Start Up: TurboTax’s dark patterns, fiduciary Facebook, Rome’s (real) collapse, Xiaomi and GoPro?, and more

Cocktail party? Google can listen in on specific voices for you. Well, for itself. Photo by James Vaughan 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. Who knows what the GDPR says. I’m @charlesarthur on Twitter. Observations and links welcome.

TurboTax UX and dark patterns critiqued • Medium

Brandon Read:


It’s important to recognize that TurboTax is one of many tax-preparation corporations lobbying against legislation that could greatly simplify the filing process for millions of Americans. This means it’s in their best interest to perpetuate the existing convoluted tax system so that they may continue to generate massive profits each year. We’ll explore concrete examples of unethical design strategies TurboTax employs to generate these profits.

TurboTax dissuades customers from using their Free offering by exaggerating the benefits of their mid-tier (or “recommended”) paid service and by fabricating obstacles that trick users into paying for unnecessary upgrades. While the paid service offers benefits that may be applicable to some users (such as specialist support and increased security), most filers would be perfectly happy (and just as successful) filing their taxes through the Free product. Instead of surfacing this reality, TurboTax buries it by manufacturing the illusion of complexity and time-scarcity. The following UX teardown shines a critical light on these Dark Patterns, and offers users tips on how to stay in control when navigating the modern freemium landscape.


There are so many dark patterns in this stuff that it’s like a black knitted blanket.
link to this extract

We may own our data, but Facebook has a duty to protect it • The New Yorker

Nathan Heller:


Two years ago, Jack M. Balkin, a constitutional-law professor at Yale, published a fifty-page article in the U.C. Davis Law Review examining what he called problems “at the intersection of information privacy and the First Amendment.” On one hand, he noted, people want to protect private information. On the other, information businesses tend to challenge regulation as infringements of free speech. Balkin ran through some prospective solutions. The government could regulate the collection and use of information, or the time, place, and manner of expression. Companies could treat data as commercial speech, or as a commodity. Platforms could make privacy contracts with their users. Yet Balkin found all these options lacking. Instead, he offered the idea of the “information fiduciary.” Fiduciaries, in traditional contexts, are defined by two responsibilities. They must be loyal to their clients’ interests, and they must show a “duty of care.”

It was no surprise to find Balkin’s article mentioned during Mark Zuckerberg’s testimony to the Senate Judiciary and Commerce Committees earlier this week. To a striking degree, the fiduciary model was the one toward which discussion slowly and chaotically converged. The hearing revealed little about Facebook, the company that Zuckerberg founded, and a lot about the committees, which at times seemed hair-raisingly ill-equipped for their task…

…What a duty of care might look like for a company such as Facebook was the meat of Balkin’s paper in 2016. One benefit of an information fiduciary, he argued, is that it has widely and easily understood obligations that go beyond a written policy. (You can reasonably expect your tax accountant not to send your financial information to the nearest newsroom, regardless of whether there’s an explicit agreement to that effect.)


link to this extract

The “GDPR consent” email I’d like to receive • informationrightsandwrongs

Jon Baines:


“Dear Jon

You know us. We’re that firm you placed an order with a few months ago. You may remember that at the time we took your order we explained we were going to send occasional marketing emails to you about similar products and services, but you could opt out then, and at any subsequent point.

We know that since 2003 (with the Privacy and Electronic Communications Regulations) (PECR) it’s been unlawful to send unsolicited marketing emails except in circumstances like those above.

We’re contacting you now because we’ve noticed a lot of competitors (and other firms) who are either utterly confused or utterly misrepresenting a new law (separate to PECR) called the General Data Protection Regulation (GDPR). They’re claiming it means they have to contact you to reconfirm your consent to receive marketing emails.

GDPR actually says nothing of the sort. It does explain what “consent” means in data protection terms in a slightly more strict way, but for companies like us, who’ve respected our customers and prospective customers all along, it makes no difference…”


Plenty more. The GDPR is one of the least understood laws around, one suspects.
link to this extract

Rome wasn’t built in a day but these days it feels as if it may collapse in one • The Guardian

Tobias Jones on the sinkholes cropping up all over Rome:


In November – and this is a sure sign things are serious – Lazio’s football match against Udinese was postponed due to torrential rain. Last week, there was more flooding of the subway. In the past month, central Italy has had 141% more “anomalous rainfall” than average.

Rain is a problem because of the city’s geology. Much of Rome is built on unconsolidated (ie soft) sediments, like the floodplain of the river Tiber. That means that water washes away small deposits that give the ground additional rigidity. Soft soil also amplifies not just earthquake tremors (hence the missing south side of the Colosseum) but also the vibrations of the city’s incessant traffic, causing what the president of Lazio’s guild of geologists calls “the liquefaction of the ground”. It’s like shaking a sieve full of water and clay below the asphalt: soon enough, the water will whisk away the grit and you’ll be left with a jelly-like blob to support the heavy traffic.

Additional water comes not from the skies but from the creaking subterranean infrastructure. Ancient aqueducts, such as the Vergine one that supplies the Trevi fountain, are still in use. Because of leaks, 50% of water is lost between the Lazio region’s freshwater lakes and Romans’ taps. Many of the city’s sewers are so old they’re made of cracked brick and tiles. And the fact that there are 32 sq km of tunnels, cavities, catacombs and quarries beneath the surface of the city hardly helps.

In many ways, the city council has exacerbated the problem: it is perennially corrupt and chronically incompetent. Last December, it was unable even to buy a green evergreen for Christmas. The tendering process for road repairs and reconstruction has been dragging on for years, because Roman bureaucracy is like treacle. And when a contract is finally awarded, companies often cut corners, patching roads badly because that way there will be more work in future.


Remarkable piece; it makes it feel as though Italy, or at least Rome, is on the verge of collapse.
link to this extract

Google works out a fascinating, slightly scary way for AI to isolate voices in a crowd • Ars Technica

Jeff Dunn:


The company says this tech works on videos with a single audio track and can isolate voices in a video algorithmically, depending on who’s talking, or by having a user manually select the face of the person whose voice they want to hear.

Google says the visual component here is key, as the tech watches for when a person’s mouth is moving to better identify which voices to focus on at a given point and to create more accurate individual speech tracks for the length of a video.

According to the blog post, the researchers developed this model by gathering 100,000 videos of “lectures and talks” on YouTube, extracting nearly 2,000 hours worth of segments from those videos featuring unobstructed speech, then mixing that audio to create a “synthetic cocktail party” with artificial background noise added.

Google then trained the tech to split that mixed audio by reading the “face thumbnails” of people speaking in each video frame and a spectrogram of that video’s soundtrack. The system is able to sort out which audio source belongs to which face at a given time and create separate speech tracks for each speaker. Whew.


Creepy machine learning! Let’s continue that thread…
link to this extract

Facebook uses AI to predict your future actions for advertisers, says confidential document • The Intercept

Sam Biddle:


The recent document, described as “confidential,” outlines a new advertising service that expands how the social network sells corporations’ access to its users and their lives: Instead of merely offering advertisers the ability to target people based on demographics and consumer preferences, Facebook instead offers the ability to target them based on how they will behave, what they will buy, and what they will think. These capabilities are the fruits of a self-improving, artificial intelligence-powered prediction engine, first unveiled by Facebook in 2016 and dubbed “FBLearner Flow.”

One slide in the document touts Facebook’s ability to “predict future behavior,” allowing companies to target people on the basis of decisions they haven’t even made yet. This would, potentially, give third parties the opportunity to alter a consumer’s anticipated course.

Here, Facebook explains how it can comb through its entire user base of over 2 billion individuals and produce millions of people who are “at risk” of jumping ship from one brand to a competitor. These individuals could then be targeted aggressively with advertising that could pre-empt and change their decision entirely — something Facebook calls “improved marketing efficiency.” This isn’t Facebook showing you Chevy ads because you’ve been reading about Ford all week — old hat in the online marketing world — rather Facebook using facts of your life to predict that in the near future, you’re going to get sick of your car. Facebook’s name for this service: “loyalty prediction.”


AI for everything!
link to this extract

How I implemented iPhone X’s FaceID using Deep Learning in Python • Medium

Norman Di Palo:


Performing classification, for a neural network, means learning to predict if the face it has seen it’s the users’s one or not. So, it should use some training data to predict “true” or “false”, basically, but differently from a lot of other deep learning use cases, here this approach would not work. First, the network should re-train from scratch using the new obtained data from the user’s face.

This would require a lot of time, energy consumption, and impractical availability of training data of different faces to have negative examples (little would change in case of transfer learning and fine tuning of an already trained network). Furthermore, this method would not exploit the possibility, for Apple, to train a much more complex network “offline”, i.e. in their laboratories, and then ship it already trained and ready to use in their phones.

Instead, I believe FaceID is powered by a siamese-like convolutional neural network that is trained “offline” by Apple to map faces into a low-dimensional latent space shaped to maximize distances between faces of different people, using a contrastive loss. What happens is that you get an architecture capable of doing one shot learning, as they very briefly mentioned at their Keynote. I know, there are some names that could not be familiar to many readers: keep reading, and I will explain step by step what I mean.


If you’re into machine learning, this is quite a read.
link to this extract

My 9.7 iPad (2018) review: Drawn, written, edited, and produced with an iPad • iMore

Serenity Caldwell:


It’s no secret to say that the iPad has changed how I work and think on my devices. I use it for work, roller derby, casual sketching and idea generation, watching movies, and so much more. And it’s why I’ve continually been bullish on the device, even when sales lagged and great multitasking was but a rumor on the road map.

To me, the 2018 base-model 9.7-inch iPad is a special beast: It hits a line drive right through the company’s fabled intersection of technology and liberal arts — and at the right price point. The iPad Pro did it first, but at a cost unattainable for all but the tinkerers and serious artists, and without iOS 11’s crucial multitasking features. At $329, the iPad offers a low-end tablet experience unlike any other on the market. Add an extra $99 for Apple Pencil, and Apple has created the best device for all-purpose education, period.


This is absolutely amazing.
link to this extract

Elon Musk says Autopilot will never be perfect but can ‘reduce accidents by a factor of 10’ • BGR

Yoni Heisler:


Earlier [on Sunday], CBS This Morning aired a new clip from its sit-down interview with Tesla CEO Elon Musk. The clip below centers on Tesla’s Autopilot feature, a topic that has been in the news quite a bit over the past few weeks following a tragic crash that saw a Model X in Autopilot mode careen into a highway divider before the vehicle’s battery pack burst into flames.

In the wake of the accident, Tesla said that the car’s Autopilot system warned the driver to place his hands on the wheel in the seconds leading up to the crash, warnings that Tesla claims were not heeded.

“The crash happened on a clear day with several hundred feet of visibility ahead,” Tesla said earlier this week, “which means that the only way for this accident to have occurred is if Mr. Huang was not paying attention to the road, despite the car providing multiple warnings to do so.”

As part of the interview, CBS This Morning co-host Gayle King went on a drive with Musk in a Model 3 where the two talked briefly about all things Autopilot. When asked about the benefits of Autopilot if the feature requires users to keep their hands on the wheel, Musk responded: “Oh, it’s because the probability of an accident with autopilot is just less.”


Musk is pushing this hard, but I think that this case is not going to break in Tesla’s favour in the way that its response to the NY Times car critic did in 2013.
link to this extract

Former operator of Android app pirate site Applanet gets three years’ probation • Android Police

Jason Hahn:


Aaron Buckley, who was an enterprising 15-year-old when he launched Applanet from his parents’ home in Mississippi, pleaded guilty to two counts of his indictment: conspiracy to commit copyright infringement and criminal copyright infringement. The Northern District Court of Georgia announced on April 11th that Buckley, now in his mid-20s, will be placed under three years’ probation and will also be put into a home-incarceration program for 365 days. He will also have to complete 20 hours of community service, work toward his GED, pay a $200 “special assessment” fee, and refrain from owning a firearm or possessing a controlled substance.

Buckley’s attorney pushed for a lenient sentence from US District Judge Timothy Batten, framing Buckley’s life since launching the site for pirated Android apps as one of community work and taking a leadership role in a support community for LGBT teenagers. He also spoke of unspecified difficulties in Buckley’s personal life.

“I really respect the government and the judge in their sentencing and am extremely grateful that they took into account all concerns of my health and life situation in regards to possible sentences,” Buckley told TorrentFreak.


The tiny bit that struck my eye was the “refrain from owning a firearm”. I don’t see why operating an app pirating site would make you unsafe to own a gun. Would it?
link to this extract

Xiaomi could be just the hero GoPro needs • Bloomberg Gadfly

Tim Culpan:


GoPro’s problem is that it hasn’t done much in 16 years. Its product line is little changed, with mere iterations of the same tiny rugged camera, and the company still relies on its home market for the bulk of sales. Consider that in 2004 – when GoPro released its first camera – Apple Inc.’s hottest product by units was the iPod.

The few attempts to diversify have failed. An entry into the drone market in 2016 lasted less than 15 months at a time when DJI and others were enjoying booming growth. Asia accounts for just 21% of revenue.

Xiaomi, meanwhile, can’t be accused of standing still. The Chinese smartphone startup has its fingers in so many pies that it’s hard to keep up. So it makes sense that it would consider making a a bid for GoPro, as The Information reported. Xiaomi may offer up to $1bn, but doesn’t want to overpay, the news website said.

A tie-up with another device maker is exactly the future I envision for GoPro. Right now it’s a technical feat to film a day on the slopes, then take it back to show on the TV in your ski lodge. For many, it’s just easier to shoot with an iPhone and a selfie stick, which is the crowd Woodman should be chasing. A combination with Roku Inc., the provider of streaming content players, is one I have advocated for a while. Xiaomi has MiBox, as well as routers and other connected devices.

A $1bn outlay for Xiaomi shouldn’t damage its balance sheet, and the upside could be immense.


Agree – this could be just what GoPro needs (though I imagine a wailing at the idea of an American company being bought by a Chinese one). For good measure it could buy Fitbit too, which also needs a white knight while its smartwatch business seeks liftoff.
link to this extract

Errata, corrigenda and ai no corrida: none notified

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.