Start Up: the ICO scam that wasn’t, millennial streamers, machine learning and chaos, Intel shutters smart glasses, and more

Mycelia: they could make an impressive leather replacement for vegans. Photo by Amadej Trnkoczy on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 11 links for you. Use them wisely. I’m @charlesarthur on Twitter. Observations and links welcome.

Millennial streamers and music on smart speakers • Global Web Index

Olivia Valentine:


Millennials are the most passionate generation when it comes to music streaming, with 68% engaging. But just as the music-streaming business is experiencing continued user growth, the interest in smart voice-controlled devices (e.g. Amazon Echo, Google Home) is trending quickly upwards, too, pointing at a potential to shake up the music streaming landscape. 

Two-thirds of Millennial Music Streamers say they currently use or are planning to purchase one of these smart devices, putting this group 9 percentage points ahead of the Millennial average for this figure across the globe.


So quite a lot of room for Apple to expand into, if it gets it right. This is in the early stages.

(The definition of “music streamer” is someone who says they stream music daily.)
link to this extract

Palantir knows everything about you • Bloomberg Businessweek

Peter Waldman, Lizette Chapman, and Jordan Robertson offer a huge rundown on Peter Thiel’s data-gathering company, which does work mostly for law enforcement and finance (it seems); this little example from Los Angeles where the LAPD is using its “Gotham” system points to the problems:


In 2016, [22-year-old Manuel] Rios was sitting in a parked car with an Eastside 18 [gang] friend when a police car pulled up. His buddy ran, pursued by the cops, but Rios stayed put. “Why should I run? I’m not a gang member,” he says over steak and eggs at the IHOP near his home. The police returned and handcuffed him. One of them took his picture with a cellphone. “Welcome to the gang database!” the officer said.

Since then he’s been stopped more than a dozen times, he says, and told that if he doesn’t like it he should move. He has nowhere to go. His girlfriend just had a baby girl, and he wants to be around for them. “They say you’re in the system, you can’t lie to us,” he says. “I tell them, ‘How can I be in the hood if I haven’t got jumped in? Can’t you guys tell people who bang and who don’t?’ They go by their facts, not the real facts.”

The police, on autopilot with Palantir, are driving Rios toward his gang friends, not away from them, worries Mariella Saba, a neighbor and community organizer who helped him get off meth. When whole communities like East L.A. are algorithmically scraped for pre-crime suspects, data is destiny, says Saba. “These are systemic processes. When people are constantly harassed in a gang context, it pushes them to join. They internalize being told they’re bad.”


You don’t finish this thinking that Palantir are on the up and up.
link to this extract

Machine learning’s ‘amazing’ ability to predict chaos • Quanta Magazine

Natalie Wolchover:


Half a century ago, the pioneers of chaos theory discovered that the “butterfly effect” makes long-term prediction impossible. Even the smallest perturbation to a complex system (like the weather, the economy or just about anything else) can touch off a concatenation of events that leads to a dramatically divergent future. Unable to pin down the state of these systems precisely enough to predict how they’ll play out, we live under a veil of uncertainty.

But now the robots are here to help.

In a series of results reported in the journals Physical Review Letters and Chaos, scientists have used machine learning — the same computational technique behind recent successes in artificial intelligence — to predict the future evolution of chaotic systems out to stunningly distant horizons. The approach is being lauded by outside experts as groundbreaking and likely to find wide application.

“I find it really amazing how far into the future they predict” a system’s chaotic evolution, said Herbert Jaeger, a professor of computational science at Jacobs University in Bremen, Germany.

The findings come from veteran chaos theorist Edward Ott and four collaborators at the University of Maryland. They employed a machine-learning algorithm called reservoir computing to “learn” the dynamics of an archetypal chaotic system called the Kuramoto-Sivashinsky equation. The evolving solution to this equation behaves like a flame front, flickering as it advances through a combustible medium.


link to this extract

What’s the length of the shortest bit sequence that’s never been sent over the internet? • Sean Cassidy

Sean Cassidy:


A friend of mine posed this brain teaser to me recently: “What’s the length of shortest bit sequence that’s never been sent over the Internet?”

We can never know for sure because we don’t have a comprehensive list of all the data.

But what can we say probabilistically? Restating it like so: “At what value for X is there a 50% chance there’s a sequence of X-bits in length that hasn’t been transmitted yet?”

What does your intuition say? Obviously every 8-bit sequence has been sent, since there’s only 256 values. By downloading this HTML page over TLS you’ve probably used up every 8-bit value. Has every 100 byte message been sent?

This is how my intuition went: it’s probably less than 128 bits because UUIDs are 128 bits, and they’re universally unique. It’s probably greater than 48 bits because of how common collisions are at that end for hashes and CRCs, and the Internet has generated a lot of traffic.

How would we determine the right value?

I decided to model data as each bit sent is like flipping a coin. This isn’t strictly true, of course, but with encryption becoming more prevalent, it’s getting to be close.


If you’re at all into maths, it’s fun just to pause and try this before you go to his solution.
link to this extract

The disgraceful dockless drama: what dockless bikes/scooters are exposing •

Have A Go, following cease-and-desist orders sent to three electric scooter companies in San Francisco:


Dockless bikes and scooters are not actually the problem.

For decades now, cars have gotten the royal treatment. Users were able to pick up their cars and drop them off anywhere in the city. Automobile parking is all around. From street parking, to business parking lots, to single family homes with driveways and garages, to large parking structures. Thus, the user experience for drivers is essentially go anywhere, park anywhere.

It was simply expected that anywhere one goes in a city, one could be guaranteed a free, giant space to park one’s private 4000-pound box, no questions asked.

Sure, in some dense areas, payment is now required. Yet private vehicle parking is essentially considered a right. We know this because when we can’t find parking for more than two minutes, we get upset. 5 minutes? We get very upset.

We also know this because it is quite literally law with parking minimums mandates for homes, businesses, and just about every building that is built or remodeled. We’ve mandated that the space that could otherwise be utilized for affordable housing, parks, cafes, or other human uses is legally required to serve as public storage for urban tanks and a free subsidy to oil and car companies.

If a city all of a sudden woke up to find the same amount of parking for cars as there is now for bikes/scooters, there wouldn’t be a few angry tweets (as there is now for new dockless bikes/scooters), but riots in the streets!


Fair point.
link to this extract

Mylo, a material that looks like leather, made from plants • Bolt Threads


How do you make Mylo™?

We use corn stalks and supplemental nutrients to feed and grow our mycelium. We precisely control growth conditions like temperature and humidity to encourage the mycelium to grow upward and self-assemble into an organized mat of interconnected cells. Their connections give the material strength. We then use a natural tanning process and compress the mat to be as thin or thick as we’d like the final material to be. At this point the mycelium is no longer growing. The final step is to imprint any desired pattern, which gives us the final material.

Is this the same technology you use to make Microsilk™?

No, our Microsilk™ technology involves engineering yeast to produce protein materials. With Mylo™, we grow mycelium, which we process into the final Mylo™ material.

Our friends at Ecovative pioneered this mycelium fabrication technology, which literally grew out of the great work they’ve been doing in creating soft flexible foams. We were blown away, and thrilled when they agreed to allow us to help develop it into a commercially viable new material. We’ve established a long-term partnership with Ecovative to optimize this technology and put processes in place to produce commercial-ready Mylo™ material and bring products to market that consumers will love.

Are these mycelium genetically engineered?

No, instead we carefully control the wild spores’ growth conditions like temperature and humidity to engineer the final material’s properties.


Like to know the tensile strength compared to leather, and how it copes with rain and wear.
link to this extract

Facebook moves 1.5bn users out of reach of new European privacy law • The Guardian

Alex Hern:


Facebook has moved more than 1.5 billion users out of reach of European privacy law, despite a promise from Mark Zuckerberg to apply the “spirit” of the legislation globally.

In a tweak to its terms and conditions, Facebook is shifting the responsibility for all users outside the US, Canada and the EU from its international HQ in Ireland to its main offices in California. It means that those users will now be on a site governed by US law rather than Irish law.

The move is due to come into effect shortly before General Data Protection Regulation (GDPR) comes into force in Europe on 25 May. Facebook is liable under GDPR for fines of up to 4% of its global turnover – around $1.6bn – if it breaks the new data protection rules.

The shift highlights the cautious phrasing Facebook has applied to its promises around GDPR. Earlier this month, when asked whether his company would promise GDPR protections to its users worldwide, Zuckerberg demurred. “We’re still nailing down details on this, but it should directionally be, in spirit, the whole thing,” he said.

A week later, during his hearings in front of the US Congress, Zuckerberg was again asked if he would promise that GDPR’s protections would apply to all Facebook users. His answer was affirmative – but only referred to GDPR “controls”, rather than “protections”. Worldwide, Facebook has rolled out a suite of tools to let users exercise their rights under GDPR, such as downloading and deleting data, and the company’s new consent-gathering controls are similarly universal.

Facebook told Reuters “we apply the same privacy protections everywhere, regardless of whether your agreement is with Facebook Inc or Facebook Ireland”. It said the change was only carried out “because EU law requires specific language” in mandated privacy notices, which US law does not.


In other news, leopards’ spots remain unchanged.
link to this extract

Intel plans to shut down smart glasses group • The Information

Aaron Tilley:


The division, formed in 2013, made fitness trackers and smart glasses. Despite an investment of several hundred million dollars by Intel, including through acquisitions of other companies, the group never made much of an impact in the wearables market.

The closure is likely to lead to some layoffs. The department reportedly had 200 people earlier this year, down from as many as 800 in 2016, although the current size isn’t known. Employees who can’t find a position in other divisions of Intel will be laid off, the people said.

In February, Bloomberg reported that Intel was looking for outside investment for the smart glasses project. Intel valued the smart glasses division at $350m with around 200 employees, according to Bloomberg. The closure suggests Intel wasn’t able to raise any fresh investment. That same month, The Verge reported on the smart glass project, known internally as Vaunt.

In a statement, Intel said it is “continuously working on new technologies and experiences. Not all of these develop into a product we choose to take to market.” It added that Intel will continue to take a “disciplined approach as we keep inventing and exploring new technologies, which will sometimes require tough choices when market dynamics don’t support further investment.”

The unit’s closure is the latest sign of how Intel has failed to diversify beyond its core chip business. Intel has tried various other steps, including buying security firm McAfee and internet of services business Wind River, without success. Last year it sold a majority stake in McAfee and recently sold Wind River.


Wearables are tricky – look at Nokia giving up on Withings – but it’s hard not to feel that Intel is getting out of this at the wrong time. Unless it has discovered things about AR and similar which tell it that this is an utter dead end.
link to this extract

Amazon recruits Best Buy to sell Fire TV edition smart TVs • Engadget

Steve Dent:


Best Buy has partnered with Amazon to sell voice-controlled Fire TV Edition branded TVs in its stores and on its website, the companies announced. CEOs Jeff Bezos from Amazon and Best Buy’s Hubert Joly said the retailer will sell 11 Amazon-powered TVs, including 4K and HD models, starting this summer with Toshiba models. At the same time, Best Buy will become a merchant on Amazon’s website and get exclusive rights to sell Amazon Fire TVs.

“Amazon and Best Buy have a long history of working together,” said Bezos, referring to Best Buy sales of Kindle readers. “Today we take our partnership to a new level.”

Amazon unveiled its first Fire TV Edition sets last year with TV maker Element, starting at $449 for a 43-inch model. The TVs expand on what you can do with a Fire TV stick, letting you see live TV alongside streaming options, detect devices connected to your TV, and consult a channel guide, to name a few features.

At the same time, you can use Alexa to control not just your TV, but also Hue lights and other smart home devices. Unlike with an Echo device, you have to hit a button to reach Alexa, as it’s not listening for a wake word.


Two things: first, this is bad news for Roku, which Best Buy used to sell with TVs; second, it increases the pressure on Apple to integrate the Apple TV into a TV set and just sell the all-in-one. Well, it worked for the iMac, right?
link to this extract

The marriage-saving robot that can assemble Ikea furniture… sort of • New Yorker

Ronan Farrow:


The researchers bought a basic chair from Ikea—the stefan model, made of lacquered pine—and, to simulate the human experience, took the pieces out of the box and put them “randomly within the environment.” First, the robot assessed where everything was; that took three seconds. Then it used an algorithm to plan out what to move where, and how; that took eleven minutes and twenty-one seconds. Actually building the chair took just under nine minutes. The sequence of steps was hard-coded in advance—the robot essentially followed the manual—but everything else was done on the fly. “The challenge is to quickly and consistently find fast, collision-free motions in a highly cluttered environment,” the researchers note, which pretty well describes every Ikea-furniture-building undertaking ever.

The Ikea bot’s arms move in extreme slow motion; it’s like watching two people try to put a chair together while stoned. (My point of reference here is Hikea Productions, a seemingly defunct YouTube channel featuring videos of humans assembling micke desks and nordli dressers while tripping on psychedelics.)


The robots though are kinda.. big? There’s a video. You wouldn’t want them in the living room. But I like the idea of Hikea productions. (Style note: the New Yorker calls it “ikea” but even the store calls itself Ikea, surely, so I’ve changed it. What I also don’t understand is the NYer’s spelling out of numbers like “twenty-one”, which are faster to read as “21”.)
link to this extract

We tracked down the SaveDroid motherfscker and need your help • Crypto Briefing

Adam Selene on the followup to that “cryptocurrency exit scam” which featured yesterday:


Crypto Briefing tracked Yassin Hankir’s location to an Egyptian resort town.

UPDATE: By now, many of you will know that Savedroid perpetrated what they consider to be a giant prank designed to illustrate that exit scams are a part of the crypto market.

Here’s the deal.

When you take $50m in funds that are raised by investors who believe in your project, you don’t then pretend to disappear with that money.

What happens if you do? People start looking for you.

Yassin Hassir may consider himself a genius for drawing the eyes of millions to his stunt. But to his investors, he caused them anxiety, pain, and distress.


Hassir broke that rule. And while there are those who will criticize Crypto Briefing for trying to help those investors track down their money, and the man who claimed to have stolen it, we are proud of our response.

We took the trouble to search for hours online to match a photo to a location in Egypt. We called the police in Frankfurt. We did what we could to be part of the REAL community – the kind of people you’d want on your side if this had been the real thing.

As to the prank itself? Not so clever. Not so funny. And having dragged his company’s name through the mud, not so good for Hassir’s investors.

We will not be removing the original article, printed below, because we are – as we say – proud to be acting in the best interests of our community.


Hassir posted a photo of a beach and an Egyptian beer online; the community pored over photos of beaches in Egypt, and tracked him down to his hotel. He says it’s a PR stunt. The equivalent of running with scissors while lighting matches in a firework factory.
link to this extract

Errata, corrigenda and ai no corrida: none notified

The axes of HomePod evolution: don’t judge what you can’t yet see

The MacBook Air, in an envelope just like it came in. Photo by yasuhisa yanagi on Flickr.

There’s been a lot of discussion about Apple’s HomePod, and the claim from Mark Gurman writing at Bloomberg that it “hasn’t lived up to expectations” in sales terms. Though if you ask analysts in the field, such as Ben Wood of CCS, they thought that 5 million in a year would be impressive. “Clearly my expectations were way lower than others,” commented Ben, which to me has echoes of the vastly inflated numbers that people expected smartwatches, and especially the Apple Watch, to sell, in the first year.

More relevant, I think, is the question of how the HomePod (or just “HomePod” – Apple never uses the definitive for its products, just as a parent wouldn’t for a child) is going to evolve.

And for that, it’s important to bear in mind how every single Apple product tends to evolve: from MVP, aka minimal viable product, to thing that people buy by the million.

Let’s go all the way back to the Bondi Blue iMac, from 1998, since that’s where the story of the modern Apple really begins. This was Apple trying to compete again in the PC market, and choosing to do so in an orthogonal way to pretty much everything else out there. It was all-in-one, it used USB (a new connector at the time), had no floppy drive (this alone was reckoned to spell its doom), and upgrading the RAM and hard drive was difficult.

But the next update showed the trajectory that Apple was on. The Bondi iMac gained colours, and it got more powerful, and added a DVD burner drive if you wanted. It didn’t revert to old connectors, but did add better sound.

iPod, youPod

Next, iPod.

Some of the iPod family – though there were more models and colours than this. Photo by Zengame on Flickr.

The first one was “expensive”, had a black and white and very pixel-y display, used a proprietary connector, had an unusual yet intuitive method for scrolling through songs, held 1,000 songs. As it evolved through the years, it generated variants that were smaller, had bigger and better screens, more memory, flash memory, but broadly the same controller and interface. (The iPod shuffle is like the Galapagos of iPods, but anyway.)

The MacBook Air (as at the top of the post). The first version had a very limited SSD drive, and underpowered processor. But it had those qualities of being thin and light and offering lots of battery power that people who could afford it really loved – especially when you compared to the average Windows laptop, which weighed tons more, had a DVD drive and floppy drive that road warriors in coffee shops didn’t need, and lasted much less long. You’d still recognise the first model today.

A calling for the iPhone

The iPhone X doesn’t look that different from the 6S, which doesn’t look that different from the original. Photo by Lucy Takakura on Flickr.

The iPhone is probably the poster child for MVP-ness. The first version’s software was limited (no MMS; no 3G; no text forwarding; no copy/paste), it was really expensive, its battery lasted a day when rival phones could last a week. But it could do so many things that others couldn’t, because of that touch screen and the concept of being a computer for your pocket, not a phone for your email. Subsequent versions have improved along pretty much every axis possible, apart from that battery life stuff (though the iPhone X is a big advance here).

The iPad. Look at these varieties.

iPad sizes have changed, but you’d still recognise the original if you’d only seen the newest, eight years on. Photo by MakeUseOf on Flickr.

The first version arrived in early 2010 and didn’t have a Retina screen (the iPhone 4, to be released later that year, would; but Retina screens at the size of an iPad were too expensive to contemplate for some time. What it did offer was something that wasn’t a PC trying to be squashed into a tablet form (which Dell and many others had been trying since 2001) and an approach to a big touch-driven screen that harked back to Jeff Haan’s remarkable TED demo of 2005. Since then, its screen has improved (hugely), it has gained an optional keyboard and pencil, and processor power has risen exponentially. But eight years on, the old and new look completely like siblings.

The Apple Watch. The first version could just about take you through a day if you didn’t exercise for too long. There has been plenty of tinkering with how the interface between apps works, but none with the basic concept of how you interact: lift-to-wake, or touches. The addition of GPS and phone data/calling has been welcomed, but if you hold the original Series 0 beside the latest Series 3, it’s essentially still the same thing: a device which hands off tasks from your phone to your wrist.

Bearing all this in mind, what should we expect from the HomePod’s evolution?

Again, look backward first. Look at the axes on which the previous devices evolved.
Evolution: colour, extras, price, screen resolution, interfaces.
No evolution: size, shape.

Evolution: colour, size, shape, control system, price, screen (colour-capable, video display-capable), output (to TV), interface (from proprietary Firewire, to 30-pin-Firewire and 30-pin USB, to Lightning; and Wi-Fi/Bluetooth in the iPod Touch), processor power.
No evolution: actually, pretty much everything about the iPod changed. (You could however argue that the iPod Touch is actually a cut-down iPhone, not an iPod, and that iPod evolution ended with the iPod shuffle of 2010 and iPod nano of 2012 – the latter being what some thought could be a precursor of an Apple wearable.)

MacBook Air:
Evolution: processor speed, screen size (smaller, never bigger than 13in), weight (reduced), disk size, price.
No evolution: screen quality (after all these years still isn’t Retina), shape, colour (there’s never been a black or rose gold MacBook Air; you want one, it’s aluminium).

Evolution: processor speed (duh), thickness, price, weight, colour, screen size, screen quality, login interface (from passcode to TouchID to FaceID), location capability (GPS, added in 2009’s iPhone 3GS), cameras (front-facing camera arrived on 2010’s iPhone 4).
No evolution: number of buttons (until the iPhone X, which removed the Home button), general interface, general portability, battery life

Evolution: size, colour, processor speed, screen resolution, screen capability (Tru-Tone etc), price, functional accessories (Smart Keyboard, Pencil), interface (USB/30-pin to USB/Lightning).
No evolution: screen size ratio, battery life

Apple Watch:
Evolution: battery life, straps, GPS, 4G connectivity, price (by selling older models at lower prices, rather than having differently priced new models, as happened with the iPhone)
No evolution: screen size

All right. Bearing in mind all the above, how should we expect the HomePod hardware to evolve, and not to evolve?

An evolution before your eyes

A “HomePod evolution” concept. Photo by Martin Hajek on Flickr.

Softly spoken, software

Given the way that everything gains software capability, I expect it will gain the capability to play more services than just Apple Music. The ability to play Spotify (which is, don’t forget, the world’s biggest streaming music service) is an obvious piece of low-hanging fruit if you’re looking to tempt people to buy a better-sounding device. Remember, Apple is into hardware sales; Services may be the thing that it talks up, but hardware is the motor that drives the engine.

In which case, why sell HomePod v1 without Spotify capability? Because the v1 is always the MVP – the minimum viable product. Look at the iPhone. Look at the iPad. Look at the MacBook Air. Look at the Apple Watch. They all started out lacking capabilities that seemed obvious (the first iPad didn’t even have alarms, which the iPhone did) and then gained them. Apple has been circling around what the HomePod should do for years – it’s been in development since Siri came on board – and the fact that it took this long to get out of the design labs suggests the usual cautious approach. The people who have bought the first model are obviously, self-definingly, going to be people who like Apple stuff; it can’t, therefore, do any harm to only offer Apple Music as a music service. But putting Spotify on? That’s just a question of an API to Spotify, and an instruction set for Siri so it recognises “play X on Spotify” or “play the X playlist from Spotify”. It can probably be done through a software update.

Other software? Besides playing music, smart speakers’ utility seems to lie in (1) checking the weather forecast (2) setting kitchen timers (3) streaming music (4) setting alarms. Below that, the proportion of people who say they’ve ever done this stuff falls below 50% of smart speaker owners (per Comscore) and it’s hard to know how often people do it.

So – weather, timers, music, alarms. Dig down to the 30% level and there’s also home automation, product ordering (that’s going to be Amazon), calendars, and games/jokes/general questions. (The HomePod can also do iMessage sending and receiving, and FaceTime alls; those don’t come up in the “things people do” listing above 13%, but it’s not something you imagine people wanting to do a great deal.)

These are all things that you can do now. So when people complain about the HomePod’s capability, they’re really complaining that it doesn’t have other music services, and about Siri. The first is a software update, and the second is – well, Apple seems to be working on it.

Hard wearing, but what hardware?

What about hardware? What can we expect there? Is the HomePod more like the portable iPod, which had multiple axes of evolution, or the deskbound iMac? In truth, it might be even closer to a device which I didn’t mention in the list above: the Apple TV.

Like the ATV, the HomePod has a limited interface (via a remote, or by voice), and in general once put somewhere it stays there essentially forever. The ATV has hardly evolved at all – there are a couple of varieties (4K/not 4K) and storage variants, but its onscreen interface is unlike anything else that Apple does. That’s been forced by the limitations of interactions with a TV screen, which one typically views from across a room, and that seems to have limited what it can do.

There has been no evolution of size or colour, and little on price (aside from selling the older model at lower cost). The competition from lower-priced rivals such as Roku, Google’s Chromecast and Amazon’s Fire Stick seems to have kept Apple stuck upmarket, and guarding its content (TV and movies bought on iTunes) jealously: you can’t get them on any of those three rivals without some DRM-fighting shenanigans.

There are signs of the HomePod taking up the same position. You can’t stream Apple Music on the Echo (though Amazon says it’s “open” to it) or Google Home. It’s possible Apple is going to treat the smart speaker market as being like the TV set-top box market – one to be fought over rigidly. Possibly that’s what caused the delay in its initial release: big internal fights over its future trajectory, for these things are all mapped out a couple of years ahead before the first product gets out of the door. (For example, iPhones are designed at least two years ahead.)

But I think that to make the HomePod as “closed” as the Apple TV would be a mistake, and given the way that other successful Apple products have evolved – different shapes and sizes and price points (to fit in with the way that people live their lives), greater software capability (to make the product indispensable, not just nice to have) – I’d expect to see more colours of HomePod, and lower-priced ones too.

It took Sonos years to diverge from its high-end music amps down to the Sonos One, but it’s the latter that was the hit because it found the sweet spot on price. The HomePod is more versatile than the Apple TV because it has more functions than just displaying content on a screen. It’s a voice-driven speaker, and that has lots of implications.


So that’s my thinking: adding Spotify is an open goal, HomePod 2 is a certainty, and we could see smaller HomePods in time if Apple decides that this is a market which is worth winning, rather than just taking part in (the latter being its approach with the Apple TV).

But the early sales numbers? They don’t tell us a lot. Because to lean on those as telling the story means to ignore the ironclad rule of Apple products: the first is the MVP. It’s the ones after that which tell you the trajectory of the device.

Start Up: Zuck’s part 2, adblocking politics, how YouTube creates human clickbait, China’s AR challenge, and more

Dots: Netflix can’t see them in email, Gmail can. Bad news. Photo by Leon Lee on Flickr.

A selection of 10 links for you. Augment that reality. I’m @charlesarthur on Twitter. Observations and links welcome.

Zuckerberg denies knowledge of Facebook shadow profiles • TechCrunch

Taylor Hatmaker:


Commerce Committee, New Mexico Representative Ben Lujan cornered Mark Zuckerberg with a question about so-called “shadow profiles” — the term often used to refer to the data that Facebook collects on non-users and other hidden data that Facebook holds but does not offer openly on the site for users to see.

In one of the handful of slightly candid moments of the past few days, Rep. Lujan pressed Zuckerberg on the practice today:

Lujan: Facebook has detailed profiles on people who have never signed up for Facebook, yes or no?

Zuckerberg: Congressman, in general we collect data on people who have not signed up for Facebook for security purposes to prevent the kind of scraping you were just referring to [reverse searches based on public info like phone numbers].

Lujan: So these are called shadow profiles, is that what they’ve been referred to by some?

Zuckerberg: Congressman, I’m not, I’m not familiar with that.


Lujan really takes Zuckerberg to task. It’s quite a thing to see. The House of Representatives, despite the funding they get from Facebook, were much tougher than the Senate; partly because they’re younger, but also had wider experience – including one with a computer science degree (which is more than Zuck has).
link to this extract

Bitcoin would be a calamity, not an economy • MIT Technology Review

James Surowiecki:


The problem with a world in which there are lots of different private currencies is that it massively increases transaction costs. With a single, government-issued currency that’s legal tender, you don’t have to think about whether or not to accept it in exchange for goods and services. You accept dollars because you know that you will be able to use them to buy whatever you want. Commerce flows more smoothly because everyone has implicitly agreed to use the dollar.

In an economy with lots of competing currencies (particularly cryptocurrencies unbacked by any commodity), it would work very differently. If someone wants to pay you in Litecoin, you have to figure out whether you think Litecoin is a real cryptocurrency or just a scam that could shut down any day now. You have to consider who else might accept Litecoin if you want to spend it, or who would trade you dollars for it (and at what exchange rate and transaction fee). Basically, a proliferation of currencies tosses sand into the gears of commerce, making transactions less efficient and more costly. And any currency that is hard to use is less valuable as a medium of exchange.

This isn’t speculative. we actually have a historical example of how this works. In the United States in the decades before the Civil War, there was no national currency. Instead, it was an era of what was called “free banking.” Individual banks issued bank notes, theoretically backed by gold, that people used as money. The problem was that the farther away from a bank you got, the less recognizable (and therefore the less trustworthy) a bank’s note was to people. And every time you did a deal, you had to vet the note to make sure it was worth what your trading partner said it was worth. So-called wildcat banks sprang up, took people’s money, issued a host of notes, and then shut down, making their notes worthless. To be sure, people came up with workarounds—there were volumes that were a kind of Yelp for banking, displaying the panoply of bank notes and rating them for reliability and value. But the broader consequence was that doing business was simply more complicated and slower than it otherwise would have been. The same will be true in a world where some people use Ethereum, others use Litecoin, and others use Ripple.


link to this extract

Tesla issues strongest statement yet blaming driver for deadly crash •

Dan Noyes on the latest regarding Walter Huang, who died when his Tesla, on Autopilot, drove into a crash barrier:


Tesla sent [ABC News] a statement Tuesday night that reads in part, “Autopilot requires the driver to be alert and have hands on the wheel… the crash happened on a clear day with several hundred feet of visibility ahead, which means that the only way for this accident to have occurred is if Mr. Huang was not paying attention to the road.”

“We know that he’s not the type who would not have his hands on the steering wheel, he’s always been (a) really careful driver,” said [Walter Huang’s brother] Will.

The family’s lawyer believes Tesla is blaming Huang to distract from the family’s concern about the car’s Autopilot.

“Its sensors misread the painted lane lines on the road and its braking system failed to detect a stationary object ahead,” said lawyer Mike Fong.

You can already see the arguments forming for the lawsuit.


If Huang had driven down the road before in the same car in the same way, Tesla will have records. If this happened after a software update, it’s Tesla’s fault: Huang would have had a reasonable expectation that the car would (as previously) avoid the obstacle. (Recall the videos of how this could happen from a few days ago.)
link to this extract

Your Facebook data is only worth $5.20 on the dark web – MarketWatch

Maria LaMagna:


Whether or not you were impacted by the Cambridge Analytica incident, there’s a depressing aspect of many recent privacy violations: The most important parts of your identity can be sold online for just a few dollars.

Consumers have to spend hours of their time — and, sometimes, their own money — when they find out their driver’s license, Facebook “likes” or Social Security number have been exposed to hackers. But those who sell them are making only petty cash.

That’s according to a new report from the content marketing agency Fractl, which analyzed all the fraud-related listings on three large “dark web” marketplaces — Dream, Point and Wall Street Market — over several days last month.

The “dark web” is part of the internet that people can only access by using special software. To create this report, Fractl accessed the dark web through the browser Tor. People buy other risky or illegal substances on the dark web, including drugs, pirated content like movies or music and materials that help with scams, including credit-card “skimmers.”

Facebook logins can be sold for $5.20 each because they allow criminals to have access to personal data that could potentially let them hack into more of an individual’s accounts. The credentials to a PayPal account with a relatively high balance can be sold on the dark web for $247 on average, the report found.


link to this extract

The dots do matter: how to scam a Gmail user • James Fisher

Fisher got a valid email from Netflix saying it was having trouble with his credit card payment. He was going to update it – but the credit card it had didn’t match his own. What gives?


I finally realized that this email is to I normally use, with no dots. You might think this email should have bounced, but instead it reached my inbox, because “dots don’t matter in Gmail addresses”:

If someone accidentally adds dots to your address when emailing you, you’ll still get that email. For example, if your email is, you own all dotted versions of your address:

Netflix does not know about this Gmail “feature”. Externally, and are different identities, and should have their own Netflix accounts. I signed up for Netflix account N1 backed by in 2013. But in September 2017, someone, let’s call her “Eve”, created a new Netflix account N2, backed by

Eve has access to account N2 because she set its password when signing up, but I also have access to the account because I own, and so I can follow the password reset process for this account. I did so.

Eve loves her TV! She’s watched 587 titles in six months, all from her “Android Device” in Alabama. She watched three seasons of Trailer Park Boys over a single day in October. She consumed nearly every day until 22nd March, when Netflix put her account “on hold” due to payment failure. Eve had paid for these shows. She paid $13.99 every month for her Premium plan, until February when her card **** 2745 (also billed to Huntsville, Alabama) was declined.

Perhaps this was all a mistake? Perhaps Eve is actually one of the twelve James Fishers in Huntsville, AL, and perhaps he typed his email address in wrong when he signed up months ago. Netflix doesn’t do any email address verification when you sign up; you can start watching shows straight away.

But perhaps this was not a mistake but a scam. I was almost fooled into perpetually paying for Eve’s Netflix access, and only paused because I didn’t recognize the declined card.


Google is proud of this “feature”, but like Fisher, I think it’s a bug. I get tons of scam emails like this.
link to this extract

Trump’s company is suing towns across the country to get breaks on taxes • ProPublica

Katherine Sullivan:


Since becoming president, Trump’s companies have filed at least nine new lawsuits against municipalities in Florida, New York and Illinois, arguing for lower tax bills, ProPublica has found. Some of those lawsuits have been previously reported. At stake is millions of dollars that communities use to fund roads, schools and police departments.

Real estate owners dispute property taxes frequently, and some even sue. The president has a long track record of doing so himself. But experts are troubled that he’s doing so while in office.

No president in modern times has owned a business involved in legal battles with local governments. “The idea that the president would have these interests and then those companies would sue localities is really a dangerous precedent,” says Larry Noble, of the nonpartisan Campaign Legal Center. The dynamic between local and federal governments is impossible to ignore in these cases, says Noble. Municipalities “rely on resources from the federal government and the federal government can make your life easier or much more difficult.” The concern arises because the president did not fully separate from his businesses, he says.

A spokesman for the Trump Organization said, “Like any other business or property owner when property taxes become inflated it is not uncommon to challenge the process to ensure fair treatment. This is a routine practice and any suggestion otherwise is simply ridiculous.”


And also obvious. What happens when the president doesn’t disentangle himself from his companies.
link to this extract

One woman got Facebook to police opioid sales on Instagram • WIRED

Nitasha Tiku:


Eileen Carey says she has regularly reported Instagram accounts selling opioids to the company for three years, with few results. Last week, Carey confronted two executives of Facebook, which owns Instagram, about the issue on Twitter. Since then, Instagram removed some accounts, banned one opioid-related hashtag and restricted the results for others.

Searches for the hashtag #oxycontin on Instagram now show no results. Other opioid-related hashtags, such as #opiates, #fentanyl, and #narcos, surface a limited number of results along with a message stating, “Recent posts from [the hashtag] are currently hidden because the community has reported some content that may not meet Instagram’s community guidelines.” Some accounts that appeared to be selling opioids on Instagram also were removed.

The moves come amid increased government concern about the role of tech platforms in opioid abuse, and follow years of media reports about the illegal sale of opioids on Instagram and Facebook, from the BBC, Venturebeat, CNBC, Sky News and others. Following the BBC probe in 2013, Instagram blocked searches of terms associated with the sale of illegal drugs.


Zuckerberg was asked about opioid adverts on Facebook by the House of Representatives committee; he said (paraphrased) they couldn’t do much and that they’d have to wait for better AI.
link to this extract

Ad blocking as a radical political act • Terence Eden’s Blog

Terence Eden:


Aside from unavoidable billboards and the occasional magazine, I just don’t see advertising any more. I’m not sure why any sane person would want to.

Even when I worked in the mobile ad industry, I blocked ads. Everyone did. The first thing that the IT helpdesk said to people who complained that they couldn’t log into their work email was “yeah mate, you need to turn your ad-blocker off…”

I’ve been blocking Facebook adverts since before it was fashionable. As a result, I’m bemused by the claims that my information has been microtargetted and used to manipulate me.

I thought it was common knowledge that you could set your Facebook preferences to block creepy use of your data for advertising purposes. Even if you didn’t want to block adverts, why wouldn’t you do that?

Perhaps Facebook themselves have been subtly manipulating what stories they choose to show me. Perhaps my friends are activated Manchurian Candidates swamping me with fake news. Or perhaps I just block the obviously dodgy news sources and unfriend anyone daft enough to share them.
Perhaps we need a word to describe the people who willingly watch adverts? The technology to block them is simple to use, and information about blocking is widely disseminated.

People who watch adverts are like anti-vaxxers – blissfully unaware of the benefits of herd-immunity.


Advertisers (and a lot of publishers) see it quite the other way round, of course.
link to this extract

China is catching up to Apple in AR, says KGI Securities’ Ming-Chi Kuo • Business Insider

Kif Leswing:


The most-closely followed Apple analyst warned in a Wednesday note that Chinese smartphone companies are rapidly catching up to Apple in augmented reality technology, which CEO Tim Cook has called “profound” and a “core technology” for the company going forward. 

The example Ming-Chi Kuo provides is a Tencent game called Honour of Kings, which will release an augmented reality version in May. It’s a big game, with over 200 million players worldwide.

It’s also a much more advanced augmented reality experience than Pokemon Go, he writes, and uses algorithms from $3bn artificial intelligence startup SenseTime.

“Apple’s first-mover lead in AR eroded by OPPO,” reliable Apple analyst Ming-Chi Kuo at KGI Securities wrote.

Apple launched ARKit last summer, which easily allows developers to make rich experiences where computer models interact with surfaces in the real world. Apple was the first major technology company to announce software like that, and had a chance to capture the entire development market. 

“However, since the debut of the ARKit nearly a year ago, there has been no heavyweight AR application on iOS,” Kuo wrote. 

Which is why he believes Apple should be concerned that it’s launching on Oppo phones running Android at the same time as iOS, on less-advanced hardware.


SenseTime again. However, I think that AR’s struggles (Pokemon Go aside) are going to remain the same: is it as engaging to have a virtual object in the real world as it is to have a virtual object in a virtual world that you control more precisely?
link to this extract

Your pretty face is going to sell • Open Space at SF MOMA

Joe Veix on the peculiar phenomenon of “YouTube Face” (YTF) – the strange, overplayed expressions that you see people adopting on videos in order to make arresting preview frames for the time when they’re in the “up next” lineup and want to be chosen, oh please choose me, for the next click:


Getting attention on social media platforms requires creating content designed to perform well within their ecosystems. Everything must contort to please the almighty Algorithmic Gods. It requires some guesswork, as these algorithms exist at such an ever-increasing scale and complexity that even their creators don’t — can’t — understand them. The Algorithm Gods work in mysterious ways.

This has odd and often unexpected effects on the physical world. Restaurants attempt to create Instagram-friendly environments with nauseatingly kitschy interior designs. Hamburger buns are glazed to make them more aesthetically appealing. Extremist political campaigns are won partially on the strength of their shitposting. Perhaps the emergence of YTF hints at one of the many ways these algorithmic forces might begin to shape our physical appearances.

We’re also witnessing tactics common to the advertising industry, especially those of late-night infomercials, being utilized autonomously by individuals. People simulate the behavior of corporate brands, while corporate brands simulate people, hiring teams of flacks to help make something like, I don’t know, fracking seem “authentic” and “cool.”

So begins the Great Brand Singularity. Corporations, humans, and machines merging in a banal orgy of commerce. The tech is currently primitive, but it’s easy to imagine scrolling through some future feed and seeing the faces of long-deceased relatives digitally grafted onto advertisements for #FappuccinoHappyHour; close friends suddenly revealed to be replicants working for foam mattress startups; augmented reality Pillsbury Doughboys stalking us on late night walks home, their soft footsteps squishing confidently along.


Or as he also says, “YouTube Face is clickbait made human”.
link to this extract

Errata, corrigenda and ai no corrida: none notified

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

Start Up (holiday bonus): Facebook redux, what the Uber car should have seen, the fake NHS pay rise, and more

That’s a forgery! But will the blockchain spot it? Photo by Yersinia pestis on Flickr.

Today’s is just a bonus, because honestly, there’s a lot of tech stuff that needs noting, isn’t there? But truly, I am on holiday as you read this. I just wasn’t when I wrote it.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 12 links for you. Use them wisely. I’m @charlesarthur on Twitter. Observations and links welcome.

Bitcoin will eventually be the single global currency: Twitter’s Jack Dorsey • CNBC

Ari Levy:


Jack Dorsey, the CEO of Twitter and Square, expects bitcoin to become the single global currency within the next decade, he told the Sunday Times newspaper.

Dorsey, a personal investor in bitcoin, expects the cryptocurrency to be used for simple things like coffee and said its ascendance to world’s currency will occur over 10 years, “but it could go faster,” the U.K.-based paper reported.

Square said in November that it would start enabling the buying and selling of bitcoin on its Cash app. Dorsey is also an investor in a star-up called Lightning Labs, which is developing technology to make bitcoin faster and easier to use.

When it first came into use, Bitcoin was touted as an alternative to the dollar and even gold. However, the cryptocurrency has been on a wild ride in recent months, soaring to a record near $20,000 before crashing below $8000 last month.

Dorsey told the Times that bitcoin is “slow and it’s costly, but as more and more people have it, those things go away.”


I’m perfectly happy to take the other side of that bet and come back in ten years. How about you, Jack?
link to this extract

Facebook scraped call, text message data for years from Android phones • Ars Technica

Sean Gallagher:


If you granted permission to read contacts during Facebook’s installation on Android a few versions ago—specifically before Android 4.1 (Jelly Bean)—that permission also granted Facebook access to call and message logs by default. The permission structure was changed in the Android API in version 16. But Android applications could bypass this change if they were written to earlier versions of the API, so Facebook API could continue to gain access to call and SMS data by specifying an earlier Android SDK version. Google deprecated version 4.0 of the Android API in October 2017—the point at which the latest call metadata in Facebook users’ data was found. Apple iOS has never allowed silent access to call data.

Facebook provides a way for users to purge collected contact data from their accounts, but it’s not clear if this deletes just contacts or if it also purges call and SMS metadata. After purging my contact data, my contacts and calls were still in the archive I downloaded the next day—though this may be because the archive was still the same cache I had requested on Friday.

As always, if you’re really concerned about privacy, you should not share address book and call-log data with any mobile application. And you may want to examine the rest of what can be found in the downloadable Facebook archive, as it includes all the advertisers that Facebook has shared your contact information with, among other things.


Jelly Bean was released in September 2012, but it took until October 2013 for that version (or later) to be on more than 50% of Android phones.
link to this extract

Fact check: your call and SMS history • Facebook Newsroom


You may have seen some recent reports that Facebook has been logging people’s call and SMS (text) history without their permission.

This is not the case.

Opt-in features in Facebook Lite and Messenger
Call and text history logging is part of an opt-in feature for people using Messenger or Facebook Lite on Android. This helps you find and stay connected with the people you care about, and provide you with a better experience across Facebook. People have to expressly agree to use this feature. If, at any time, they no longer wish to use this feature they can turn it off in settings, or here for Facebook Lite users, and all previously shared call and text history shared via that app is deleted. While we receive certain permissions from Android, uploading this information has always been opt-in only.

We introduced this feature for Android users a couple of years ago. Contact importers are fairly common among social apps and services as a way to more easily find the people you want to connect with. This was first introduced in Messenger in 2015, and later offered as an option in Facebook Lite, a lightweight version of Facebook for Android.


Unsigned. Isn’t going to make it any more welcome. “Yeah, you agreed to that in the gazillion-page agreement. Remember? OK so it looked like something else. Get over it.”
link to this extract

How can I download a copy of my Facebook data? What is included – and what isn’t? • Big Brother Watch

You can download your information from your settings. To download your information:

1. Click at the top right of any Facebook page and select “Settings”
2. Click “Download a copy of your Facebook data” at the bottom of General Account Settings
3. Click “Start My Archive”
You will be prompted to confirm that you have requested the archive from your associated email account.

This archive will typically contain a large amount of very sensitive personal information, including contact information, addresses, photos and private messages (see below). You should be careful to store it securely.

I now have my Facebook archive. Where can I find the contact information it has stored about me?

See the ‘contact info’ tab under ‘html’. If you have closely controlled your privacy settings, you won’t see much here.

However, many people find comprehensive contact details from their phone and email accounts.

Some even find extensive call and text logs, likely to arise from app permissions that have been granted.

Why does my contacts list include people that are not on Facebook?

When you first sign up to Facebook, you are asked to hand over your contact lists and address books so Facebook can “Find Friends” for you.
link to this extract

‘Oh my God…It’s fake’: Far right falls for hoax about Broward County sheriff • POLITICO

Marc Caputo:


In the wake of the Marjory Stoneman Douglas High School massacre, the far-right fever swamps buzzed with false information and conspiracy theories about student “crisis actors” who were paid to lie about the mass shooting.

But ironically, conspiracy-minded conservatives fell for a political hoax involving a different kind of actor. The subject? Broward County’s Democratic sheriff, Scott Israel.

Israel for the past month has been assailed as everything from a “rapist” to a philanderer to a crooked cop thanks to three old YouTube videos in which a mystery woman accused him of impregnating her when she was 17 and forcing her to get an abortion. The videos together have been viewed almost 130,000 times since the Feb. 14 shooting.

But all of it was a lie, the woman and her attorney, Yechezkel Rodal, now tell POLITICO, which found her by combing internet videos and social media.

“I was paid to say these things. I didn’t even know what I was saying,” said the woman, who spoke with POLITICO on condition of anonymity because she fears political retribution from Internet trolls or from the sheriff’s office, which does not know her identity. “I’m sorry … It’s fake.”

The revelation comes amid growing concerns about the spate of conspiracy theories and “false flag” attacks surrounding recent mass shootings — especially in Florida — that are surfacing on right-wing and fringe media sites.


This happens at both extremes of political belief, of course.
link to this extract

Despite its mystique, Cambridge Analytica didn’t offer advertisers anything special • AdExchanger

James Hercher:


One agency found Cambridge Analytica was effective for campaigns with specific parameters and targets. The agency used Cambridge Analytica for a campaign heavy on earned media after it claimed it could drive new cycles and engagement.

“It worked, but we chose them because we knew we were targeting a Trump-like audience and they’d have models for that,” said the agency exec, who hasn’t worked with Cambridge Analytica since.

Cambridge Analytica was fairly effective, according to an executive from a news publisher that piloted a subscription campaign with the company, but the program was dropped because it was more expensive than similar optimization tech companies on the market.

Where Cambridge Analytica found success and longer-term work was in Washington, DC, where it positioned itself as an outside commercial option for Republican candidates losing the narrative on data and technology.

Besides need, the Republicans also presented opportunity. They had fewer vendors compared to the Democratic ecosystem, according to a former Cambridge Analytica executive and a digital media executive who worked closely with the company during the election.

“Republican candidates and committees had frankly been overpaying conservative vendors for a long time because really no competition was allowed,” said one political tech executive who worked closely with Cambridge during the campaign and refused to comment publicly due to a nondisclosure.

Cambridge Analytica’s technology may have been standard market fare, he said, but it was competing with overpriced platforms that had long attached big premiums to conservative media buys based on a vague sense that campaigns should have a more political-first media approach and, mostly, out of partisan loyalty.

“The truth is, Facebook or about any commercial DMP can do that better even if their employees want you to lose,” he said.


AdExchanger doesn’t want to tell us what DMP is. Jargon for “data management platform“, since you ask.
link to this extract

Indian agency denies security lapse in ID card project; ZDNet defends report • Reuters

Malini Menon:


Tech news site ZDNet said on Sunday it stood by its report that identified a security vulnerability in data-linked to Aadhaar – India’s national identity card project, after a semi-government agency that manages the database sought to discredit the report.

ZDNet reported that a data leak on a system run by a state-owned utility company could allow access to private information of holders of the biometric “Aadhaar” ID cards, exposing their names, their unique 12-digit identity numbers, and their bank details.

The Unique Identification Authority of India (UIDAI), which manages the Aadhaar program, said “there is no truth in this story,” in a statement late on Saturday.

ZDNet’s global editor-in-chief Larry Dignan said in an email to Reuters on Sunday the publication stood by its report. Dignan said they spent weeks compiling evidence and verifying facts.

“We spent weeks reaching out to the Indian authorities, specifically UIDAI, to responsibly disclose the security issue, and we heard nothing back — and no action was taken until after we published our story,” said Dignan.

UIDAI sought to downplay the report stating that even if the claims in the story were true, it would raise security concerns with the database of the utility company and not with the security of UIDAI’s Aadhaar database. UIDAI said it is “contemplating legal action against ZDNet”.


There have been so many reports of Aadhaar breaches that they can’t all be fake.
link to this extract

#PutinAtWar: trolls on Twitter • Medium

The Atlantic Council’s Digital Forensic Lab looks at how a poll about the Skripal poisoning by a British user was hijacked by a Russian account which spread it to others bots:


From Lisitsa, the retweet cascaded to dozens of other, primarily Russian-language accounts, forming the most substantial cluster of retweets throughout the scan.

None of these Russian accounts has an organic focus on, or interest in, UK politics; their content is dominated by pro-Kremlin messaging, mostly in Russian or English. Their purpose in retweeting the poll therefore seems to have been to spread it to a Russian audience which could be expected to vote against the UK government.

This intervention was small in itself, impacting one poll, from one account. However, the source account was an influential member of a politically vocal UK community; thus, by targeting it, the Russian accounts may have hoped to reinforce their message among UK opposition supporters.

If so, they succeeded. @Rachael_Swindon is not a member of this troll community; it has had no interactions with @malinka1102 or @rixstep, and does not post on hot-button Kremlin topics such as Crimea or MH17.

However, still on March 17, the account had a conversation with @ValLisitsa, at the end of which @Rachael_Swindon claimed, based on its own poll, that the “mood of the British public is starting to shift.”


If these researchers find it this easy to find Russian trolls, why can’t Twitter? Also, anyone who takes the slightest notice of a Twitter poll needs telling off.
link to this extract

Why you need an Untouchable day every week • Harvard Business Review

Neil Pasricha:


Now when I get home after work, I soak in time with my wife and two little boys. Nothing is or will ever be as precious to me, and I resist insight from anyone who isn’t making space for loved ones.  I realized that what I needed was a practical way to get more work done without taking more time. And, to be honest, I needed it fast. Why? Because in my first year as a full-time author, I actually started feeling my productivity slipping — even though I had quit my full-time job. It wasn’t just disheartening; it was also embarrassing. “So how’s the new book coming?” “Oh, now that I quit my job? Terribly!”

I finally found a solution that I feel has saved my career, my time, and my sanity. If you’re with me right now, I bet you need this solution too: I call it “Untouchable Days”.

These are days when I am literally 100% unreachable in any way…by anyone.

Untouchable Days have become my secret weapon to getting back on track. They’re how I complete my most creative and rewarding work. To share a rough comparison, on a day when I write between meetings, I’ll produce maybe 500 words a day. On an Untouchable Day, it’s not unusual for me to write 5,000 words.  On these days, I’m 10 times more productive.

How do I carve out Untouchable Days?

I look at my calendar sixteen weeks ahead of time, and for each week, I block out an entire day as UNTOUCHABLE. I put it in all-caps just like that, too. UNTOUCHABLE. I don’t write in all-caps for anything else, but I allow  UNTOUCHABLE days to  just scream out to me.

Why sixteen weeks ahead? The number of weeks isn’t as important as the thinking behind it. For me, that’s after my speaking schedule is locked in — but, importantly, before anything else is. That’s a magic moment in my schedule. It’s the perfect time to plant the Untouchable Day flag before anything else can claim that spot.


Not sure this works for those who aren’t authors who aren’t obliged to go to meetings and offices, but included just in case you’re looking for a new way to make your boss say “You’re WHAT?”
link to this extract

Police chief said Uber victim “came from the shadows”; don’t believe it • Ars Technica

Timothy B. Lee:


In this nighttime video, posted to YouTube by Brian Kaufman on Wednesday, the scene of the crash can be seen around 0:33. Features at the sides of the road—including curbs, signs, and bushes—are clearly visible. No pedestrians walk into the road during the video, but it seems clear that Herzberg would have been visible much earlier if the Uber video had been taken with this camera.

Mill Ave. at night.
Another YouTuber, Dana Black, posted this video. His camera work isn’t as good as Kaufman’s—the video is blurry and he doesn’t hold his camera steady. But his video supports the same basic conclusion. “It’s not as dark as that video made it look,” Black says in the video as he drives past the point in the road where Herzberg was hit (around 0:33). “My footage is from my Pixel XL and looks pretty similar to real life,” he writes in the YouTube description.

To be fair, there are a few other cars on the road in Black’s video, which might be adding some illumination. But Kaufman’s car appears to be the only vehicle on the road, and visibility is still much better than in Uber’s dashcam video.

It’s not surprising that the road was actually more brightly lit than the Uber video makes out. Think about it: the Uber car was going 38 miles per hour (61km/h), and people on pitch-black country roads drive faster than that all the time. That would be extremely reckless if—as the video implies—headlights can’t illuminate the road two seconds ahead at that speed.

The video implies that the Uber car’s headlights had a range under 110 feet (33 meters). For comparison, here’s a diagram from the Insurance Institute for Highway Safety showing headlight ratings for the car in question, a Volvo XC90:


IIHS shows the XC90 with a range just under 250 feet (76 meters) with “low beams” on. The car’s headlights are rated poorly by the IIHS compared with other cars on the market. Still, 250 feet is more than 4 seconds of illumination for a car driving 38 miles per hour. If the Uber car’s headlights really didn’t illuminate Herzberg until less than two seconds before the crash, there was something seriously wrong with them.


As I said previously, cameras don’t give you a good idea of how people see them, but the Uber dashcam really seems to be making it look a lot darker than it was. Uber doesn’t have an alibi.
link to this extract

Those eye-popping 6.5% to 29% NHS pay rises are a lie – and I can prove it •

Rachel Clarke is an NHS doctor, but used to be a journalist for ITV:


Pages 11-13 of the Framework Agreement purport to demonstrate, for each level of seniority of NHS staff, their “new” pay rise over three years. To expose the statistical sleights of hand deployed, take the example of staff on “point 24” of the payscale, screenshot below:

The total pay rise for a staff member on this point of the payscale would be, allegedly, 14.02% over three years. However, during that same three-year period, their pay would have risen anyway on the old payscale by 10.48% (from £29,626 to £32,731), as they received their annual incremental pay awards, reaching point “27” on the old payscale. In other words, their actual pay rise on the proposed new pay deal is a mere 3.54%, spread over three years.

That’s not even close to the promised minimum pay rise of 6.5%. It’s barely greater than 1% per annum.

Deploying the same simple arithmetic with the outlandish-sounding upper limit pay rises reveals, again, the dishonesty of the government’s figures. Let’s look at that alleged 29% pay rise. Here are those lucky individuals, on point “26” of the payscale:

But, once you deduct the increase in salary these staff members would have received anyway on the old payscale (from £31,696 to £35,577 = 12.24%), you find the headline figure of 29% shrinks down to an actual pay rise of 16.8%.

In short, the government – and the 13 unions who have agreed to sign up to these bogus figures, with the notable exception of the GMB – have misled NHS staff into thinking their pay rises over the next three years are vastly greater than they actually will be.


How surprising that the government would misrepresent a pay award in a way that favours it. Meanwhile I highly recommend Clarke’s book “Your Life In My Hands“.
link to this extract

This startup is using blockchain to fight art forgers • Bloomberg

Vivianne Rodrigues and Rob Urban:


Contemporary artist Philip Colbert, whose colorful, high-spirited art is finding buyers around the world, had been toying with the idea of creating his own catalog system to prove the authenticity of his expanding body of work.

“I had a dealer in Japan who had been telling me I needed to have better forms of certification for my artwork, because people are buying art as an investment,” said the British artist, who appropriates pop culture images in his paintings, fashion and furniture. “Art is a currency in a way; at the end of the day when they come to auction, the provenance is a very important element of their value.”

Then he met Rob Norton, the founder of Verisart, a U.S.-based startup that’s using blockchain, the ledger technology underlying Bitcoin, to verify the authenticity of artwork. It’s a problem as old as art itself, said Norton, and artists have long been unreliable when it comes to documenting their own work. As far back as the 17th century, Rembrandt’s dealer complained of his client’s poor record-keeping, Norton said.

Blockchain creates an immutable, traceable record of every transaction, whether it’s art changing hands or Bitcoin. Widespread adoption of the technology could give a boost to the market for art online, which has yet to explode…

…Colbert’s certificates, for example, contain small reproductions of the piece itself called “image hashes,” along with all of the relevant information about its creation, ownership and movement, such as whether it was part of an exhibition. He’ll have a show in Tokyo in September and Beijing next February.

Since Verisart uses the unaltered Bitcoin blockchain rather than a customized version, one risk may be that their effort can be easily replicated, since it brings little in the way of new technology. Some collectors, particularly those who buy and sell privately may also be reluctant to share their information in such a public way.

“The blockchain is a more efficient method of verification,” Colbert said. “You’re not worried about the authentic value of your work, because it’s all about locking down the time and place. Then all those fakes aren’t doing you any damage. All those fake Mona Lisas don’t do the Mona Lisa any harm.”


But how do you know that the image of the original thing that you hashed is authentic? Art faking often starts right at the point where the art enters the system. The first buyer thought it was a Monet; turns out it was a fake all along. Now do you do to the blockchain entry?
link to this extract

Errata, corrigenda and ai no corrida: none notified

Beware taking your Mac running High Sierra to the Genius Bar: APFS might surprise them

“Your Mac. On top”, according to the caption. But it might also be a source of confusion at the Genius Bar. Photo by tua ulamac on Flickr.

I wrote a while back about the problems I had with my 2012 retina MacBook Pro, and its strange shutdowns – which I suspected, but couldn’t absolutely prove, were due to the graphics card problem that these models have been known to suffer from: when the discrete graphics card was activated, there was a chance it would go completely off the rails.

Finally it shut down and didn’t seem to want to start. At that point, I turned over to using the iPad Pro full-time, which is another story. You might find it entertaining.

The only way to absolutely prove that the problem with the computer was the graphics card, of course, was to take it to a Genius Bar. After eventually getting an appointment (the Mac Geniuses are rare, compared to the iPhone/iPad Geniuses), I turned up with the rMBP which I’d left for dead.

“It was dead, honest”

Of course the first thing it did when the guy plugged it in was to start up and happily proceed to the login screen. Talk about embarrassing. I logged in. All seemed OK. But we were there for a Genius appointment, so we were going to do that. He restarted, booted from a network drive which has Apple’s Mac OS9-style hardware test, and gave it a once-over.

Mac OS 9: if you take your Mac to get Geniused, you’ll probably see a screen that looks something like this. Photo by Rodrigo Vera on Flickr.

(That’s OS9. It’s not what my machine looked like, except for how the progress bar looked.)

The checks all looked OK, except for the VST – video system test. That didn’t work – it said it couldn’t find the relevant drivers (which it looked for on the hard drive). So he ran Disk Utility off the network drive to see whether there was something wrong with the hard drive. (rMBPs from that time all have SSDs.).

At this point things got very confusing. Disk Utility showed that there were two drives – an SSD with 500GB capacity, and another also with 500GB capacity. One of them seemed to be called Macintosh HD, and the other seemed to have partitions such as “private” and “mem”.

Except that the rMBP only had a 500GB SSD. How had it got two? “Have you got a Fusion drive in here?” the Genius asked.

Me (cooperative citizen pulled over by policeman): “Er.. no.”

At this point the Genius said he suspected there was something wrong with my SSD. So he restarted the machine, held down the key to boot from a network drive, whizzed over to the one marked 10.12 and started up. Could Disk Utility read my drive now? No, it couldn’t.

“Do you have this backed up?” he asked, indicating the computer. I sure did – SuperDuper clones, and a Time Machine backup. “I think we might need to reinstall the operating system.”

(This is the point at which the smart readers are saying “ohhhh” because they’ve spotted the mistake that both he and I made.)

“You backed it up, of course?”

So yes, it was time to wipe the drive and reinstall everything. Thankfully the Wi-Fi in Apple Stores is really fast, so you can get your OS reinstalled quickly. Everything seemed to be going fine. I hung around, knowing that all the important stuff was backed up either in the cloud or on the backup disk. For fun, because I knew that it would take forever on our super-slow broadband, I decided to re-download my iCloud Photo library. All 16,800 or so of them. Look, the Wi-Fi’s free.

The photo download was fast. Everything seemed to be fine. The machine wasn’t crashing. I was now starting to question myself: was it really just something scuzzy on the hard drive?

Then, fatefully, I tried to look at a photo that had downloaded at full resolution. Click on the photo, it begins to fill the screen, and––

BAM. The machine shut down, just like that. It had been nothing to do with the hard drive or the OS. As I thought, whenever the discrete graphics card was called on to do something, it knocked the machine out.

Some more consultation, and another Genius. More network booting, and this time the video test seemed to indicate that, ehhhh, the video card might not be in good shape.

At this point I suddenly realised why the network-booted versions hadn’t been able to run the video card drivers from my hard drive, and now they could. The network drives were formatted with HFS+, which is the file system Apple has been using on Mac OSX (and MacOS) for a couple of decades. That’s why the version number on the network drive was 10.12 – the file format for Sierra.

13 is the magic number (it makes stuff disappear)

High Sierra! It was staring us in the face. Sorta.

But I had brought the machine in running 10.13 – High Sierra – which uses the APFS filesystem. APFS brings a number of benefits, such as “instant” file copying, disk space saving, better encryption performance, inherent SSD trim, and faster boot times, but it isn’t back-compatible with HFS+; if you boot from an HFS+ drive, it won’t be able to read a file encoded in APFS. (APFS can work read HFS+, of course.)

That, of course, is why the Genius’s HFS+-encoded 10.12 network drive couldn’t understand my APFS-encoded SSD. So I’d wiped the hard drive for nothing. If there had been an APFS-encoded network drive, it could have booted up the machine and run the video test and found the problem. (Well, probably.)

Anyhow – they took the machine in and replaced the entire logic board, and they did it under the guarantee (even though that had run out a little while earlier). In my case, the problem had gotten much worse after I got the battery replaced in November 2017; possibly doing that disturbed the logic board and the card’s connections, and led to a cascade of trouble.

When I got home, there was an email waiting for me from Apple – the standard “how did we do?” questionnaire. I had by then realised what had happened, and so I filled it in, explaining that they needed to update their procedures to take High Sierra/APFS users into account. I got a followup phone call from the Apple Store where I’d taken my machine, and they sounded interested and honestly grateful for the feedback. So this has possibly already been implemented. (I mean, you’d hope, right?)

Problem mostly solved

So what’s the solution? Apple obviously needs to implement some network drives formatted with APFS. Which might mean an overhaul of how it does some stuff in-store; but it should expect that there are going to be more and more people coming in with machines that are APFS-encoded.

As for me – I came home with a machine that was fixed; it’s got a new battery (since November) and now a totally new logic board. What’s really nice is how clean the machine is. All the dust and crumbs and fingermarks are blown and cleaned off. It’s like getting a brand-new machine. (I think they may even have replaced my letter “A”, which had taken a lot of punishment over the past five and a bit years.)

Getting the hard drive back to status quo ante took an afternoon or so. The one thing that’s not working? Can’t get my old Time Machine drive (which is of course HFS+ because Time Machine hasn’t moved forward to APFS yet) to believe that the new machine is essentially the same one. If I try to run a backup, it thinks for 12 hours or so (it’s not an SSD backup drive) and then declares that it’s going to need vast tracts of storage. I’ve tried command-line invocations (yes, I’ve read the manual for tmutil) without success. Strangely, I am apparently able to restore files from the drive; I just can’t put fresh ones onto it.

But that’s not a huge problem, because I have hardly any files that live exclusively on my Mac – doing so seems like a liability, and a self-inflicted handicap in these days of cloud synchronisation via iCloud and Dropbox – and the very few that there are can be handled by SuperDuper!, which does bootable backups (yes, even of APFS).

Back to my Mac

Oh sure, you’re wondering: what’s it like being back on the Mac after weeks exclusively on the iPad? Three key things: it’s damn heavy; I can type a bit faster; the screen (15in) is really big, which makes it feel like all the fonts are gigantic and yet can fit stuff onto the screen.

More generally, I find it easier to get distracted on the Mac. There are pings from emails, there’s the time spent waiting for the newsreader to load articles or web pages to load, during which one may decide to go and look at another app (email, Twitter) and get lost in that, breaking the flow of whatever I was doing before. I’m much more focussed when working on the iPad – I can go most of a day without looking at email, which is how I like things to be, unless it’s VIP email, in which case I get pinged. Also, I don’t get spinning beachballs on the iPad. Sure, it’s newer and it’s doing less simultaneous processing, but even so, as a user interface experience, Beachballs Are Bad, and one notices them even more when returning from a platform where you never, ever see them.

But anyhow, if your Mac does break down, and you’re on High Sierra, make sure to tell them if you’re on APFS when they come to the diagnostics. And if they tell you that you need to wipe your drive and start again, just make sure to ask them: “are you certain it’s that, or could it be your network drive can’t read the APFS file system on my machine?” It can’t hurt to ask.

Life on an iPad

IPad Pro with Smart Keyboard, and devil

The Mac went to the day of the dead, so the iPad had to step in

A couple of weeks ago, I opened my Macbook Pro as usual. The keyboard lit up, as usual. I waited – there’s that pause while the display gathers itself (it’s a 2012 model) and the processor pulls everything together and presents the login window.

Except this time, nothing. The display didn’t light. There was the quiet sound of the fans going, but nothing. Oh dear. Closed the display, opened it to catch it unawares – no, that wasn’t going to fool it. After a bit more futzing around, I concluded that it was not in the mood to work. But I had work to do, and so I turned to my iPad Pro.

That was, as I say, a couple of weeks ago. Since then I’ve been doing everything I’ve done on this iPad – a 12in iPad Pro, with Smart Keyboard. That means email, writing articles for papers, editing chapters for my book, composing The Overspill’s daily Start Up post, and so on.

A few years ago, this would probably have been impossible. I wouldn’t have contemplated it. Now? Getting along fine. In a number of ways, the iPad is preferable – particularly weight and connectivity. In only a couple of ways is it worse (the most notable being “lappability”).

The big advantage these days is that if you trust your documents to iCloud, then moving between Mac and iPad isn’t a problem. (OK, for the chapters in the book that has been a mixture of iCloud – for the Pages edits of Word documents – and Dropbox, which is where Scrivener, which I used to write it originally, stores documents.) I was able to go straight to my iPad and have all the tools I needed.

Let’s run through a few of those things.

• writing my book: wait, you didn’t know I had a book coming out? Yes, in May – Cyber Wars, looking in detail at seven big hacking incidents: how and why they happened. You can order it on Amazon. (US, UK.) I wrote it in Scrivener, which is wonderful, and has an iOS as well as MacOS version. Documents and “projects” are synced via Dropbox, and it detects if you’ve done something in one place or the other and offers to sync them up.

• editing book chapters: the publishers sent back chapters as Word documents with Track Changes. Import those to Pages (on the iPad), run through the Track Changes, export to Word documents (also in iCloud Drive) and send back. All lives in iCloud Drive, so will be available when (if?) the Mac revives.

• writing articles for papers: this is generally easiest in Google Docs (because a lot of papers are on Google Apps). Weirdly, although Google will let you write things in Google Docs in Safari on MacOS, it absolutely won’t let you do that on Safari on the iPad, even if you request the desktop site. You have to use the app. This is the only case I’ve come across where you can’t do it on the site and have to use the app.

• making and recording Skype calls for work.

• curating The Overspill. This involves spotting links, selecting content from them, perhaps adding a comment and an image from that link, and then collating all the links together in a specific format (using particular HTML formatting), and putting those into a timed WordPress post with a Flickr CC-BY licensed picture at the top. (The CC-BY has to include a link back to the original photo and the photographer’s name or username.)

The most complex part of those is composing The Overspill, where I use a mixture of Instapaper, Pinboard and WordPress for the raw content. On the Mac, I collect the links and content and comment using the Javascript supplied by Pinboard. But that’s not available (or wasn’t) on the iPad, so I used Workflow to write an Action Extension: when I’m on a page, I select the text, hit the Share button and choose “Run Workflow”, and I can put the selected text – with a comment – into Pinboard. It’s actually better than on MacOS, because Workflow has options so you can grab the author name from the page meta-content.

Workflow script for iOS

Workflow (now owned by Apple) means you can script across applications on iOS

On the Mac, I compose the daily Overspill post using a custom Applescript I wrote (it queries Pinboard and posts to MarsEdit). Fortunately, well before the Mac went into a coma I’d translated the script into Python for iOS, using Pythonista – which is a damn useful program that lets you write and run Python programs which will interact with web pages, web APIs, and the OS itself. I wish there was something like it on MacOS; it makes writing Python programs to do tasks so much easier than doing them in the Terminal and other interfaces. (Pythonista apparently can also sync files between devices, as Workflow does, if you enable a setting – I wasn’t aware of it.)

Python script, in Pythonista, with console output

Pythonista on iOS means you can run Python scripts – it’s even more convenient than on the Mac

(Please don’t laugh at my coding. It just has to get the job done, not be pretty.)

I don’t do any podcast recording (which I understand is still a problem on iPads, as Garageband fights with Skype), nor any video editing. But what I’m trying to do is “real work”, at least for me. It’s work that earns money, and isn’t that what we’re after?

So a couple of weeks in, here’s what I find to be the good and the bad points of working full-time on an iPad to do things I used to do on a Mac laptop.

The good

the weight. It’s so much lighter (even with the Smart Keyboard) than a laptop. Put it in a bag and go and you hardly know it; that’s a big difference compared to toting around the MacBook Pro (5.6lb, v 1.6lb for the iPad plus 0.75lb for the Smart Keyboard – so half the weight).

• battery life. If it doesn’t last a couple of full days, then I’m disappointed and slightly surprised. Compare that with the laptop, where you’d expect to get a morning and an afternoon, and then be hunting for a power outlet.

• connectivity. I’ve got a PAYG sim from Three, with 1GB of credit, and I use that if I find myself somewhere without Wi-Fi: just hook into the mobile network. Yes, I know you can do this by setting up a hotspot from your phone to your laptop, but being able to have the device do it on its own is far more satisfying.

• focus. The iPad lets you work on two – max three – apps at once on the screen. If you tailor notifications correctly, you can get a lot done. So if I don’t want to be disturbed by email, then I don’t let it notify me, and I can go literally hours without being interrupted. (I don’t use email in the browser.) Then you go to your email and deal with it. Remember, you might think of it as “my inbox” but it’s actually composed of messages sent out of your control by other people. In general, “your” inbox is not under your control at all; it’s other peoples’ ideas of what you should do – a task manager compiled by other people. Not looking at email is good.

• aptitude. By which I mean that some of the scripts I write (with Workflow, with Pythonista) can do more than equivalents on MacOS. My Workflow one can get the name of the author of a page/article, which the standard Pinboard bookmarklet doesn’t. (Possibly a little bit of Javascript hacking could sort that, but when you roll your own you see the gaps in what you’re provided with.) On the Mac I use Viewfinder to get details of Flickr CC-BY photos, but the Pythonista script I’ve written gets the photographer name too, which Viewfinder doesn’t offer.

• the keyboard. I really like the keyboard. The odd thing is that I don’t much like the keyboard on the new MacBooks/MacBook Pros, but the Smart Keyboard uses the exact same key design. The crucial difference is that the Smart Keyboard covers them in a layer of fabric, which has two huge advantages: it makes them much quieter (because oh my lord the bare keys are CLACKY), and it proofs them against the specks of dust which have been the downfall of recent designs. Double win. If they could make a MacBook with these keys covered in fabric they’d have solved their problems, but I’m guessing that there would be thermal dissipation problems with that – most laptops vent plenty of heat out of the gaps around the keys.

The bad

• ‘lappability’. Laptops have the huge advantage that they’re designed to work in your lap: the big flat base sits on your lap and the screen can be adjusted to your taste, and then the keyboard has a solid base too. With the iPad and Smart Keyboard, it’s difficult to get the same effect, because it’s so light and the screen angle is fixed. (That wouldn’t be improved by the Microsoft Surface’s adjustable leg, because I can’t adjust the length of my femur to cope with where the slide must rest for a specific screen angle.) With the iPad Pro, you really want a table to rest it on – or else something that can go on your lap. (I’m writing this sitting in a car, so it’s not impossible.)

• lack of keyboard shortcuts for one’s own scripts. On MacOS, I can use Keyboard Maestro (and some of Apple’s custom keyboard shortcut offerings) to create a keyboard shortcut to invoke scripts which do tasks such as adding text or HTML to a clipping. On iOS, there’s no such option. So I invoke the Share menu and Workflow a lot.

• grab problems. Sometimes it’s difficult to select a chunk of text, especially if it goes past a photo on a page.

• can’t grab inline image URLs. There isn’t a way that I can see on the iPad to find the URL to an image on a Safari page and directly copy that. It might be possible with a bit of scripting (input some text before and after the picture; script grabs the source, looks for image links between those words). Solution: presses on the image and choose “open in new tab” and grab the link from that tab. But it’s an extra step, and isn’t always available – take the example below from Techcrunch.

Confusing choices on an image menu on Safari on the iPad

An image on Techcrunch (though it happens on many sites). It’s not obvious that “copy” means “copy the image URL to the clipboard” rather than “copy the whole image to the clipboard”.

(OK, so people on Twitter have pointed out that the “copy” there is “copy the image URL”. I have to say that isn’t self-evident, and I didn’t try it because I didn’t want to destroy what was already on the clipboard. Anyway, there you have it: the solution is to “copy”.)

• information density. If you compare the number of pixels on an iPad with those on a laptop, it doesn’t seem like that many more. But the action targets (the things you have to hit with the mouse) are way smaller on the laptop than on the iPad, where they’re larger because it must expect that people will only use their fingers to operate it – even if the Pencil is an option. Smaller targets and more pixels means a lot more space can be used for information.

• you’re using a tablet? Some sites still don’t expect that. Yes, I’m looking at you, Flickr, and your impossible-to-copy text in the “embed” link. On a Safari page, this comes up as it does on the desktop – a floating window with some HTML. But trying to tap-to-select the necessary part of that code – which begins “https://farm…” and ends “.jpg” is a fight, and copying precisely what you want a truly vexing process. (I haven’t found a way to script the grabbing of the necessary code, and the Flickr app is unbelievably useless: can’t restrict a search to a specific licence, can’t do a view by date/relevance, and so on. It’s mindless crap meant for the most passive user imaginable.)

• missing web page functions. I use Instapaper to collect links through the day/week; in Safari on Mac you get icons to delete links after you’ve used them. Not on Safari on iPad. (The Instapaper app does, though.)

• easier to miss stuff. The Overspill Start Up daily email requires a specific set of things to be correct about the WordPress blogpost (correct category, launch before a certain time). I’ve made more mistakes with the WordPress interface in Safari on the iPad in three weeks than I did on the laptop in three years, which has led to missed blogpost launches and missed email deliveries (sometimes both, sometimes just one). It’s very annoying; partly it’s that some of the work was done before by Marsedit (see below) and that the web interface for WordPress is appalling when it comes to the scheduling/category stuff. (Ought to be at the top of the page; instead is relegated to the side, sometimes well down the side.)

• miss having a newsreader. I use NetNewsWire on the Mac, which I’ve been using for about 15 years now. (It’s not as good as it was.) I know it’s available for iOS; I just hadn’t set up the synchronisation, so it would have been a big slog.

Room for improvement

• Mail needs work. Quite a bit to bring it up to speed. Though you can filter your inbox(es) by all/unread/flagged/has attachment/to me/cc me, there are no Smart Mailboxes (I can’t create a virtual inbox of messages with particular characteristics, or from a particular sender or domain or set of domains). I also want to be able to see more emails on the left-hand pane – if you get any appreciable number of emails per day, they’re going to overwhelm those you were previously dealing with, which screws up your workflow.

• I’d really like a good blogpost editor, ideally scriptable – basically, MarsEdit for iPad. MarsEdit is a wonderful blogpost writing/editing program which can deal with multiple blogs, and is also scriptable so that you can fire up a script and get things done. The WordPress app (in which I’m writing this) is OK, but not very intuitive. Although – as soon as I made this complaint to myself, I realised there might be a solution. And so there was, via Workflow (which can control WordPress – you can do pretty much everything up to scheduling the post) plus Python(ista). With a bit of finagling, I had a solution which did slightly more than the version I run on my Mac.

In many ways, this post is like the real-life experience that I wrote about more as a theory in Benjamin Button moves from an iPad Pro to a MacBook Pro. But it’s reality. Necessity may be the mother of invention, but sometimes it’s just the mother of getting on and discovering what tools are actually available.

I eventually got a Genius Bar appointment for the Mac. After a lot of to-ing and fro-ing (including a disk wipe that turned out to be unnecessary, but that’s for another post) the diagnosis was a problem with the NVidia video card – a known fault on my model. Weird that it took over five years to become calamitous, but that’s computers.

In the meantime, I’ve got the iPad, and over the weekend wrote a combination Workflow/Pythonista script which automates almost the whole process of compiling and scheduling The Overspill. Of course, rather as we redefine artificial intelligence to be “anything that computers can’t yet do” (where the goalposts move from “beat humans at chess” to “beat humans at Go” to “be better than us at Where’s Waldo”), the definition of “real work” has probably moved so that, because everyone knows you can’t do “real work” on an iPad, it’s now all the things that I don’t or didn’t try to do – the podcasts and video editing.

But you know what? It works for me. Plus it’s improved my Python.

TL;DR: want to do pretty much everything you do on a Mac, but on an iPad? Get Workflow and Pythonista.

The Apple Watch Series 3 ripoff: how carriers want to charge for zero data use

The Apple Watch Series 3 can take phone calls. But you’ll pay for that. Photo by portalgda on Flickr.

On first trying the Apple Watch, in 2015, my reaction was that it did a lot of things pretty well. I still wished that it had an always-on screen. But earlier this year I started taking exercise more seriously. At that point, it suddenly comes into its own: the workout apps, the heart monitoring, the calorie estimator. Add AirPods – I was quick enough to snag a pair when they went on sale in the UK last Christmas – and you have a terrific combo for running: store some music on your watch, connect AirPods, go running. No wires, no phones, and no, they don’t fall out.

When I’m out I see other runners with phones strapped to their wrists, with headphone wires all over the place. They give me odd looks. I give them an odd look right back. Exercising without wires is how it’s meant to be. (If you’ve got a Watch then I recommend the HeartWatch app, which gives you the granular detail of your heart rate, especially during workouts.)

Since you can add Apple Pay, the Watch becomes a device that can do everything while you’re out and about, even without a phone. Except.. if you don’t have a phone you can’t take phone calls, or receive and respond to text and other forms of messages, or get new data for Maps, or activate Siri, etc, etc.

Adding mobile (“cellular”) capability makes perfect sense there. Now you really can leave the phone at home, because you can receive calls anywhere you get coverage – with good LTE this means plenty of places, such as the middle of a lake, as in the Apple demo – and make them, because your contacts list is in the phone, and failing that there’s a Big Buttoned Virtual Keypad.

And generally in technology, if someone can, someone will. Samsung had already gone there, but its device was big and bulky, and it didn’t have the same phone number as your phone. Apple has solved that.

Zero data, zero incentive

What doesn’t make sense is the price that carriers are looking to charge for hooking your Watch to their network. In the US, the price is put at $10/month; in the UK, at £5 per month, on EE.

These are outrageous prices, on a par with the ludicrous data charges that carriers used to apply before the iPhone. In those days, up to mid-2007, to want data on the move marked you out as someone with money to burn, or else a raging desire for debt.

Why outrageous? Because Watch cellular data use is not additive; it’s substitutive. If you’re pulling in data on your cellular Watch, you must have left your phone behind. Ergo, you’re doing nothing with the phone, so it’s consuming (next to) no data. The data consumption has shifted to your Watch.

(Just to be clear: Apple says that your Watch uses the best available connection with your phone. If you’re in Bluetooth range, it uses that. If you’re on the same Wi-Fi network (or even, magically, a Wi-Fi network that your phone knows how to connect to, even somewhere distant) then it’ll use Wi-Fi. Now, if you’re not in range of either of those, the Watch will connect to the data network when it has to. But most of the time, and especially when you have your phone with you, it won’t be connecting to the mobile network.)

If anything, you’ll be consuming less data while you’re Watching solo – you won’t be loading Facebook pages, or giant email attachments, or scrolling through Twitter, or watching YouTube. Sure, you might be listening to music streamed from Apple Music. But you might well have been doing that anyway; if you like streaming music while you run, you’ve probably been doing that already, but with a phone around your arm. (And you can get music onto the Watch just by downloading it from the phone, rather like one used to with iPods. This is probably the biggest use case of music on the Watch even if you can stream, because runner like to create their own playlists, not rely on stuff in the cloud.)

Nor do the carriers have to send you a physical SIM; it’s done in software, in the Watch. Nor do they have to open a new account; you’re already a customer. There might be a mild bit of back-end administration to inform the cell network that two different IMEIs (mobile device IDs) have the same phone number. (Side note: the fact this can be done implies that spying on your phone calls may be easier than it seems?)

But there’s nothing in there which justifies $10/month or £5/month. And think of what that adds to the cost of the device: $120 or £60 per year. That’s a substantial chunk of the upfront price, and it never stops. On Twitter, Marine Engelvuori points out that EE ties you to a 24-month contract if you buy the watch from them, and that you have to add VAT; suddenly that device which costs £399 on its own has added £200-odd of costs over the contract lifetime.

If the cost were $1 or £1 per month, that would be tolerable; one can concede that carriers could charge for the tiny bit of administration cost that might be involved, and maybe eke a profit on the fact of this device’s new qualities. But more than that is just absurd, and it will stifle purchases by anyone who might be a marginal buyer of the service.

This is a real pity. The Series 3 is a remarkable piece of engineering: turning the screen into the aerial (I don’t even know how they do this) and maintaining the thin profile is just amazing. All the software functionality, such as heart rate monitoring and so on, is top class. People could benefit from cell-connected smartwatches, and not only the ones made by Apple. (It might encourage people to spend less time staring at screens, weirdly enough.)

But the price that the carriers are trying to charge is stupid.

Third-party like it’s 2006

It really is 2006 in wearable land; the time before carriers woke up to the broader benefit of offering services at prices which encourage people to use them. Wearables are, arguably, still at the same stage in their evolution as the smartphone was in 2006. This doesn’t mean though that the carriers couldn’t act as the midwives to help things along a little.

Remember, they’re trying to charge this amount for something which will use no extra data over you using your phone, and for which they don’t have to provide a physical item.

There is a precedent for doing this well: Amazon and the Kindle. The deal it cut for “Whispernet” meant you could download books anywhere and all you paid for was the extra 3G functionality in the upfront price. No ongoing fees. I can imagine that Apple’s board gulped a bit at the potential cost of doing that for the Watch, when people would no doubt eagerly take the chance to stream music all day and all night long forever for the extra £70. Kindle files are pretty small compared with music files, and Amazon had a monopoly on that market. So it was probably a non-starter for Apple to shoulder the cost. (This doesn’t mean there’s a cost to the carriers – as I said above, it’s substitutive. But it would be all new costs for Apple to pay for Watch data.)

Maybe the first carriers are just hoping to rake it in before competition opens up and drives prices down. Here’s hoping.

It took the iPhone, and Steve Jobs’s negotiating genius, to get carriers to adopt a flat rate model for data. It’s a disappointment that Apple hasn’t managed to push the future of connectivity forward in the other place where it matters – not on your wrist, because they’ve solved that; but in your wallet.

Fining Google: a slow train coming

“Slow Train Coming”, the artwork from the cover of Bob Dylan’s album. Photo by Logos: the Art of Photography on Flickr.

The cover of Bob Dylan’s album “Slow Train Coming” shows people literally laying a railway just ahead of a train which is, in theory, a-comin’. Just very slowly. The European Commission’s antitrust decision against Google is just such a train. A €2.42bn train. Big, but deathly slow.

(If you need any background about the EC and Google and why this all matters, I wrote about it in 2015. Slow train.)

• Google has been squashing rival shopping sites since mid-2006;
• the EC was alerted in summer 2009 after many efforts by sites to get responses from Google failed;
• do we seriously think Google’s going to change its behaviour?
• why isn’t Foundem getting a slice of the fine?
• antitrust moves too slowly in the modern era

The European Commission’s fine of €2.42bn on Google has been just like that train: a damn long time coming. The original complainant, the “vertical search” site Foundem, first noticed something funny happening to its position in search results back in 2006: it was being penalised for no apparent reason.

The penalty (search) box

Foundem was the brainchild of Shivaun and Adam Raff (it really is like their child, and they are brainy; I’ve met them on several occasions as this antitrust case has inched its way through the system). By this time the site was only six months old, focussed on what it saw as a gap – or at least growing niche – in the market: “vertical search”, comparing one specific product, rather than “horizontal search” as practised by Google and Bing (and many also-rans). You can probably think of other “vertical search” sites: Kelkoo was very big at one point. There’s also one called Amazon, though at that time it did a lot of the fulfilment as well; Foundem would find results from other shopping sites, so that it was like a meta-search engine. Amazon, at the time, wasn’t, though as it has become more of a marketplace rather than a fulfilment company that description is increasingly accurate.

But for Foundem in June 2006, this was remote. It had been hit by an algorithmic search penalty which hit lots of vertical search companies. It filed “reconsideration requests” to Google, which it says the company ignored.

(See the timeline for yourself at Foundem’s site.)

In August 2006 it was hit by an “AdWords Penalty”: this suggested that “landing pages” people arrived at were such low quality that it would have to pay much more to be able to buy an AdWord (Google advertising position). How much more? It was raised, they say, from about 5p/click to £5/click.

It’s summer 2006, and as the Raffs put it in their timeline, “Foundem was excluded from Google’s natural and paid search results, both of which are essential channels to market for any internet-based business.” That would be near enough a death penalty for any consumer-facing business; fortunately they found other outlets, such as powering shopping searches on magazine websites for IPC, Bauer and others.

The Raffs kept lobbying Google for reconsideration, and kept being brushed off; meanwhile Google launched Universal Search (integrating Google Maps and Google News and YouTube results into a box at the top which favoured Google products and pushed rival services further down the search rankings).

In December 2008 a TV show named Foundem the UK’s best price comparison site. Google meanwhile didn’t relent on its penalty against Foundem’s position in search results.

Finally, in July 2009 Foundem had its first meeting with the EC’s DGComp – the arm of the European Commission which investigates antitrust cases.

Eight years and more of hurt

That’s almost exactly eight years ago. It’s taken absolutely ages for the EC to act on this, giving Google plenty of time to tighten its grip on the business, and even for the whole search landscape to shift – from one where the desktop has primacy to one where many searches begin on mobile, inside apps.

There’s lots of applause today from Europeans about the fact that Margethe Vestager didn’t give up on this case, and that a record fine has been imposed (and that if Google doesn’t alter its behaviour in 90 days, the daily fine will be eye-watering). “Better later than never, but seven years have been still an eternity for some market players, in particular European SMEs [small and medium enterprises],” to quote the MEPs Ramon Tremosa I Balcells and Andreas Schwab.

There’s the usual eye-rolling from a number of American observers, who say “which AMERICAN company will be next?”, and ignore the fact that the Federal Trade Commission’s investigation in 2011/12 discovered that Google’s own user testing found that people preferred seeing other vertical search engine results in the organic search results; and also ignore the fact that DGComp fines all sorts of European companies for antitrust and cartel actions of all flavours. (The decision before Google was fining three car lighting system producers over cartel behaviour.)

Also, for those eye-rolling American (and other) readers: European antitrust doctrine differs in one very significant way from the American flavour. In the US, if you use a monopoly in one space to take over another but consumers benefit overall, there’s no case to answer. This was why the FTC dropped its case (on a 4-0-1, ie one abstention, no opposition) decision. Scroll down in that FTC release to “search bias” where it says the introduction of Universal Search “could be plausibly justified as innovations that improved Google’s product and the experience of its users.” A bit milquetoast, that recommendation.

In the EU, however, the question is whether antitrust stifles competition, not what happens to consumers. This refusal to consider “consumer surplus” infuriates and astonishes a significant number of American observers, but it’s how it’s done here.

But, but, but. I very much expect that Google will appeal this before the 90-day deadline, and that this will mean it doesn’t yet have to change its behaviour, nor pay the fine. Do you think that this might be a long-drawn-out process which will grind interminably through the courts, during which Google won’t change how it displays results? I do.

Meanwhile Foundem and all the other vertical search companies which the EC is ostensibly protecting have been almost crushed. If there were any justice, they’d be getting a slice of the fine. After all, companies which report cartel action either get some payment, or (if they’re part of the cartel) let off some of the fine.

Slow train, now arriving

This is the reality of antitrust: in technology especially, the dominance of these companies and the power of their networks means that the decision comes too late to help those who were originally affected. It was certainly the case with Microsoft and Netscape; it’s clearly the case here. Who knows how big Foundem and Kelkoo and all the others might have been if Google hadn’t been able to use its dominance in straight search to annexe the vertical search space?

Some would really like the fine to have teeth. Tremosa i Ballcells commented: “When it comes down to the fine, I always said: first, you pay the fine and, then, you restore competition and the level playing field like it was the case with Microsoft. I believe that the fine should be retroactive for each year since the beginning of the wrongdoing by Google. This fine is far from the theoretical fine of 10% of Google annual revenues. The fine should be multiplied by the number of years since the start of the damage to competitors. Moreover, the behaviour of Google since the SO [Statement of Objections] and from today should be taken into account as well. Time helps monopolies, not SMEs.”

The argument of course is that antitrust actions serve to make the dominant company change its future behaviour: a fine of that size, and the threat of continuing fines, and particularly the tedious legality of it all, burdens the company’s decision-making process so that its executives all act as though someone suggested they play on the electrified railway when the idea of moving into “adjacent” business comes up. (It certainly worked with Microsoft.)

This will be the real acid test of the EC’s action: will it make Google’s internal culture change? We won’t know the answer to that for some time. Slow train coming.

In search of the early adopter (HomePod edition)

An Apple HomePod. If your home is this nice, wouldn’t it already have a Sonos Play:1 or similar there? Photo by portalgda (via Apple) on Flickr.

Ahead of Apple’s WWDC, research companies were falling over themselves to offer their forecasts about how the “smart speaker” market would grow over time.

Here, for example, is what Strategy Analytics reckoned the world would look like – note that it’s using installed base, not sales – over time:

Smart speaker market - Strategy Analytics forecast to 2022

The way they see it, Amazon’s early lead with the Alexa is going to be eroded by Google, while Apple, “others” and Baidu (in China) will take the rest.

What I wonder about is: who’s left who wants to buy a HomePod? Who are the early adopters?

If you want a “smart speaker”, you’ve been served for quite some time in the US by Amazon’s Echo and Dot products, which are passably cheap. For Amazon, they’re the peace dividend of losing badly in the smartphone wars when the Fire Phone turned out to be a clunker.

More recently you’ve been able to buy Google Home – which managed to get a passing mention in a Modern Family episode in a recent series, so that is clearly reckoned to have reached far enough into the public consciousness not to merit special highlighting. (Sure, it could have been product placement, but there’s no point placing a product nobody’s heard of. Modern Family likes to play with modern tech obsessions: in 2010 one of the episodes was about Phil Dunphy’s mad desire to get an iPad on first-day release.)

And there’s even a (Microsoft) Cortana speaker from Harmon Kardon.

A recent survey of 1,000 people found that “smart speakers” are the most popular category of “smart home” device: about a quarter of US households have some smart home gizmo, and of those 56% reported that they “own and use” a smart speaker. And they really use it regularly. Half of respondents (we’re at about 12% of households) use it at least daily; another 39% use it several times per week.

Top five uses, in order: play music, ask for the weather, get news, get basic facts and trivia, get or set calendar and/or scheduling info.

Not mentioned in this, because it doesn’t do “smart” (so far, and probably not ever) is Sonos, which most people are probably familiar with: it provides single- or multi-room music and speech streaming for pretty much any service, with a range of high-quality audio speakers, as well as TV soundbars. It has been going since 2002, focussing just on multi-room music; I’ve liked the Sonos idea ever since I saw it in 2005, and I think that when it released the Play:1 speaker it found the sweet spot of price and audio quality. (I own a number of Sonos devices.)

Sonos, one should note, has hit some rough times, laying people off in March 2016. It isn’t clear how big its installed base is – it only talks about serving “millions of rooms” – but it’s very likely that it has users in multiple millions of homes. Update: this person says Sonos has annual sales of about $1bn, equating to 5m speakers per year. “Not mainstream,” they say. Though I’d say that 5m per year – more than a million a quarter – isn’t too bad.

And how does the install base look at present? According to CIRP, about 11m Amazon customers have an Echo, of which 52% (6.6m or so) have the cheap Dot. Amazon has about 70% of the market, says eMarketer – though others put the figure higher.

All this leaves one wondering: hasn’t the early adopter market, who might have been keen to buy a comparatively expensive smart speaker (and even slightly more expensive than Sonos) been tapped already? There are millions of those things out there already, playing music and telling people what they could figure out themselves by gazing out of the window.

After all, if you have a product which does something comparably new, then the general thesis is that you have to tap the innovator market (about 2.5% of the total who will buy it) and then the early adopter market (13.5%), and then spread the news to the “early majority” and “late majority” who each comprise 34%. Then finally you mop up the 16% of laggards. Note that even when your market is saturated, you don’t necessarily reach 100% of the population. Not everyone will want your gadget.

So at what stage are “smart speakers” – and multi-room speakers, since the HomePod does both?

We can probably be confident that multi-room speakers have breezed into their late majority by now. If Sonos is having trouble finding fresh buyers, that’s a sign of market resistance.

And if smart speakers are already in 12% of US households, then those are nudging well into the early adopter market.

This is different from the situation when Apple has launched previous products. The Macintosh, all those years ago, was entirely new in sporting a graphical user interface. The iPod came early in the MP3 player revolution – though many people had portable CD players, most people didn’t have an MP3 player, nor an MP3 library. The iPhone was out on its own in the smartphone category through its all-touch interface. The iPad defined an entire all-touch “slate” tablet market. (There had been tablets before, which had poor interfaces for touch.) Even the most recent new device, the Apple Watch, came when smartwatches had very small user penetration.

So in that sense the HomePod is coming at a very different time: three years (and some months) after the original Echo, and what’s more it’s coming in December, giving everyone else a chance to get their marketing in first.

How then is Apple intending to sell it? Apple is framing it as a really music speaker with a bit of intelligence thrown in. Here’s the quote:

“Apple reinvented portable music with iPod and now HomePod will reinvent how we enjoy music wirelessly throughout our homes,” said Philip Schiller, Apple’s senior vice president of Worldwide Marketing. “HomePod packs powerful speaker technology, Siri intelligence and wireless access to the entire Apple Music library into a beautiful speaker that is less than 7 inches tall, can rock most any room with distortion free music and be a helpful assistant around your home.”

This makes it sound like Sonos with benefits. It’s also saying: those things that Amazon and Google can do? We can do that, but sound better. It’s looking for the things that those can’t (yet?) do, and aiming for them. Early user tests (in very controlled environments) suggest that the HomePod sounds better than the Sonos Play:3 (which is comparably priced). Could be, though I’m not sure why a Sonos owner would give up the latter for the former. And we know from the triumph of MP3 over CD that in music, people prefer convenience over sound quality. Update: two points made to me after first publishing this post. First, Apple could find a market in China, where Amazon and Google are effectively excluded. Second, the privacy angle could be attractive; some people just don’t like the idea that Amazon and Google are going to try to sell them stuff. (Google Home has already started doing this, and will carry on; Amazon’s Echo/Dot/etc are intended to be mainlining for shopping lists.)

Overall, it feels as though Apple decided – as usual – that there was only one place it could thrive in this market: at the premium end. But again, if someone has the sort of money needed to buy a HomePod, why haven’t they already bought some Sonos kit? Or if they want a smart speaker, wouldn’t they just get an Echo/Dot/Home?

That’s what I’m puzzled by: how Apple is going to scoop up enough of the early majority market to make this work. Early majority buyers can be more price-sensitive than innovators and early adopters. Apple’s pitch seems to be for those who haven’t bought a Sonos, saying: look, you can control this with your voice. There’s surely a market there – Strategy Analytics thinks it’s tolerably big – but it’s noticeable that even there it’s a comparatively small slice of a sizeable market. Sometimes it’s good to let the early players sort the market out for you. But I think this might be one case where it isn’t.