Benedict's
Newsletter

This is a weekly newsletter of what was interesting and important in tech. This is the free edition, with 150,000 subscribers.

Not a subscriber yet? Sign up here.

I decided to make the column in this week's issue of the newsletter free. You can upgrade here to get every issue.

My essays

Amazon's profits, AWS and advertising. The bigger Amazon gets, the more it's worth reading the accounts. Does AWS subsidise the whole thing? Is the revenue $250bn - or $450bn? And is that ad business just a footnote, or is it bringing in more cash than AWS? Link

News

Facebook and the US election. Mark Zuckerberg announced that no new political ads will be allowed on FB in the week before the US election, as well as a whole range of things aimed at trying to stop misinformation about the voting process. A few things to think about:

This is a significant amount of power for one unelected 36-year-old to hold, regardless of whether he takes the right decisions. This wasn't what he had in mind when he launched thefacebook.com and it's partly why Facebook pushes for both regulation and governance boards: who should decide?
Most such misinformation, and rumours, and any misleading ads, comes and will come from real Americans, not fake accounts created by Russians. It's not Facebook's fault that the USA has national politicians who want to spread misinformation - BUT...
Facebook built an unprecedented platform for spreading information without any of the previous layers of gatekeeping and filters. Unlike 2016, it's now very conscious of how that can go wrong, but it's not entirely clear what a structural solution would be, as opposed to playing whack-a-mole.
Meanwhile - if Trump makes a speech saying X, Y or Z on election night, how will the old-fashioned analogue media report it?
Zuck's announcement: Link

Banning Indian politicians on FB. After a lot of internal argument, Facebook banned an Indian politician, T. Raja Singh, for inflammatory statements that break its rules. There were suggestions his behaviour had previously been ignored due to his political influence - he's a member of the ruling BJP and India is FB's biggest market by users. FB may have screwed up here by not banning him before, but as in the previous item - whose decision is this, and (how) can Facebook be the arbiter of political speech for every country on earth? Link ($)

PR manipulation: Facebook said a US PR firm was involved in 'co-ordinated inauthentic behaviour' around elections in Venezuela, Mexico and Bolivia. Link

India is banning more Chinese apps, including PUBG Mobile (Korean, but published by Tencent) and over 100 others. Some of this is about security on your phone, but it's mostly about an escalating fight with China per se where banning these is a convenient way to signal. Link

Apple has finally published a human rights policy. Apple talks a lot about privacy, but also stores Chinese user data unencrypted in China and bars VPN apps from the Chinese app store. If you're going to operate globally, how do you reconcile your own views of how to protect users with obeying local law? (And again - is it your decision?) Google had to leave China, but Apple has a big business there and has to think about this. Link

Australia's link tax: Facebook followed Google last week in reacting to the Australia proposal to tax links to newspaper websites: it said that it might block links to news in Australia (or anything else that creates an automatic liability). This is predictable: bad laws create unintended consequences. Link

Apple postpones mobile ad-tech armageddon. As I wrote last week, iOS 14 (due out this month) makes tracking of user behaviour between apps (using IDFA, roughly equivalent to a cookie) opt-in, which will cripple it. Now Apple says it will delay implementation of that until 2021 to give the ecosystem more time to adjust. Link

Fake reviews on Amazon?! Amazon UK deleted 20k user product reviews after the FT pointed out that most of the top reviewers were obviously faking. Amazon reviews are effectively useless for anything where there's enough money involved to attract incentives. Link ($)

Netflix free previews: Netflix is putting some movies and TV shows (or just the first episode) outside the pay wall to market to new customers. Link

Hollywood antitrust: In 1949 an anti-trust case against the Hollywood studios forced them to split off their tied chains of movie theatres: they'd previously been using them to lock in a distribution cartel. Now the world has changed so much that the decree is being unwound. Limited implications for all except perhaps Disney, but symbolic. Link

Reading

A Florida sheriff uses 'data' to guess who will commit crime, and sends his deputies to 'hunt down' and harass them. When terrible police use computers, they're still terrible police. Link

Facebook is launching a proposal for an external, peer-reviewed, data-based academic study of social media effects on elections. Link

Audio interview with US politician David Cicilline (on the anti-trust panel from last month): he thinks we need a 'Glass-Steagall Act' for the internet to stop platform companies competing on their own platforms. He doesn't seem to understand that this would ban, say, Google Maps or Microsoft Office, and I don't think this will happen, but useful to listen to get a sense of where he's coming from. Link

"The effects of political advertising are small regardless of context, message, sender, or receiver" (in other words, stop freaking out about micro-targeting). Link

Amazon drivers are hanging smartphones in trees to get more work. Too good to check. Link ($)

For Lebanese students post-explosion, WhatsApp is their classroom. Link

Facebook has some new research and tech on spatial audio, as a building block for VR and potentially AR in future. Link

Interesting things

The Dallas-Fort Worth freeway system. This is the kind of website no-one makes anymore, and amazing. Link

A taxonomy of Window Manager graphical user interfaces, from 1988. When things looked much less clear. Link (PDF)

Make your own cyborg cockroach. Not entirely sure how I feel about this. Link

A 'digital' pregnancy test is just an analogue paper strip test with a light sensor to read the result for you. Link

Stats

The Stanford Cable TV News Analyser - machine learning for media analysis. Link

Epic filed another motion in its war with Apple, which isn't very interesting, except that it disclosed some data: iOS DAUs are down 60% since Epic started the fight, 116m of Fortnite's 350m registered users are on iOS, and iOS users have played 2.86bn hours. Link

Smartphone contact tracing is effective even at low take-up. Link

Column: new spheres of speech

We don't really know what we think about speech online. We spent hundreds of years evolving complex, mostly implicit social norms, institutional structures and laws around speech, where speech exists in many different spheres, from a private phone call to a bar to a newspaper. Actual laws in which the government stops you saying things are pretty limited, both in scope and by country, but there are great thickets of custom, ethical standards, convention and peer pressure about what can get onto the front page of a newspaper, the window of a bookshop or the evening news, and indeed what you can and can't say in a bar.

That is, 'What can you say?' is different depending on where you are, and it's enforced (or not) by different gatekeepers - by your friends, colleagues or peers, or by a barman, or by a newspaper editor or bookseller or TV station and their peers and conventions, and very occasionally by a judge. And, of course, those norms are different: we can mostly agree about incitement to violence or CSAM, but the UK, France and the USA have very attitudes to, say, libel law or privacy.

But none of those norms map neatly to the internet. It's not just that there are no or very different gatekeepers - the spheres of speech don't map either. An SMS or a WhatsApp message still counts straightforwardly as a private conversation, but at what point does a FB Group become public? With 10 members? 100? If I reply to your tweet, is that public? It depends. If I post to my Facebook feed, how many followers must I have before I'm 'broadcasting'? Is a tweet from a journalist that goes viral a publication? Who's their editor? And if you follow me and tap 'like' on all of my posts, and Instagram shows you my latest post at the top of your feed, is Instagram promoting me?

The internet and then social platforms imploded all of our social norms around speech, and yet somehow Facebook / Google / Twitter are supposed to recreate that whole 200-year tapestry of implicit structures and consensus, and answer all of those questions, from office parks in the San Francisco Bay Area, for both the USA and Myanmar, right now. We want them to Fix It, but we don't actually know what that means.

You can see a microcosm of this in the US debate abut political ads on Facebook. Do you run ads that tell lies? Newspapers do, and US TV stations aren't allowed to block ads from qualifying candidates. Meanwhile a ban on advertising is good for incumbents, who already have organic reach, and populists and trolls, who can get it, but shuts out moderates and new entrants. But a lie on Facebook, spread with money, reaches new people. Who decides? For now, one 36-year-old called Mark.

This is part of the challenge: everyone at a big social media company thinks about his, but they do so very conscious that they don't have social or democratic legitimacy to make those kinds of decisions. They thought they were making a 'social app', not a systemically important part of democracy. In 2015, most people in Silicon Valley would have said censorship was bad, and also unscalable - now ML means you can at least try to scale it (with tens of thousands of human moderators) and everyone understands how bad things can get and the responsibility to do something. But what? How does a 30-something PM in Palo Alto decide the basis of political speech in Malaysia? This is why Facebook calls for regulation, and sets up advisory boards, and that might be better, but doesn't feel like the end point. I wrote last week that moderation may not be scalable, but the other question is who decides what should be 'moderated' in the first place, and we don't have a good answer to that either.

Forwarded this? Subscribe here

You signed up to this at http://www.ben-evans.com/newsletter. Want to change how you receive these emails? You can update your preferences or unsubscribe from this list.

Our mailing address is: Benedict Evans, 2 Golden Square, London, W1, United Kingdom