Thread by @AlecMuffett

I am somewhat horrified by this; here in the privileged west we are openly amazed at countries which "turn off the internet when there are school exams", yet our key infosec luminaries propose rate limiting as acceptable mitigations towards "fake news"

As claimed by one proponent, this paper purports to take us beyond speculation of how rate-limiting might affect online group-chats and communities; yes, but reading it I see no discussion of how online group-chats & communities will adapt to rate limiting

https://twitter.com/chinmayiarun/status/1175783953908862978

Circumstance restricts behaviour, but also behaviour adapts to circumstance; to go all Jeff Goldblum for a moment: "life, or communication, will find a way" - if you work in a platform anti-abuse team you will see it's a war of attrition as both bad-and-good people adapt.

Anti-abuse is whack-a-mole, and if we begin to accept rate limits for everyone, we constrain all expression for the benefit of stopping a small fraction.

Similarly I've spoken with people @Twitter who had to explain that the very calls for limits to "organic reach" which <some activists> were calling for, would literally destroy those activists' own campaigns.

@Twitter I offer this in hope that @gvrkiran and other people calling for "rate limits" will revise their work to include the negative consequences - to look in the mirror and ask "if we do this, whom else will we impact and what would the cost be?"

Because they're not doing that yet.

@Twitter @gvrkiran links: bbc.co.uk/news/world-afr…

@Twitter @gvrkiran links: researchgate.net/publication/33…

@Twitter @gvrkiran ps: if I was going to make a bet, it'd be that rate-limiting social networks will give rise to an additional pressure: smaller chat-groups. This would be to combat the rise of "Sibyl" groups that bypass rate limiting by morphing messages and resharing them under separate identity

@Twitter @gvrkiran The "cheap and easy" solution to this would be to restrict group sizes, so that the size of Sibyl groups stays beneath the rate-limiting bound.

So, you wanna get into your school's parent-teacher chat group? Sorry, it's too big, because of fake news in India.

@Twitter @gvrkiran The amusingly ironic cyber-dystopia-flipside to this, of course, is that the smaller groups will contain 1 human being and 29 sibyl-bots, all talking to the 1 victim and radicalising them without any other human intervention. I am sure @nealstephenson must have done this already.

@Twitter @gvrkiran @nealstephenson Presumably there will even be software written by "bad actors" — probably advertisers selling fake sunglasses — which will watch for "<Parent has left the group>"-messages and will subscribe their sibyl-friends to fill the vacant slot, to gain more "leverage"

@Twitter @gvrkiran @nealstephenson Eventually there will be small groups entirely populated by Sibyls and bots, sending nothing but hate-messages to each other, which mean nothing until a Journalist or Academic somehow stumbles across it and writes a clickbait exposé about "Nazi Bots" and demands "action".

@Twitter @gvrkiran @nealstephenson Whimsy aside, though, I am mostly worried about this. Bad speech is best combatted by good speech, but there are no "good speech" sibyl networks, so it's down to individuals.

We should try to fix the people, and that can't be done with software.

https://twitter.com/AlecMuffett/status/1176044721367658496

@Twitter @gvrkiran @nealstephenson Elsewhere in "…we should not allow bad people to be able to communicate with others en-masse, wait, no, I didn't mean that you should censor _me_ — I'm a good person!"

src: reddit.com/r/geek/comment…