London Interdisciplinary School ©2019

The Fine Line Of Free Speech

400 hours of content are uploaded to Youtube each minute and a billion hours are watched every day. Internet companies are expected to remove extremist material within an hour of it being flagged, but firms struggle to meet these targets.

Develop a regulatory strategy which aims to contain the problem, without enabling discrimination or empowering oppressive regimes.

01. Ethics

Humanity through time and space

Moral philosophy is a branch of philosophy that involves systematizing, defending, and recommending concepts of right and wrong conduct. The field of ethics, along with aesthetics, concerns matters of value, and thus comprises the branch of philosophy called axiology. Read more

An enormously broad topic but is the fundamental basis of not only how we judge the moral content being uploaded, but also how we judge our actions to how we deal with that content.

02. Sensitivity To Fairness

Humanity through time and space

In an illustration of our relative sense of well-being, we are careful arbiters of what is fair. Violations of fairness can be considered grounds for reciprocal action, or at least distrust. Yet fairness itself seems to be a moving target. What is seen as fair and just in one time and place may not be in another. Read more

This plays particularly hard when deciding where to draw the line on free speech. Many views started as extremist before they became mainstream. For instance women used to be suppressed for advocating for the vote, Nelson Mandella was labelled a terrorist. Drawing a line when it is constantly moving is hard but extremely important.

03. Machine learning

Frontier Technology

Machine learning (ML) is the scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without using explicit instructions, relying on patterns and inference instead. It is seen as a subset of artificial intelligence. Read more

The current way many platforms believe that they will contain this extremist content at scale is through using machine learning, rather than human moderators.

04. Tragedy of the commons

Capital incentives

The concept states that in a system where a common resource is shared, with no individual responsible for the wellbeing of the resource, it will tend to be depleted over time. Read more

This is a hotly debated concept and there are many examples of commons which have specific usage rules which enable them to flourish. Assuming the platforms do not feel responsible for the wellbeing of the content, they quickly fall into this category.

05. The Lindy effect

Capital incentives

The Lindy Effect refers to the life expectancy of a non-perishable object or idea being related to its current lifespan. If an idea or object has lasted for X number of years, it would be expected (on average) to last another X years. Read more

In the world of fast paced online media prioritising content and views which will have the largest impact is vital for effective prioritisation.

06. Intangible economy

Capital incentives

As companies have come to invest more in intangible assets (data) than tangible ones (buildings, books etc), it’s had a big effect on business models and social division. While the bigger companies are getting bigger, smaller businesses are faltering because they struggle to get investment. Frontier companies are breaking away from the laggards and the data suggests these divisions will only widen. Read more

Rules will not only have to be considered for the large companies like Google, but also for smaller companies who might not have the resources to develop novel policing solutions, further increasing this economic gap.

07. Regulatory Capture

Capital Incentives

Regulatory capture is an economic theory that says regulatory agencies may come to be dominated by the industries or interests they are charged with regulating. The result is that the agency, which is charged with acting in the public's interest, instead acts in ways that benefit the industry it is supposed to be regulating. Read more

How does the government become informed by the world’s best experts on how to filter content at scale (i.e. the platforms that host the content) while balancing public interest?

08. Red Queen Effect

Complex systems

An evolutionary hypothesis which proposes that organisms must constantly adapt not merely to gain reproductive advantage but to survive. The name comes from the Alice in Wonderland quote from the Red Queen “Now, here, you see, it takes all the running you can do, to keep in the same place.”Read more

Also seen with co-evolutionary arms races, any mechanisms which are put in place to filter out extremist content will cause the content to evolve slightly to pass the filters, which will in turn cause the filters to evolve. Designing the system for improving the filter, rather than simply designing the filter is essential.

09. Asymmetric Information

Complex systems

Asymmetric information, occurs when one party to an economic transaction possesses greater material knowledge than the other party. For example, doctors typically know more about medical practices than their patients. Read more

The result of which causes severe inefficiencies and issues. Here the uploader will typically know the content is extremist and will be able to conceal this, putting the platform at a significant disadvantage. However, this dynamic will also play out between the government and the platforms.

10. Bottlenecks

Complex systems

In engineering, a bottleneck is a phenomenon by which the performance or capacity of an entire system is severely limited by a single component. The component is sometimes called a bottleneck point. The term is metaphorically derived from the neck of a bottle, where the flow speed of the liquid is limited by its neck. Read more

Given the excessive volume of content uploaded every minute, introducing human moderators (or even using machine learning) will likely lead to bottlenecks with content. These will need to be balanced as volumes fluctuate so not to disrupt the influencers.

11. Second Order Thinking

Productivity and performance

First-order thinking is fast and easy. It happens when we look for something that solves the immediate problem. For example, I’m hungry so let’s eat a chocolate bar. Second-order thinking is more deliberate. It is thinking in terms of interactions and time, understanding that despite our intentions our interventions often cause harm. This means thinking about the consequences of repeatedly eating a chocolate bar when you are hungry and using that to inform your decision. If you do this you’re more likely to eat something healthy. Read more

A powerful tool when thinking about any tool, but particularly important when thinking about problems which will have such deep and wide impacts on society.

More real-world problems