announced that he combined different solutions into one, creating the biggest innovation of a generation. Schumpeter defines that “innovation combines factors in a new way, or that it consists in carrying out New Combinations.” Jobs was a real innovator for the textbooks.
Turns out that all of the building blocks — including the microchip, internet, cellular communication, GPS, touchscreen, voice-recognition — have been government-funded. Independent from the fact if this government funding is hidden in a national defense budget or straight-out in form of public subsidies (as done by the European Union), public funds have always played a huge role in directing innovation into the future. This stems from the simple fact that basic research is way too risky for any private investor. The vast majority of it fails. The odds are better in a casino. The only systematic way to drive this game forward is to shoulder this financial risk among all of us and use taxpayer money to venture into the yet unimaginable.
As the current times of social distancing glue us to our digital reality, we get ample opportunity to think about what we’d like future generations of algorithms to look like.
This series consists of 7 articles. Each providing a complementary perspective (no need to read all, or in order)
Social Media Distancing: An Opportunity to Debug our Relationship with our Algorithms
Social Media Distancing Context: Algorithms Found our Weaknesses
Social Media Distancing 1: Industry Self-Regulation
Social Media Distancing 2: Public-Sector Regulation
Social Media Distancing 3: Technological Innovation
Social Media Distancing 4: Human Innovation
Social Media Distancing Exit Strategy: Toward a Healthy Relationship with Our Algorithms
An accompanying series of video presentations can be found here (part of the online course “Digital Technology & Social Change”, at the University of California).
