" /> net.wars: May 2020 Archives

« April 2020 | Main | June 2020 »

May 29, 2020

Tweeted

sbisson-parrot-49487515926_0c97364f80_o.jpgAnyone who's ever run an online forum has at some point grappled with a prolific poster who deliberately spreads division, takes over every thread of conversation, and aims for outraged attention. When your forum is a few hundred people, one alcohol-soaked obsessive bent on suggesting that anyone arguing with him should have their shoes filled with cement before being dropped into the nearest river is enormously disruptive, but the decision you make about whether to ban, admonish, or delete their postings matters only to you and your forum members. When you are a public company, your forum is several hundred million people, and the poster is a world leader...oy.

Some US Democrats have been calling Donald Trump's outrage this week over having two tweets labeled with a fact-check an attempt to distract us all from the terrible death toll of the pandemic under his watch. While this may be true, it's also true that the tweets Trump is so fiercely defending form part of a sustained effort to spread misinformation that effectively acts as voter suppression for the upcoming November election. In the 12 hours since I wrote this column, Trump has signed an Executive Order to "prevent online censorship", and Twitter has hidden, for "glorifying violence", Trump tweets suggesting shooting protesters in Minneapolis. It's clear this situation will escalate over the coming week. Twitter has a difficult balance to maintain: it's important not to hide the US president's thoughts from the public, but it's equally important to hold the US president to the same standards that apply to everyone else. Of course he feels unfairly picked on.

Rewind to Tuesday. Twitter applied its recently-updated rules regarding election integrity by marking two of Donald Trump's tweets. The tweets claimed that conducting the November presidential election via postal ballots would inevitably mean electoral fraud. Trump, who moved his legal residence to Florida last year, voted by mail in the last election. So did I. Twitter added a small, blue line to the bottom of each tweet: "! Get the facts about mail-in ballots". The link leads to numerous articles debunking Trump's claim. At OneZero, Will Oremus explains Twitter's decision making process. By Wednesday, Trump was threatening to "shut them down" and sign an Executive Order on Thursday.

Thursday morning, a leaked draft of the proposed executive order had been found, and Daphne Keller had color coded it to show which bits matter. In a fact-check of what power Trump actually has for Vox, Shirin Ghaffary quotes a tweet from Lawrence Tribe, who calls Trump's threat "legally illiterate". Unlike Facebook, Twitter doesn't accept political ads that Trump can threaten to withdraw, and unlike Facebook and Google, Twitter is too small for an antitrust action. Plus, Trump is addicted to it. At the Washington Post, Tribe adds that Trump himself *is* violating the First Amendment by continuing to block people who criticize his views, a direct violation of a 2019 court order.

What Trump *can* do - and what he appears to intend to do - is push the FTC and Congress to tinker with Section 230 of the Communications Decency Act (1996), which protects online platforms from liability for third-party postings spreading lies and defamation. S230 is widely credited with having helped create the giant Internet businesses we have today; without liability protection, it's generally believed that everything from web comment boards to big social media platforms will become non-viable.

On Twitter, US Senator Ron Wyden (D-OR), one of S230's authors, explains what the law does and does not do. At the New York Times, Peter Baker and Daisuke Wakabayashi argue, I think correctly, that the person a Trump move to weaken S230 will hurt most is...Trump himself. Last month, the Washington Post put the count of Trump's "false or misleading claims" while in office at 18,000 - and the rate has grown over time. Probably most of them have been published on Twitter.

As the lawyer Carrie A. Goldberg points out on Twitter, there are two very different sets of issues surrounding S230. The victims she represents cannot sue the platforms where they met serial rapists who preyed on them or continue to tolerate the revenge porn their exes have posted. Compare that very real damage to the victimhood conservatives are claiming: that the social media platforms are biased against them and disproportionately censor their posts. Goldberg wants access to justice for the victims she represents, who are genuinely harmed, and warns against altering S230 for purposes such as "to protect the right to spread misinformation, conspiracy theory, and misinformation".

However, while Goldberg's focus on her own clients is understandable, Trump's desire to tweet unimpeded about mail-in ballots or shooting protesters is not trivial. We are going to need to separate the issue of how and whether S230 should be updated from Trump's personal behavior and his clearly escalating war with the social medium that helped raise him from joke to viable presidential candidate. The S230 question and how it's handled in Congress is important. Calling out Trump when he flouts clearly stated rules is important. Trump's attempt to wield his power for a personal grudge is important. Trump versus Twitter, which unfortunately is much easier to write about, is a sideshow.


Illustrations: Drunk parrot in a Putney garden (by Simon Bisson; used by permission).

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.

May 22, 2020

The pod exclusion

Vintage_Gloritone_Model_27_Cathedral-Tombstone_Style_Vacuum_Tube_Radio,_AM_Band,_TRF,_Circa_1930_(14663394535).jpgThis week it became plain that another bit of the Internet is moving toward the kind of commercialization and control the Internet was supposed to make difficult in the first place: podcasts. The announcement that one of the two most popular podcasts, the Joe Rogan Experience, will move both new episodes and its 11-year back catalogue to Spotify exclusively in a $100 million multiyear deal is clearly a step change. Spotify has also been buying up podcast networks, and at the Verge, Ashley Carman suggests the podcast world will bifurcate into twin ecosystems, Spotify versus Everyone Else.

Like a few hundred million other people, I am an occasional Rogan listener, my interest piqued by a web forum mention of his interview with Jeff Novitzky, the investigator in the BALCO doping scandal. Other worth-the-time interviews from his prolific output include Lawrence Lessig, epidemiologist Michael Osterholm (particularly valuable because of its early March timing), Andrew Yang, and Bernie Sanders. Parts of Twitter despise him; Rogan certainly likes to book people (usually, but not always, men - for example Roseanne Barr) who are being pilloried in the news and jointly chew over their situation. Even his highest-profile interviewees rarely find, anywhere else, the two to three hours Rogan spends letting them talk quietly about their thinking. He draws them out by not challenging them much, and his predilection for conspiracy theories and interest in unproven ideas about nutrition make it advisable to be selective and look for countervailing critiques.

It's about 20 years since I first read about Dave Winers early experiments in "audio blogging", renamed "podcast" after the 2001 release of the iPod eclipsed all previously existing MP3 players. The earliest podcasts tended to be the typical early-stage is-this-thing-on? that leads the unimaginative to dismiss the potential. But people with skills honed in radio were obviously going to do better, and within a few years (to take one niche example) the skeptical world was seeing weekly podcasts like Skepchick (beginning 2005) and The Pod Delusion (2009-2014). By 2014, podcast networks were forming, and an estimated 20% of Americans were listening to podcasts at least once a month.

That era's podcasts, although high-quality, were - and in some cases still are - produced by people seeking to educate or promote a cause, and were not generally money-making enterprises in their own right. The change seems to have begun around 2010, as the acclerating rise of smartphones made podcasts as accessible as radio for mobile listening. I didn't notice until late 2016, when the veteran screenwriter and former radio announcer and DJ Ken Levine announced on his daily 11-year-old blog that he was starting up Hollywood & Levine and I discovered the ongoing influx of professional comedians, actors, and journalists into podcasting. Notably, they all carried ads for the same companies - at the minimum, SquareSpace and Blue Apron. Like old-time radio, these minimal-production ads were read by the host, sometimes making the whole affair feel uncomfortably fake. Per the Wall Street Journal, US advertising revenue from podcasting was $678.7 million last year, up 42% over 2018.

No wonder advertisers like podcasts: users can block ads on a website or read blog postings via RSS, but no matter how you listen to a podcast the ads remain in place, and if you, like most people, listen to podcasts (like radio) when your hands are occupied, you can't easily skip past them. For professional communicators, podcasts therefore provide direct access to revenues that blogging had begun to offer before it was subsumed by social media and targeted advertising.

The Rogan deal seems a watershed moment that will take all this to a new level. The key element really isn't the money, as impressive as it sounds at first glance; it's the exclusive licensing. Rogan built his massive audience by publishing his podcast in both video and audio formats widely on multiple platforms, primarily his own websites and YouTube; go to any streaming site and you're likely to find it listed. Now, his audience is big enough that Spotify apparently thinks that paying for exclusivity will net the company new subscribers. If you prefer downloads to streaming, however, you'll need a premium subscription. Rogan himself apparently thinks he will lose no control over his show; he distrusts YouTube's censorship.

At his blog on corporate competition, Matt Stoller proclaims that the Rogan deal means the death of independent podcasting. While I agree that podcasts circa 2017-2020 are in a state similar to the web in the 2000s, I don't agree this means the death of all independent podcasting - but it will be much harder for their creators to find audiences and revenues as Spotify becomes the primary gatekeeper. This is what happened with blogs between 2008 and 2015 as social media took over.

Both Carman's and Stoller's predictions are grim: that podcasts will go the way of today's web and become a vector for data collection and targeted advertising. Carman, however, imagines some survival for a privacy-protecting, open ecosystem of podcasts. I want to believe this. But, like blogging now, that ecosystem will likely have to find a new business model.


Illustrations: 1930s vacuum tube radio (via Joe Haupte).

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.

May 15, 2020

Quincunx

Thumbnail image for sidewalklabs-streetcrossing.pngIn the last few weeks, unlike any other period in the 965 (!) previous weeks of net.wars columns: there were *five* pieces of (relatively) good news in the (relatively) restricted domain of computers, freedom, and privacy.

One: Google sibling Sidewalk Labs has pulled out of the development it had planned with Waterfront Toronto. This project has been contentious ever since the contract was signed in 2017 to turn a 12-acre section of Toronto's waterfront into a data-driven, sensor-laden futuristic city. In 2018, leading Canadian privacy pioneer Ann Cavoukian quit the project after Sidewalk Labs admitted that instead of ensuring the data it collected wouldn't be identifiable it actually would grant third parties access to it. At a panel on smart city governance at Computers, Privacy, and Data Protection 2019, David Murakami Wood gave the local back story (go to 43:30) on the public consultations and the hubris on display. Now, blaming the pandemic-related economic conditions, Sidewalk Labs has abandoned the plan altogether; its public opponents believe the scheme was really never viable in the first place. This is good news, because although technology can help some of urban centers' many problems, it should always be in the service of the public, not an opportunity for a private company to seize control.

Two: The Internet Corporation for Assigned Names and Numbers has rejected the Internet Society's proposal to sell PIR, the owner of the .org generic top-level domain, to the newly created private equity firm Ethos Capital, Timothy B. Lee reports at Ars Technica. Among its concerns, ICANN cited the $360 million in debt that PIR would have been required to take on, Ethos' lack of qualifications to run such a large gTLD, and the lack of transparency around the whole thing. The decision follows an epistolary intervention by California's Attorney General, who warned ICANN that it thought that the deal "puts profit above the public interest" and that ICANN was "abandoning its core duty to protect the public interest". As the overseer of both it (as a non-profit) and the sale, the AG was in a position to make its opinion hurt. At the time when the sale was announced, the Internet Society claimed there were other suitors. Perhaps now we'll find out who those were.

Three: The textbook publishers Cengage and McGraw-Hill have abandoned their plan to merge, saying that antitrust enforcers' requirements that they divest their overlapping businesses made the merger uneconomical. The plan had attracted pushback from students, consumer groups, libraries, universities, and bookstores, as well as lawmakers and antitrust authorities.

Four: Following a similar ruling from the UK Intellectual Property Office, the US Patent and Trademark Office has rejected two patents listing the Dabus AI system as the inventor. The patent offices argue that innovations must be attributed to humans in order to avoid the complications that would arise from recognizing corporations as inventors. There's been enough of a surge in such applications that the World Intellectual Property Organization held a public consultation on this issue that closed in February. Here again my inner biological supremacist asserts itself: I'd argue that the credit for anything an AI creates belongs with the people who built the AI. It's humans all the way down.

Five: The US Supreme Court has narrowly upheld the right to freely share the official legal code of the state of Georgia. Carl Malamud, who's been liberating it-ought-to-be-public data for decades - he was the one who first got Securities and Exchange Commission company reports online in the 1990s, and on and on - had published the Official Code of Georgia Annotated. The annotations in question, which include summaries of judicial opinions, citations, and other information about the law, are produced by Lexis-Nexus under contract to the state of Georgia. No one claimed the law itself could be copyrighted, but the state argued it owned copyright in the annotations, with Lexis-Nexus as its contracted commercial publisher. The state makes no other official version of its code available, meaning that someone consulting the non-annotated free version Lexis-Nexus does make available would be unaware of later court decisions rejecting parts of some of the laws the legislature passed. So Malamud paid the hundreds of dollars to buy a full copy of the official annotated version, and published it in full on his website for free access. The state sued. Public.Resource lost in the lower courts but won on appeal - and, in a risky move, urged the Supreme Court to take the case and set the precedent. The vote went five to four. The impact will be substantial. Twenty-two other states publish their legal code under similar arrangements with Lexis-Nexus. They will now have to rethink.

All these developments offer wins for the public in one way or another. None should be cause for complacence. Sidewalk Labs and other "surveillance city" purveyors will try again elsewhere with less well-developed privacy standards - and cities still have huge problems to solve. The future of .org, the online home for the world's non-profits and NGOs, is still uncertain. Textbook publishing is still disturbingly consolidated. The owners of AIs will go on seeking ways to own their output. And ensuring that copyright does not impede access to the law that governs those 23 American states does not make those laws any more just. But, for a brief moment, it's good.

Illustrations:

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.

This week's net.wars, "Quincunx", wakes up to discover a confluence of (relatively) good news in the last few weeks of computers, freedom, and privacy: https://www.pelicancrossing.net/netwars/2020/05/quincunx.html

May 1, 2020

Appified

china-alihealth.jpegAround 2010, when smartphones took off (Apple's iPhone user base grew from 8 million in 2009 to 100 million in early 2011), "There's an app for that" was a joke widely acknowledged as true. Faced with a pandemic, many countries are looking to develop apps that might offer shortcuts to reaching some variant of "old normal". The UK is no exception, and much of this week has been filled with debate about the nascent contact tracing app being developed by the National Health Service's digital arm, NHSx. The logic is simple: since John Snow investigated cholera in 1854, contact tracing has remained slow, labor-intensive , and dependent on infected individuals' ability to remember all their contacts. With a contagious virus that spreads promiscuously to strangers who happen to share your space for a time, individual memory isn't much help. Surely we can do better. We have technology!

In 2011, Jon Crowcroft and Eiko Yoneki had that same thought. Their Fluphone proved the concept, even helping identify asymptomatic superspreaders through the social graph of contacts developing the illness.

In March, China's Alipay Health got our attention. This all-seeing, all-knowing, data-mining, risk score-outputting app whose green, yellow, and red QR codes are inspected by police at Chinese metro stations, workplaces, and other public areas seeks to control the virus's movements by controlling people's access. The widespread Western reaction, to a first approximation: "Ugh!" We are increasingly likely to end up with something similar, but with very different enforcement and a layer of "democratic voluntary" - *sort* of China, but with plausible deniability.

Or we may not. This is a fluid situation!

This week has been filled with debate about why the UK's National Health Service's digital arm (NHSx) is rolling its own app when Google and Apple are collaborating on a native contact-tracing platform. Italy and Spain have decided to use it; Germany, which was planning to build its own app, pivoted abruptly, and Australia and Singapore (whose open source app, TraceTogether, was finding some international adoption) are switching. France balked, calling Apple "uncooperative".

France wants a centralized system, in which matching exposure notifications is performed on a government-owned central server. That means trusting the government to protect it adequately and not start saying, "Oooh, data, we could do stuff with that!" In a decentralized system, the contact matching us performed on the device itself, with the results released to health officials if the user decides to do so. Apple and Google are refusing to support centralized systems, largely because in many of the countries where iOS and Android phones are sold it poses significant dangers for the population. Essentially, the centralized ones ask you for a lot more trust in your government.

All this led to Parliament's Human Rights Committee, which spent the week holding hearings on the human rights implications of contact tracing apps. (See Michael Veale's and Orla Lynskey's written evidence and oral testimony.) In its report, the committee concluded that the level of data being collected isn't justifiable without clear efficacy and benefits; rights-protecting legislation is needed (helpfully, Lilian Edwards has spearheaded an effort to produce model safeguarding legislation; an independent oversight body is needed along with a Digital Contact Tracing Human Rights Commissioner; the app's efficacy and data security and privacy should be reviewed every 21 days; and the government and health authorities need to embrace transparency. Elsewhere, Marion Oswald writes that trust is essential, and the proposals have yet to earn it.

The specific rights discussion has been accompanied by broader doubts about the extent to which any app can be effective at contact tracing and the other flaws that may arise. As Ross Anderson writes, there remain many questions about practical applications in the real world. In recent blog postings, Crowcroft mulls modern contact tracing apps based on what they learned from Fluphone.

The practical concerns are even greater when you look at Ashkan Soltani's Twitter feed, in which he's turning his honed hacker sensibilities on these apps, making it clear that there are many more ways for these apps to fail than we've yet recognized. The Australian app, for example, may interfere with Bluetooth-connected medical devices such as glucose monitors. Drug interactions matter; if apps are now medical devices, then their interactions must be studied, too. Soltani also raises the possibility of using these apps for voter suppression. The hundreds of millions of downloads necessary to make these apps work means even small flaws will affect large numbers of people.

All of these are reasons why Apple and Google are going to wind up in charge of the technology. Even the UK is now investigating switching. Fixing one platform is a lot easier than debugging hundreds, for example, and interoperability should aid widespread use, especially when international travel resumes, currently irrelevant but still on people's minds. In this case, Apple's and Google's technology, like the Internet itself originally, is a vector for spreading the privacy and human rights values embedded in its design, and countries are changing plans to accept it - one more extraordinary moment among so many.

Illustrations: Alipay Health Code in action (press photo).

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.

A life in three lockdowns

squires-rainbow.jpg"For most people it's their first lockdown," my friend Eva said casually a couple of weeks ago. "It's my third."

Third? Third?!

Eva is Eva Pascoe, whose colorful life story so far includes founding London's first cybercafe in 1994, setting up Cybersalon as a promulgator of ideas and provocations, and running a consultancy for retailers. She drops hints of other activities: mining cryptocurrencies in Scandinavia using renewable energy, for example. I'm fairly sure it's all true.

So: three lockdowns.

Eva's first lockdown was in 1981, when the Communist Party in her home country, Poland, decided to preempt Russian intervention against the Solidarity workers' movement and declared martial law. One night the country's single TV channel went blank; the next morning they woke up to sirens and General Wojciech Jaruzelski banning public gatherings and instituting a countrywide curfew under which no one could leave their house after 6pm. Those restrictions still left everyone going to work every day and, as it turned out crucially, kept the churches open for business.

Her second was in 1987, and was unofficial. On April 26, 1986, her nuclear physics student flatmate noticed that the Geiger counter in his lab at Warsaw's Nuclear Institute was showing extreme - and consistent - levels of radiation. The Russian Communist Party was saying nothing, and the rest of Poland wouldn't find out until four days later, but Chernobyl had blown up. Physicists knew and spread the news by word of mouth. The drills they'd had in Polish schools told them what to do: shelter indoors, close all windows, admit no fresh air. Harder was getting others to trust their warnings at a time without mobile phones and digital cameras to show the Geiger counter's readings.

Those two lockdowns had similarities. First, they were abrupt, arriving overnight with no time to prepare. That posed a particular difficulty in the second lockdown, when outside food couldn't be trusted because of radioactive fallout, and it wasn't clear whether the water in the taps was safe. "As in COVID-19," she wrote in a rough account I asked her to create, "we had to protect against an invisible enemy with no clear knowledge of the surface risks already in the flat, and no ability to be sure when the danger passes." After 14 days, with no sick pay available, they had to re-emerge and go to work. With the Communist Party still suggesting the radiation was mostly harmless, "In the absence of honest government information, many myths about cures for fallout circulated, some looking more promising than others."

Their biggest asset in both lockdowns was the basement tunnels that connect Warsaw's ten-story blocks of flats, each equipped with six to ten entrances leading to separate staircases. A short run through these corridors enabled inhabitants to connect with the hundreds of other people in the same block when it was too dangerous to go outside. Even under martial law, with deaths and thousands of arrests on the streets, mostly of Solidarity activists, those basement corridors enabled parties featuring home-brewed beer and vodka, pickled cabbage, mushrooms, and herring, and "sausages smuggled in from Grandma's house in the countryside". Most important was the vodka.

The goal of martial law was to stop the spread of ideas, in this case, the Polish freedom movement. The connections made in those basement corridors - and the churches - ensured it failed. After 18 months, the lockdown ended because it was unsustainable. Communist rule ended in 1989, as in many other eastern European countries.

Chernobyl's effects were harder to shake. When the government eventually admitted the explosion had taken place, it downplayed the danger, suggesting that vegetables would be safe to eat if scrubbed with hot water, that the level of radiation was about the same as radon - at the time, thought to be safe - and insisted the population should participate in the May 1 Labor Day marches. Eventually, Polish leaders broke ranks, advised people to stay at home and stop eating food from the affected 40% of Poland, and organized supplies of Lugol for young people to try to mitigate the effects of the radioactive iodine Chernobyl had spread. Eva, a few years too old to qualify, calls her Hashimoto's thyroiditis "a lifelong reminder of why we must not blindly trust government health advice during large-scale medical emergencies".

Eva's lessons: always have a month's supply of food stocks; make friends with virologists, as this will not be our last pandemic; buy a gas mask and make sure everyone knows how to put it on. Most important, buy home-brew equipment. "It not only helps to pass time, but alcohol becomes a currency when the value of money disappears."

This lockdown gave us advance notice; if you were paying attention, you could see it forming on the horizon a month out. Anyone who was stocked for a no-deal Brexit was already prepared. But ironically, the thing that provided safety, society, and survival during Eva's first two lockdowns would be lethal if applied in this one, which finds her in a comfortable London house with a partner and two children. Basement tunnels connecting households would be spreading disease and death, not ideas and safety in which to hatch them. Our tunnels are the Internet and social media; our personal connections are strengthening, even with hugs on pause.


Illustrations: Sign posted on the front door of a local shop that had to close temporarily.

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.