| Home - Topics - Papers - Theses - Blog - CV - Photos - Funny |
Hillary Clinton’s electoral college defeat has provoked many questions about how so many Americans could have choosen a misogynistic, xenophobic, pathologically lying, authoritarian ideologue as their next President – and one of many targets of their blame is social media. Compounding its polarizing “echo chamber” effect, online discussions were tarnished with deliberate misinformation, including incorrect instructions on how to vote. But is the solution for social media companies like Facebook to ``do a better job’’ at curating the world’s content to suppress out false or hateful speech?
Asking technology companies to serve as the world’s arbiters of truth and civility is not only counterproductive but just as profoundly dangerous as the social maladies that social media creates or exacerbates. We need to rethink social media from the ground up to make it more open and transparent, not less – and most importantly, to infuse online debate with some semblance of democratic legitimacy. The purpose of this blog post is not to dwell on the many flaws of social media but to explore how we could get it better.
What would an ethical, secure, and democratic alternative to either conventional ``elitist’’ media or the polarizing echo chamber of social media look like? It would need to be built on a foundation of equality among all real people; it would ensure proportional speech and accurately expose biases; it would counteract polarization by ensuring that all participants are exposed to all viewpoints; and it would provide strong but proportionate and abuse-resistant protections for anonymity and free speech before the ballot box, in order to head off unpleasant surprises being revealed only at the ballot box.
When a story or claim appears with 1,000 ``likes’’, how do we know whether 1,000 real people liked the story, or only one real troll and his 999 sock puppet accounts?
Doesn’t mean all people are equally smart, equally capable, equally right, or should end up having equal power or influence in the end. It means all people should be given an equal starting point in apportioning power, and be free to ``become unequal’’ as their words and actions are subject to the review of their peers.
Need to be able to quantify the number of people
Need a mechanism to ensure that these are real people we are counting, not fake accounts created by whoever has the most time, money, or determination.
Content-based analysis versus account validation. Content-based analysis has often proven successful, but relies on complex algorithms that someone must keep updating (and making still more complex) to keep up with the ``arms race’’, as attackers develop more sophisticated algorithms to farm fake accounts that remain under the detection radar. Relying on complex algorithms that need to adapt constantly creates inherent risks that they will become opaque and vulnerable to accidental bias or deliberate attacks, such as the secret inclusion of “back doors” that produces false positives designed to close the accounts of and hence disenfranchise populations disfavored by the attacker. While I believe content-based analysis plays an important scientific role, it is unsuitable as a foundation for ensuring the equality of real people in online public discourse.
Need an objective way to avoid bias, or clearly and quantitatively expose it when it exists.
…like wearing blinders…
Pressuring Mark Zuckerberg to deploy Facebook’s minions to do a better job of censoring fake or misleading social media posts does not solve the problem, but instead shifts the power to decide between ultimate truth and falsehood to precisely the place it does not belong: to the anonymous employees and proprietary algorithms of corporations that are accountable to no one but their shareholders.
All relevant algorithms need to be public and accountable. Algorithms need to be as simple as possible but no simpler: in particular, not so simplistic as to ignore the many lessons from past failures.
A fake or misleading claim on social media, as soon as it starts collecting nontrivial interest, needs to become inextricably attached to the counter-claims that debunk it. These counter-claims must prominently include accurate, quantitative evidence that the debunking experts are trustworthy and not [paid lobbyists masquerading as scientists](Ebell guy) – together with accurate quantitative measures of the number of people (real people) who support the false statement despite its being debunked.
Even when people are willing to hear viewpoints they don’t like, there is a matter of scalability: everyone simply does not have time to listen to everything everyone else says. Speech needs to be filtered or summarized.
one of many explanations, possibly contributing but probably not only factor… other explanations include normal polling errors, hacking of the election system, the FBI, …
Some Trump supporters are certainly racist and xenophobic; most of them are probably not, but voted Trump for other reasons. In order to protect our civilization, we the people need to be fully aware of size and malignancy of any malignancy affecting it: you do not beat a cancerous tumor by ignoring it or sanitizing the doctor’s diagnosis. We need know the true, complex set of reasons that people are inclined to support someone like Trump
Need regular opportunities for speech ensuring that all voices are represented, and building anger over policies actually have a chance to influence policy before that anger explodes in a self-defeating electoral choice of an authoritarian con-artist and places freedom and democracy in existential risk.
| Bryan Ford |