Inkwell: Authors and Artists

    

It was the Left who were supposed to become the Techlash contingent,
but the loathsome Senators Cruz and Hawley immediately grabbed that
torch away from them, and these two coup-meisters are so dangerous
and disgusting that nobody is gonna want to help them out.

So instead of losing Section 230, a goal that Trump was ready to
defund the Pentagon to achieve, we're gonna get some kinda
double-down Section 460, a digital new-deal that pretends to
regulate the tech majors but also formalizes their power.  

I don't think that Big Tech wanted this, they'd much prefer to be
left alone to mint cash in their surveillance silos, but Trump's
bloody radicalism has forced their hands. Either they kiss, make up
and divvy up the turf, or they're gonna get pecked to death by
loony-right QAnon fascist squadrons.  So their time has come to
stand up; it's pretty much them or nobody.
  

    

Also, it's not enough the Trumpista gang have been hacked and
suddenly fallen silent; some of them are gonna have to go to jail. 
I don't know who's gonna hound them and how, but there's gonna be a
lot of sticks.  And, also, I'd be guessing, some pretty big carrots.
These new Dukes of Cyberspace will need to shower some of their gold
on the restive population.   They need to make the new dispensation
look likable because, considered objectively, it's like 1984 on
steroids.

This development seems kinda sudden, but it's one of the least weird
political developments I've seen in a long time.  It's a different
ruling class, but it's a reasonable and plausible ruling class that
goes where the money, power and brains are, so who knows, just maybe
they'll actually be able to govern the USA.
  

    
How does that jibe with the assertion that surveillance is BS? The
technocracy is all about surveillance. 
  

    <scribbled by jonl Sat 9 Jan 21 08:15>
  

    
<echodog> I'm not sure how that relates to the previous posts. I
don't think <bruces> is suggesting that Big Tech won't continue to
"mint cash in their surveillance silos," or that the surveillance is
any more okay that it was before. 
  

    
> but there's gonna be a lot of sticks.  And, also, I'd be guessing,
some pretty big carrots. These new Dukes of Cyberspace will need to
shower some of their gold on the restive population.   They need to
make the new dispensation look likable because, considered
objectively, it's like 1984 on steroids.



This has been underway within YouTube for quite some time. The
strange inevitability of youtube's machine-learning algorithms to
channel "next video" rabbit holes towards strange homemade
conspiratorial/holocaust-denialism/white nationalist monologues has
been a puzzle and a challenge for their team for several years. 

In 2019 they took steps to ban holocaust denial videos outright, but
the homebrew radicalization original content of Q-anon and its
related factions has been driving massive levels of engagement (i.e.
revenue) on their site, putting them in a bind over whether to kill
the golden goose, or wait for it to kill them. Because original
content earns higher levels of engagement, the YouTube sales teams
can  target advertising content more narrowly and charge more for
it. Throughout 2020 they've been trying different ways of throttling
down on traffic flow towards the unhinged sector, and encouraging
traffic towards more "legitimate" institutions, though it's
definitely cutting into their profit margins a bit. 

Anyone who's able to generate decent (and non-radicalizing) original
content that earns engagement AND partner with credible institutions
(universities, large businesses, media companies, NGOs, government
agencies, etc) will likely benefit massively from YouTube's efforts
to cultivate non-crazypants moneymaker spaces where they can sell
ad-space without having to get called into senate hearings every six
months.
  

    
[<jonl> reminded me this morning that I haven't participated here yet,
though I've been following it; the below is adapted from messages I sent to
a private mailing list that Jon and I are both on; I've edited them to
remove material from others on the list who haven't consented to sharing it]

==

Appl and Goog have removed Parler for not having an effective hate-speech
filter.

Do we think there's a "hate speech filter" that:

a) filters a meaningful fraction of hate speech, and

b) does not filter legitimate speech?

Given that we can't make a nudity filter that works (hi, Tumblr) and given
that nudity is a far less contested category that "hate speech," how would
that work?

By all means, if you think Apple and Goog should be evicting Parler, then
take that position. But the idea that Parler's failing is their
unwillingness to invest in a nonexistent technology is an obvious pretense
on Appl/Goog's part.

Some test cases for a software hate-speech filter:

Which of these is hate-speech?

* "The rioters were wearing shirts with an English translation of the motto
over the gates of Auschwitz, 'Arbeit macht frei."

* "We should send you Jew-lovers to Auschwitz where you'll learn the meaning
of 'arbeit macht frei'!"

* "You call me a 'Jew-lover' as though that is a source of shame. Of course
I'm a Jew-lover. After all, I am a Jew, and I love myself."

--

Which of these is hate speech?

* "The Trumpist mob that beset the Black woman in the streets of downtown
Los Angeles used numerous racial slurs, including [racial slur]."

* "The Dems have built an unstoppable coalition backed by [racial slur]s."

* "The senator's account was terminated after it was revealed that he had
posted numerous racially inflammatory statements to the service, including
'The Dems have built an unstoppable coalition backed by [racial slur]s'."

--

Recall that Sarah Jeong asked Jigsaw's leading sentiment analysis tool to
evaluate the hate speech she'd been subjected to, and it identified the
following statement as having the lowest possible rating for aggression:

"I’m going to rip each one of her hairs out and twist her tits clear off"

https://www.wired.com/2016/09/inside-googles-internet-justice-league-ai-
powered-war-trolls/

==

So sure, condemn Parler, condemn Apple and Google for including them in the
app stores, but please let's not pretend that "hate speech filters" exist as
anything but grifty promises from overcapitalised snake-oil salesmen
flogging their magic beans.
  

    

Someone reading the above might say that hate-speech filters are imperfect,
but they could act as an early warning signal, alerting human moderators who
can exercise judgment over speech.

That may be true, but there's not much evidence of it. Facebook and Twitter
notoriously and routinely overblock the TARGETS of hate speech for repeating
the things they've been called, or just for venting about the people who've
targeted them) and let obvious hate-speech through.

The lack of a crisp definition of "hate speech" is a big part of the
problem. For all intents and purposes "hate speech" is indistinguishable
from "speech that is almost-but-not-quite-hate-speech."

That means that if the platforms publish a definition of "hate speech," the
most dedicated producers of hate speech will produce an endless torrent of
almost-but-not-quite-hate-speech, which will be experienced as hate speech
by its targets.

Worse: dedicated trolls will use their encylopedic knowledge of hate speech
rules to goad their targets into uttering hate-speech and then demand that
their targets be removed from the system.

This is not hypothetical: it's how the state sponsored trolls in Cambodia
operate, and they have used it to completely neutralize the opposition that
mounted a nearly successful election campaign in 2013.

https://www.buzzfeednews.com/article/meghara/facebook-cambodia-democracy

There's some evidence that Cambodia's state trolls are actual private
contractors, which means that you can buy goad-hate-speech-targets-into-ToS-
violations as a service.

So what if the tech platforms instructed their moderators to remove things
that fit the hate speech definition AND anything that was *almost* hate
speech?

We know what that looks like: The "I know it when I see it" standard,
combined with huge numbers of moderators with a range of sensibilities and
common sense, leads to things like a World of Warcraft moderator shutting
down an LGBTQ-friendly guild for "provoking hate speech" by "inflammatorily"
advertising that its members are queer and
allies.

https://arstechnica.com/uncategorized/2006/02/6129-2/

This isn't a counsel of despair. We can have speech policies that work, but
they work at the scale of a community, not a global service.

BTW, If you're looking for a set of moderation principles written by
marginalized people who have been the targets of hate-speech, check out the
Santa Clara Principles, which articulate a human rights framework for
content removal and moderation:

https://santaclaraprinciples.org/
  

    
Does that mean that Goog and Appl shouldn't have taken action against
Parler? Not at all.

I am all for platforms (including app stores) having a variety of speech
policies. After all, I expect different speech standards when I'm tucking my
daughter in at night, when I'm in a professional meeting, when I'm having a
conversation around a campfire, and when I'm in a political debate. I want
to have a variety of conversational spaces that I can choose among based on
my preferences about the suitability of the house rules to the context of
the discourse I want to have.

The problem with the app stores is that a shameful, four decade, bipartisan
neglect of antitrust enforcement has contracted the possible universe of
speech policies for mobile apps into two hands, and neither store has been a
good steward of that power. Both routinely block apps for stupid reasons. I
don't expect Appl or Goog to stop making mistakes, and I am not confident
that they'll make fewer mistakes.

Rather than focusing on making Appl and Goog more successful at content
removal (something I'm in no position to do), I want to focus on making
their failures less consequential, by encouraging competition in app stores
and by legalizing adversarial interoperability, so that members of the
public who are badly served by Appl and Goog's speech policies have a remedy
apart from "complaining" and "suing" -- that remedy is "choosing someone
else's moderation policies," which, in a technical
sense, generally means "sideloading."

Sideloading (for Appl devices) requires legalizing violations of terms of
service (CFAA), bypassing TPMs (DMCA 1201), reimplementing APIs (Oracle v
Google*), and other legally fraught activities.

https://www.eff.org/deeplinks/2019/10/adversarial-interoperability

Like the pandemic, Parler is revealing the latent fragility of our systems.
Parler is vulnerable to takedown by a duopoly of app stores, by a oligopoly
of cloud providers, by a tiny coterie of payment processors.

I hate Parler and would be delighted to see it consigned to the dustbin of
history, but I know that my political adversaries (the entities behind
Parler) are perfectly capable of bringing the same pressures to bear against
services I like - app stores already routinely block apps for stupid
reasons, such as alleged "progressive bias."

It's reasonable to say that Parler should use a mix of automated tools and
human judgment to block unlawful speech and/or hate speech - but it's NOT
the reason that Appl/Goog have removed it. Both companies have made it clear
in their public communications that Parler must install an effective hate-
speech filter to qualify for inclusion in their stores.

I hope we can all agree that "effective hate-speech filter" is a nonexistent
technology, and that therefore none of the existing services whose apps
appear in either app store have such a technology in place.

I don't think this means the companies should do nothing. If Apple and
Google want to remove Parler because they think it's terrible, they should
say so: "We have removed Parler because we think it is a politically suspect
project intended to foment violence. That judgment is primarily subjective
and may be wielded against other
platforms in future. If you don't like our judgment, you shouldn't use our
app store."

I'm 100% OK with that: first, because it is honest; and second, because it
invites the question, "How do we switch app stores?"

Cory
  

    
Does Trump's threat to repeal Section 230 make sense? If that were
done the big media companies would immediately evict Trump and his
boosters to keep them from posting things that would cause the
companies to be sued.
  

    
Only if you don't mind shutting down the WELL the same day. There definitely
isn't enough money in this shoestring op to let people post stuff if The
WELL is to be held liable for what they post. But Facebook'll be fine - they
have more mods than then entire headcount at Twitter.
  

    
I wouldn't worry about the Well. Postings are mainly only readable
by its members; and it doesn't have enough members for anyone in
authority to be concerned with. 
  

    
You've misunderstood the nature of the risk: CDA230 insulates platforms from
civil liability - that is, a pissed off WELL user suing the WELL over
something another WELL user said. If you don't think WELL users are capable
of getting angry enough at one another to sue each other, you need to read
some different forums.
  

    
I understand your point, but I judge the risk to be vanishingly
small.
  

    
It's silly to make this about the Well.
  

    
Of course it's about the WELL. It's about every single forum for online
speech, from Kickstarter to Yelp, from The WELL to your PTA mailing list.
They would literally be unable to function if they were jointly liable for
their users' speech.

As to "vanishingly small" - WELL users have REPEATEDLY threatened legal
acation agains one another over things they said to one another (some have
even served notice on one another). Do you think that the service has
undergone some transformation that has made the users of this service less
prone to anger at one another? How would (or could) the WELL vet potential
new signups to determine whehter they would name the service as a party to
any suits in the event that another WELL user made them angry enough to sue?
  

    
Donald Trumo isn't demanding repeal to strike out at the WeLL. Who
put that bug in his ear, and why? It would be bad for him
personally.
  

    
He probably doesn't have a card good for $15, or he could check it
out himself. Maybe he can get a pre-paid?
  

    
Cory is right. Well members have done some memorably reprehensible things to
each other over the years, just not lately.
  

    
CDA230 is the legal underpinning that makes the Well's business
possible.
 
I don't have a management role here anymore, but I can't begin to
imagine the scramble to find out if private insurance companies
would step up and how much they would charge. Managers and owners
being able to roll their eyes at assertions of libel makes all
forum-conversation space viable. 
  

    
It looks like Amazon has a bigger (if temporary) ban-hammer than the
other big tech firms, because Parler runs on AWS. Not for long.

Amazon is kicking Parler off its web hosting service
Parler’s CEO said the site could be offline for up to a week
https://www.theverge.com/2021/1/9/22222637/amazon-workers-aws-stop-hosting-ser
vices-parler-capitol-violence

So if they want to stay up, they will need to find another hosting
service and a CDN big enough to absorb the inevitable
denial-of-service attacks. I wonder if Cloudflare would do it?

But I wonder a bit why Trump's minions didn't have a blog ready to
go. His fans would find it and syndicate it everywhere for him, and
the press would feel obligated to read it and report on it.
  

    
I'd say it's dead, not just off for a week.  Migrating from AWS to
another host (if one can be found in the US) will be a major
operation.
  

    
Portability is possible but it depends very much on the details of
their software stack. Parler's CEO claims they planned ahead, but
who knows, maybe it's just talk.

re: <74> I don't see why big tech would need a Zoom call. All they
needed was a Schelling point, and they got a big, obvious one. They
don't need to coordinate any further, any more than airlines need to
coordinate privately to match prices. The info they need is public.

Also, I'm thinking all the bosses had to do was finally give in to
the activist contingent of their mostly-liberal employees who have
been advocating for this sort of thing since before Trump got in.
It's probably pretty good for morale, overall, and after being under
siege (culturally) for years, they could use a morale-booster.
(Their minority of closeted Trumpist employees might be glum about
it, though.)
  

    
Yeah, it's very much about the WELL, or at least, what history has
done with the legacy of the WELL. The Electronic Frontier's just not
a frontier now, it's densely settled, it's got all kinds of wealth
and infrastructure to quarrel over, and it's got a blooming plethora
of economic, legal, social and ethical problems.

I'm not that upset about it.  Problems are inherent in the human
condition.  It's good that we're recognizing that the richest
companies in the world really are the richest and most influential
enterprises in the world and not kinda purple-haze cyberspace ivory
towers.  It's realpolitik and it's much healthier than QAnon
delusion.

The WELL really was a purple-haze cyberspace ivory tower for quite a
while, I can fondly remember how thrilling that was, but nowadays it
is what it is, which is a funky little mom-and-pop legacy-media
niche.  That's what the passage of time does, it's the nature of
history and futurity, there's a melancholy beauty to it.  You
shouldn't whine about it any more than you ought to wring your hands
about the Second Law of Thermodynamics.
  

    
Meanwhile, it's snowing like crazy in the country of the Tiny
Spaniard.  A major-league Greenhouse blizzard has blown in and
people are skiing on the snowdrifts in Madrid.