App stores, trust and anti-trust. The app store model has been a central part of the smartphone revolution, bringing safe, trusted software to billions of people for the first time. Breaking it would be insane. The trouble is, it also means Apple (and Google) aren’t the pirates anymore - they’re the navy, the port and the customs house, so how do they manage that, and how soon do regulators step in? Link
The ecommerce surge. We now have official stats on lockdown ecommerce for the USA and UK, and they're dramatic: The UK went from 20% ecommerce penetration to over 30% in two months, and the USA from 17% to 22%. Link
Lockdown ecommerce. The USA released e-commerce stats for Q2, and we now have several weeks of data for the UK and some European countries: see my essay linked above.
Two stores of dysfunctional regulators this week. First, Australia. As I've written before, there's a persistent idea in some newspaper companies that if they appear in a Google search, or someone shares a story on FB, then Google and FB are making money from that and newspapers should be paid for the inbound link. Australia is trying to write this into law, badly:
- News has little direct value to internet platforms: that's not where the the ads are. So, pretending this is a value exchange means the whole discussion has an Alice in Wonderland quality: it would be much better to be honest and talk about a targeted tax and a subsidy if that's what you want to do.
- Even if you want a subsidy, the Australian proposal is accompanied by a bunch of frankly insane operational demands: for example: that any changes to search or recommendation algorithms that might affect news sites' rankings need 28 days notice. In practice that means all changes need 28 days notice, which would basically shut down Google and FB's operations (and of course, make it impossible for them to stop news sites from gaming the system).
- This is all accompanied by a lot of cheerful mud-throwing and insults on the Australian model, in which the Aussie regulator generally gives the impression that it hasn't read its own proposals (of course, maybe it hasn't).
- The deeper problem: these companies can be and and probably should be regulated, but you can't invent new economics or demand things that are operationally impossible: you need to set aside wishful thinking and do the analysis.
- Links: Coverage, Google's messaging
California's 'gig economy' mess. There is an important, underlying and broad debate about whether and when an Uber driver should be an employee - but this is a story about bad regulation. Late last year California passed a law, 'AB5', that tried to force Uber and Lyft to classify drivers as employees, but the original drafting was so inept it classified anyone who does any freelance work at all as an employee, which caused chaos: a musician who played one gig at a venue would become a permanent employee of that venue (really). Now there's an ongoing process of amendments with endless carve-outs of other professions, aiming to leave only Uber and Lyft covered. This is a model of bad regulation: instead of setting a principle of what makes an employee, and working out why (and if!) a driver is an employee and a hairdresser with the same hours is not, we just have muddle. Then more chaos: on 10 August a judge issued an injunction ordering Uber and Lyft to convert all their drivers to employees in 10 days, which was of course obviously impossible, so they threatened to shut down in California, until that order was overturned. There are real questions here, Uber in the past wasn't anyone's model of a good citizen, and California has moved towards being the USA's tech regulator by default - but gross incompetence like this does't help anyone. Link
Amazon delivery. Apparently, Amazon's own fleet of vehicles is handling almost two thirds of US deliveries. Lower cost, faster delivery, more customers - straight out of the famous flywheel from 2000. Link
More Facebook content moderation, this time banning QAnon and similar extremist groups and conspiracy theories. Link
The Apple/Epic drama continues: Epic sued for an emergency injunction for Fortnite to be let back on the iOS app store while its broader case against Apple continues. Apple's response: 'there is nothing stopping Epic getting back in the store while its anti-trust case goes through the courts - just follow the rules that applied since 2011 (and iOS is only 9% of Fortnite revenue anyway)'. I am not a lawyer, but you can believe Apple shouldn't get 30% and even believe the store model should end while also believing this is a stunt. Link(PDF)
Apple's sweetheart deals. Following the discovery that Apple has given Amazon a special 15% commission rate on in-app purchase of Prime subscriptions (so much for the 'level playing field'), a newspaper trade group is asking whyt it can't have the same. Link(PDF)
App Store horror stories. As I wrote last week, there are far too many horror stories of arbitrary and inconsistent decisions around Apple's app store, and now here's a great example: Apple briefly insisted that Wordpress include add in-app payment to buy web hosting, and then backed down and apologised. You can believe that the sandboxed App Store model is a very good thing while also believing that Apple’s payment policies are unsustainable and its management of the store is dysfunctional. Link
Google Maps. Google is now feeding satellite data much more directly back into its mapping products. Machine learning, of course. Link
(This week's news seems to be mostly about people screwing up. Hopefully something more interesting happens next week 🤷🏻♂️)
Fascinating and very detailed study, based on internal data from Uber and Lyft, on the distribution of hours and income for on-demand car service drivers. How many people really are working 'full time'? What do they earn? Only covers Seattle, but has broader implications. Link (PDF)
Thoughts about how Amazon hides email receipts and what that might mean for AR glasses: will AR glasses screen-scrape the world? Who owns that data? Link
A long time ago, when iPods were cool, Apple very secretly helped the US government make a special, one-off modified version with... something inside. Link
Casey Newton being thoughtful about tech journalism. Link
With a high-quality audio recording of a key turning in a lock, you can work out what's inside and make a duplicate key. Link
A lovely 1960s commercial for the Selectric typewriter (which the Russians managed to bug in the same way) Link
Chosen almost at random: one of the Cambian explosion of interesting new video/collaboration/remote work apps: here.fm. Link (needs Chrome)
Using machine learning to generate photo-realistic (ish) images of Roman emperors from their statues and coins. Link
Yelp data on linking consumer behaviour with, potentially, COVID hotspots. Link
Nielsen on US consumer media consumption in lockdown. Link