Thread by @MiekeEoyang

Next up in #hewlettverify, @alexstamos, to talk about the challenges the platforms face in balancing privacy, security, and freedom.

@alexstamos Going through tradeoffs in benefits of platforms. non-monopolistic open tech with data portability vs. no cambridge analyticas.

Also tradeoffs in privacy vs. security. Can't identify/stop bad action on your platform if you can't see it.

All the effort is on the privacy side, not on security side. Ex. What's app and encryption where platform can't see, so aren't responsible.

Platforms acting in a quasi-governmental manner to regulate speech/activity, but without the legitimacy [in a rule of law sense].

Tahrir Square was the peak of platforms believing that the tools they were making were tools of liberation and would lead to more democracy, more accountability for governments.

But then successor governments were more savvy and uses tools to oppress.

Another challenge: you can't stop what you can't measure. Companies that don't use quantitative metrics fail. But measurement is about "are you happy with the product..." Not whether it's good for you (e.g. time on the platform).

But how do you define & measure what's harmful speech? Speech questions especially comedy and parody are hard to classify or draw boundaries.

@alexstamos is going through amplification model that the Russian Internet Research Agency used.

1. Create fake accounts & get ~100k followers
2. Amplify content using advertising on platform to reach ~5M people.
3. Then the people (crazy uncle) shared in organic network.

@alexstamos Now going through GRU model:
1. steal sensitive data
2. create content, release via cutouts to media.
3. through mass media reach 200M+

Also no discussion in media for doing what the GRU wanted, and debating own culpability.

@alexstamos Another example: India and Whats App amplification:

1. Send messages to extreme supporters.
2 Supporters send to their networks/groups.

Challenge: WhatsApp is encrypted, so the platform can't say what reach was or how it disseminated.

@alexstamos Alex calls on platforms to:

1) embrace transparency and
2) exercise restraint on moderation.

Meaning, explain your decisions clearly & specifically.

Transparency and specific explanations about decisions on the platforms sets precedents, but that also builds legitimacy in decisionmaking.

He argues that the more amplification that the mode of communication provides, the more content moderation you can impose. Meaning, little to no moderation one on one messaging but greater moderation in advertising.

But it depends on the harm that you're dealing with.

@alexstamos calls for innovation on the safety/privacy frontier.

Examples with an internal victim: bullying, threats, doxing, terrorist recruiting, child grooming, sextortion. *He notes that the last two are the worst things that happen on the internet every day.

@alexstamos Also need innovation to think about how to deal with abuse with an external victim.

Examples: CSAM trading, conspiracies, anti-vaxers, hate speech.

@alexstamos Calls for platforms to synchronize policies and blockade hate sites.

Notes we treat muslim terrorism and white violent hate groups differently on social media for blocking. How do you disrupt the chain before they get to violent content?

@alexstamos Calls for legislative change to enable data-based academic research, currently criminalized under our laws.

@alexstamos Calls for legislative change:

Critiques use of FBI as premier defensive cybersecurity agency, since there's no such thing as a casual conversation with the FBI.

And now questions from the audience...

@alexstamos talks about how everyone in 2016 assumed that @HillaryClinton would win and made decisions in light of that assumption, not just @Comey but platforms as well.

@alexstamos talks about how @facebook is the best position to respond because:

1) Its actions can (re)define industry standards. (which twitter can't)
2) the have the means and nimbleness to respond. (which google can't)

(Shade thrown was his, not mine.)

@timkmak asks about @alexstamos about need to develop media guidelines about how to deal with fake news.

Alex says media is an active participant in the problem, not just neutral observers, so has to grapple with what to do itself.

@NPRDina asks about Christchurch shooting video. @alexstamos says only 200 people watched live.

Follow up: analogy to stockmarket trading anomalies can be stopped real time. Could platforms do?

Christchurch problem was that only a small # saw on FBlive, more shared on 8chan.

Stamos says we need more research over how the propagation works.

But Christchurch shooter could have used other platforms, so it's not a platform specific problem.

Now taking a series of 3 questions to wrap session.

-Companies cannot operate in China and live up to their values in the west.
-Three models: US, EU, China.
-Now fighting for unaligned countries to adopt one of the three.