April 21, 2021

What’s Wrong with Tech Companies Banning People? (#250) | Julian Sanchez on the Rationally Speaking Podcast with Julia Galef

Check out on the Rationally Speaking Podcast Page and Episode Notes

Key Takeaways

• Julian Sanchez is fine with tech companies de-platforming users using racial slurs or other kinds of harmful speech
  • ”I don’t think anyone ought to be forced to republish or amplify content they find loathsome” Julian Sanchez
• Three main factors that affect how concerned we should be about de-platforming
  • We should be much more concerned with banning decisions “lower in the stack”
    • An Internet Service Provider banning an application is a stronger reason for concern than a social site banning a user
  • We should worry more about de-platforming for “misinformation” than for inciting violence
    • In general, social media sites should not be arbiters of truth
    • There may be exceptions when clearly false information can cause direct harm
  • We have fewer reasons to worry if we believe that “the market will solve”
    • If platforms start being too strict on speech restriction, they will cause their users to move to other platforms

Intro

• Julian Sanchez (@normative) is a senior fellow at the Cato Institute, a policy analyst, and a journalist covering the intersection of privacy, technology, and politics
  • Julian is a founding editor of the policy blog Just Security
  • Check out Julian’s personal website
• Host: Julia Galef (@juliagalef) is a writer, speaker, and the President of the Center for Applied Rationality
• In this chat, Julian Sanchez discusses tech companies banning users and highlights the three factors that affect how concerned we should be about it

Banning Users: a Legal Perspective

• From a legal standpoint, tech companies are private and have the right to decide who to host on their platforms
  • Legally, it is not a violation of the First Amendment which only restricts the government
• The ability to legally penalize speech is very low
  • Inciting violence and fraud are the few exceptions of criminally penalized speech
• Society uses a variety of mechanisms to push harmful speech to the side

Why Julian is Fine with Social Media’s Ability to Ban Users

• Julian is fine with tech companies de-platforming users using racial slurs or other kinds of harmful speech
  • ”I don’t think anyone ought to be forced to republish or amplify content they find loathsome” Julian Sanchez
• These decisions influence the kind of platform the company is creating
  • For example, many social media sites do not allow pornographic images
    • There can be other sites where you find that content
  • Each platform can shape the environment it creates
• “The barring of a particular individual from a platform is not at all the same thing as the erasure of their ideas from that platform” Julian Sanchez
  • Trump may be banned but many of his ideas are still circulating on Twitter

Generally, The Market Will Solve

• De-platforming becomes more problematic when you have one company owning the most market share
  • In a sense, it’s difficult to compare market shares among different social media platforms, in terms of the ability to express political claims
  • In this sense, Twitter and Facebook are not monopolies
    • There are many other ways to express your political preference
    • Traditional newspapers, podcasts, youtube all end up on social media feeds
    • The percentage of Americans on Twitter is not that high
• The dominance of these platforms is not set in stone
• If platforms start being too strict on speech restriction, they will cause their users to move to other platforms
  • People think that network effects are too strong for social media companies to fall
    • Friendster and Myspace are both clear examples
  • Other platforms for online discourse have been growing tremendously (Parler, Discord, Twitch)
• Is there a threshold (say the % of people using a platform) after which the network effects become too hard to displace?
  • Looking at the number of users is not necessarily a good metric
  • You can have many users who are mostly inactive on a platform

Lower vs. Higher in the OSI Stack 

• The OSI model describes the communication functions of computing systems on different layers
  • The lowest level is represented by the physical wires that make internet connection possible
  • Higher levels are the TCP/IP and HTTP protocols 
  • The applications you run on your PC are among the highest in the stack
• Julian would be much more concerned with banning decisions “lower in the stack”
  • If an Internet Service Provider banned a whole category of applications, it would be a stronger reason for concern
    • Changing broadband provider is much more cumbersome than signing up for a new social site
• Amazon’s Web Service decision to not host Parler is lower in the stack
  • It deserves more scrutiny, as it affects every user on the platform
  • Even then this decision by Amazon doesn’t prevent Parler from existing

Inciting Violence vs. “False” Speech

• It’s understandable that platforms may want to de-platform users who incite violence
• De-platforming becomes more troublesome when they ban what they deem as “false speech”
  • At a time Youtube was going to ban misinformation about COVID vaccines
  • They defined misinformation as anything that disagreed with the WHO
  • This is dangerous troublesome because we’ve seen evidence of the WHO claiming debatable statements (e.g. ineffectiveness of masks)
• Platforms should be a lot more careful at restaining information based on what’s deemed true
  • We often don’t know absolute truths with certainty
  • The possibility of raising counter-arguments allows us to test our position
    • If all counter-arguments are banned we can’t be as confident of a statement
• There might be instances where it is appropriate for them to ban content that is clearly false and harmful to people
  • If someone touted drinking bleach as a cure for COVID, it could cause many people to poison themselves
    • We would want platforms to minimize the reach of that kind of content
  • The issue of election fraud was certainly more nuanced, but it also incited violence
    • People have had the chance to present their evidence for fraud, but ultimately it didn’t turn out to be compelling

Additional Notes

• Trump’s “meritless claims” about electoral fraud were obviously harmful
  • In addition to not being true, they were inciting people to use force
  • There are many other ways that Trump can make his voice heard
• Julian thinks about speech and heroin in similar ways
  • They ought to be legal, but not necessarily that easily available

Rationally Speaking Podcast : freedom of speech, social media, tech companies
Notes By Giorgio Parlato