Self-conception: new media, new place for free expression. Twitter’s GC describes Twitter as the free speech wing of the free speech party. “The tweets must flow.” (Perhaps unintended reference to “the spice must flow,” because that’s not actually a great comparison.) The new town square.
Paper interviewed GC of Tumblr, Wordpress, some lawyers at Twitter, Facebook, Pinterest, Yelp, etc. Lots of litigation at Yelp over unmasking reviewers, which they treat like confidential sources; church/state divide between reviews and ads, influenced by newspaper history.
Three influences today that SCt watching wouldn’t tell you about—regulation of speech outside courts. (1) Power of private gatekeepers. You can report any tweet for spam, violation of Twitter’s rules. (2) International norms. FB says there’s one FB globally. 80% of users are not American. They think in terms of free expression generally, not the 1A. (3) Congressional statutes: §230 and §512. Tumblr: NYT v. Sullivan is hilariously irrelevant; to him §230 of the CDA is the greatest thing Congress has done in 2 decades, and is why internet companies can flourish.
Not saying these lawyers are perfect. In corporate interest to advocate for free speech in certain circumstances. Isn’t saying that newspapers don’t matter, or that courts don’t matter. Self-concept as speech lawyers does matter, though.
NYT has 29 million views/month; Twitter has 200 million users; many NYT visits come from Google or FB. WordPress has 70 million users and many more readers. Tumblr, likewise. Massive platforms. NYT is $2 billion company, Google is $400 billion company. Washington Post is 16,000x smaller than FB. Free expression lawyers have markets in the new media.
Tend to be platforms for others’ speech, not the speech of their own. (Interesting resonance with Crawford’s piece.) Tend to see themselves as fighting for their users. Google’s mission to organize world’s info, a speech based mission. WordPress: mission is to democratize publishing. Like NYT, have dual classes of stock; shareholders have to trust the owners. Lawyers come from community of people who’ve thought about keeping the internet free (Larry Lessig, Yochai Benkler).
“Collateral censorship”: the idea that the gov’t can make a phone call and instead of getting a court order, get the content removed. Wikileaks is the prime example. PayPal, Amazon, and others received pressure to dissociate themselves. Huge issue. But there are folks at these companies who think more about free expression than PayPal or Amazon.
FB’s global platform: doesn’t believe that FB’s rules governing speech are that different from the rules in our daily lives—the US can’t throw you in jail for hate speech, but we have a lot of social pressures/ability to remove people from private property when they engage in hate speech. People at a funeral might not even see the Westboro protesters, because they were so far away; the plaintiff father didn’t even see them. So FB is much like the US in actuality—we don’t see very much hate speech in our daily lives. (Um. What do you mean we, white man?) And it’s not that hard to find porn/hate speech online if you want it.
Other countries regulate hate speech very differently, and lawyers have to find the right policy, including whether they even go into a particular country, e.g., Google in China. Twitter is proud of policy: take down content of tweet only in the country in which there is a legal order.
Congressional power: importance of industry structure for free speech environment. #1 analogy to NYT v. Sullivan is rules that make ISPs not liable for the speech of their users. Sullivan is about NYT as intermediary: needed a standard to make intermediary unafraid to publish. YouTube gets 100 hours of video/minute. Can’t screen all for libel. §230 is greatest thing since sliced bread to these lawyers.
Marjorie Heins: Somewhat less sanguine view of FB’s power with its terms of service. FB says you won’t post hate speech, threatening content, or pornographic; incites violence; contains nudity or graphic or gratuitous violence. Some of this is unprotected by the 1A, others not. Even within the unprotected categories, there’s no judicial determination, just a best guess. We don’t have access to internal guidelines, which are interpreted by internal censors. Appeals process? Mysterious at best. Controversy over a nude statue in a Kansas public park. ACLU posted a photo with an article about the controversy on its FB page. ACLU made many attempts to communicate with FB; its status allowed it to get the picture reinstated. But during period of newsworthiness, relevant information was not available, and many people can’t get appeals. Nat’l Coalition Against Censorship: web archive, the File Room, of censorship around the world. Here, photo by Nan Golden, prominent documentary photographer; domain name host for NCAC decided they didn’t like the picture and ejected it from the domain. Were able to find another provider, but there’s no other FB now. Unpredictable censorship.
FB isn’t a common carrier and probably has its own 1A rights to censor whatever it wants. Might say the same about search engines. But people have to know what’s being censored/filtered from their search engines, and there is a distinction between social media and search engine that is basically a sophisticated mechanical tool. UK customized Google search: you accept Google ads and have to comply with ToS, which means no content Google doesn’t like—porn, hate related, violent, or offensive content. Google also censors your general searches by default (SafeSearch). Now Google makes it impossible/difficult to disable, though you can get around it by typing in sufficiently specific terms (e.g., “porn”); without such terms you won’t know what you’re missing.
Also, DMCA §512 is a big hole in §230. While §230 is indeed wonderful, the IP hole is a problem. Congress’s gift to the IP industries; machine-generated takedowns by the 100s of thousands. 10-day period of suppression follows even if there’s a valid defense. And non-copyright IP is not covered at all by §512.
What’s left of NYT v. Sullivan? There are other arbiters now. The current generation of lawyers, whatever struggles they have in France or with their own ToS, they are inspired by the spirit of Sullivan. You should read these cases because they are part of a literature and part of our culture. But it has its weaknesses, as Black and Douglas pointed out in concurrence: actual malice test invites inquiries. Can honor spirit of Sullivan while improving on it—state legislatures can make liability for defamation of public officials harder; social media can make clear that they won’t censor criticism of gov’t even if potentially defamatory; even extend that to public figures. Also need to know what our gov’t is doing to criticize it—secret gov’t torture, drone strikes, surveillance, etc. Social media could play a positive role by trimming censorship and fostering robust access to necessary info.
Jonathan Zittrain: in 50 years we may still be getting our info from Twitter and Google, in which case what we venerate will be what they tell us to venerate. Distance between Ammori and Hein is important. Ammori mentioned ability to censor; Hein was more concerned.
Models for mass communication: apocryphal 1789 model, when public can communicate among itself in bulk with public square/commons/gov’t spaces where we’ve come to understand there are great restrictions on what gov’t can do to regulate speech. Hyde Park model of unfettered speech. 1964 model: NYT as the speaker/source of news going out to public—directional arrow, and NYT reached more people than someone on a commons. Occasionally, the public could use NYT as a megaphone if it could convince NYT it was worthy of inclusion as letter or op-ed, or could write them a check to run an ad. But nobody doubted the kind of censorship/editing that the NYT would do, so it’s not really the gov’t space but a very different mode of broadcast.
These differences help us understand today’s mode: public talking to the public w/out a trusted intermediary providing content, merely routing it. No longer unidirectional. Structural similarity to 1789, but in the middle is not the gov’t but a private entity entitled to censor whatever it likes. Hein’s danger; Ammori’s answer may be that it’s not a real danger. Inability to be arrested for outlandish speech is interesting, but your ability to be exposed to that outlandish speech is more like 1964 than 1789. Also, while Google, FB and Twitter retain censorship rights, the volume is so great and the self-concept is not as provider, they don’t censor that much. (Note that NYT can also try to get your attention via Google, FB, and Twitter, which was what made Anderson unhappy. Need for good and trusted sources answering to a muse other than Mammon or complete idiosyncrasy.)
Twitter may be the least problematic for Hein because Twitter lets everything in—the tweets are what people I follow are saying. At most, Twitter suggests people to follow. Antediluvian phase, with less SEO or twitterbombing than you might think when you search a term. Geographical zoning with all of these intermediaries, but all different. Twitter: tweet withheld in your country. This is seen as better than deleting for all/for none. Only problem: Makes it so easy on countries to censor for themselves; replicates censorious structures around the world. Plus, you can also change your country settings to read the tweet. Total fig leaf. Only functions as symbolism, which may be that all Europe wants for hate speech. The other thing: small fences can keep in large mammals—may deter many from seeing the tweet.
The more this is abused, the more AI will help you filter. And this is what FB does. Secret sauce, unknown to us. Opacity is very different from is it censored or is it not, but also just as powerful (and can be affected by commercial interests). FB and others in the middle will want to exercise more control for your own good—FB apps end up asking for much more private data than they actually need. He doesn’t want 1789 or 1964 to be the baseline; he wants the baseline to be our capacities now. Lock-in exercises a subtle power.
Also worth noting how ephemeral this speech is. NYT in 1964 can be retrieved; Twitter keeps tweets for 7 days. For public sphere purposes, that’s pretty bad. There are other sites to search old tweets, contingent on access to Twitter API. We’ve studied link rot in SCt opinions—50% of links from 1995 to present don’t work any more. 72% of links in Harvard Law Review don’t work any more. Dead links are unlikely to resurrect, and live links may die, so that’s a problem. Perma.cc—designed to create permanent citation link. Built and run by libraries, according to noncommercial values—but needs to be defended against various kinds of attacks. Ammori says: you shouldn’t need links, only Google. He was surprised by that level of trust in one intermediary—not just preserved, but searchable/first result forever? If enough people can downvote a tweet and get rid of it, the private organization seems worrisome.
Private repositories of content formerly held by public organizations: every single time the word “kindle” appears in War and Peace, “Nook” has been substituted in the Nook ebook.
Scary: 2011 study found that Zappos reviews weren’t persuasive without grammar and spelling, so Zappos just went in and fixed them using Mechanical Turk, and therefore improved demand for the product. That’s not usually how we think of improvement in quality. The improvement makes the reviews less informative/less truthful, if you believe that the grammar does matter.
FB’s election experiment: randomly chose to insert in some people’s feed the fact that their friends had voted. People whose friends said they voted were more likely to vote themselves. Is there any problem with Mark Zuckerberg saying that he’ll do this only for Democrats? Not censorship, but control over the conduit of user-generated speech made possible through a trade secret algorithm. 1A freedom of association? All Z. can think of is to require honesty, but the contract will always reserve the right to do this kind of thing. De facto, very different speech environment.
We’re not just talking about speakers and readers; the book reads you. What happens when you can actually watch people watching movie/reading book; that data is now being gathered when you read a Kindle. Textbooks are trying to watch students. Microsoft applied for patent to allow Xbox Kinect to watch how many people are in the room. New monitoring opportunities for flow of speech, allowing chilling effects. Can tie your rants on FB to your ability to enter restaurants.
His charge to the lawyers at FB, Twitter, etc., and to us: look for ways to change how those services operate. Google News had a feature for a while that if you were quoted in a news story, you could write a comment about your quote and have it featured next to the story. Could we implement tweet retraction, which is now impossible?
Fred Friendly, president of CBS, said that television makes so much at its worst that it can’t afford to be its best. Sometimes we need institutions that don’t exist for success/profit, not just a GC of a for-profit firm.
Ammori: we do need more institutions. Why does he defend having no links in law reviews? They’re long, they’re ugly, they’re last checked years ago—Google/Bing would be better.
Will these companies get it wrong on what content to show you and what to censor? There’s a flood of information; their argument is that they’re doing their best. Alternatives to improve the situation: lobby them to improve, but hard to figure gov’t intervention that would work better. Agree with Hein on §230’s IP loophole. Companies are doing some litigation to improve the situation, but they see the problem, including the requirement to reveal identity in order to get content put back up. Restaurant can send a takedown for a bad review that has a picture in it. TM is another real loophole.
He likes SafeSearch. Tumblr without it is horrifying.
RT: drivers 2 and 3, international norms and the unique US §230, seem conceptually opposed, which is not a flaw in Ammori’s analysis but suggests something about the rhetoric of internet exceptionalism. A sort of ISP forehand/backhand: not our problem and not our responsibility whether we take something down or keep it up. Relatedly: “We rarely experience hate speech”: depends on the we, and this is connected to how the private policies are enforced (Ammori mentioned difficulties with FB’s handling of misogynist speech).
Ammori: both §230 and international norms influence companies’ response. Rely heavily on US law, but when you expand into new countries, that’s when there’s a greater role for int’l norms. Maybe trying to have it both ways, but both influential.
Twitter only blocks direct threats of its own volition; they think of what they do differently than FB. FB’s response: People use FB for activism because they know real people are there: Twitter accepts lots of anon/pseudonymous activity. (NB: FB’s real name policy is not a real name policy—danah boyd has said a lot about this. It’s a “look like you have a standard Western name” policy, and it can easily be used against activists.)