I disagree with your characterization of “kill them all” as “commonly used” or as a “term of art”. I’m assuming they use
Sentiment analysis - Wikipedia , which is another area making use of AI.
I wrote about censorship last year
Masinter's Musings: The Epidemiology of Bad Ideas
The back-off algorithm probably need not be exponential (as doubling the time delay would be) to be effective.
But I agree with your conclusion (“the IETF MUST NOT rely on social media platforms”), although for different reasons – I don’t think the IETF should rely on any communication platform where significant revenue of the platform comes from selling or using information about the preferences or behavior of the participants. But also because “open source” is more than an implementation strategy, it also provides for transparency and faster evolution.
--
https://LarryMasinter.net https://interlisp.org
From: Dean Willis <dean.willis@xxxxxxxxxxxxx>
Sent: Sunday, June 6, 2021 11:48 AM
To: LMM@xxxxxxx; emil@xxxxxxxxx
Cc: jared@xxxxxxxxxxxxxxx; ietf@xxxxxxxx
Subject: Re: Postel's Principle and Layer 9 protocol engineering
-------- Original Message --------
On Jun 6, 2021, 1:13 PM, Larry Masinter < LMM@xxxxxxx> wrote:
...
But banning someone is very different from delaying their posting
for a reasonable length of time. Especially if the algorithm used
is clear enough that participants who value the ability to
communicate without delay can readily choose other ways
of expressing their perspective, without resorting to hyperbole.
----
Larry, they didn't "delay my posting for a reasonable length of time. " They permanently blocked my comment, denied the appeal, and banned me from using Facebook for any purpose for 24 hours. As I understand it, they use an escalating scale, so the next time they make this mistake it's a week ban, then two, then a month...
Given that MY current employer (for better or worse) is an agency that literally runs on top of social media platforms, said 24 hour ban literally blocks my income stream and the business operations of my clients. Now, perhaps that's my problem and I have a business Facebook dependency to resolve, which I will do...
But my essential point is that given these sorts of risks the IETF MUST NOT rely on social media platforms -- in particular Facebook (because that's where we have an objective instance to analyze as a threat model) -- where said platform is likely to invoke automated censorship in such a way that users must self-censor commonly used terms of art in order to avoid being expelled from the platform.
Literally last month we were circulating a survey asking people about doing IETF work on social media platforms. And while we have had a few technical discussions on Facebook and other platforms, I don't think we've thought our way through the implications of working that way.
--
Dean
