Pixel Envy

Written by Nick Heer.

Archive for April, 2018

Personalized ‘Hey, Siri’

A new update on Apple’s machine learning blog explores their approach to speaker recognition in detecting “Hey, Siri”. It’s obviously fairly technical, but I found this bit interesting as it describes how they measure the success of the key phrase activating Siri:

The overall goal of speaker recognition (SR) is to ascertain the identity of a person using his or her voice. We are interested in “who is speaking,” as opposed to the problem of speech recognition, which aims to ascertain “what was spoken.” SR performed using a phrase known a priori, such as “Hey Siri,” is often referred to as text-dependent SR; otherwise, the problem is known as text-independent SR.

We measure the performance of a speaker recognition system as a combination of an Imposter Accept (IA) rate and a False Reject (FR) rate. It is important, however, to distinguish (and equate) these values from those used to measure the quality of a key-phrase trigger system. For both the key-phrase trigger system and the speaker recognition system, a False Reject (or Miss) is observed when the target user says “Hey Siri” and his or her device does not wake up. This sort of error tends to occur more often in acoustically noisy environments, such as in a moving car or on a bustling sidewalk. We report FR’s as a fraction of the total number of true “Hey Siri” instances spoken by the target user. For the key-phrase trigger system, a False Accept (or False Alarm, FA) is observed when the device wakes up to a non-“Hey Siri” phrase, such as “are you serious” or “in Syria today.” Typically, FA’s are measured on a per-hour basis.

I’ve been extremely impressed by the performance of “Hey, Siri” over the last couple of years. Not only does it reliably wake my device, it also does not wake my girlfriend’s — and vice-versa, when she says “Hey, Siri”.

What Siri does after that leaves much to be desired, of course.

The Other Irresponsibility

Following Mark Zuckerberg’s awkward and tedious testimony before the House and Senate came several great pieces from journalists covering it, as well as Facebook as a whole. I wanted to collect a few of the best that I found as a sort of highlight reel of irresponsibility.

Sam Biddle, of the Intercept, on Zuckerberg’s frequent claims that he didn’t know the answer to a question:

After watching the Facebook founder and CEO’s 48-hour trip to Capitol Hill, there are two possible conclusions: either Mark Zuckerberg deliberately misled Congress, or Mark Zuckerberg knows very little about his own company. Both are bad.

Again and again, before both Senate and House committees, Zuckerberg pleaded ignorance about the company he created and has controlled for 14 years. Zuckerberg wasn’t dodging questions about obscure corners of the company or corporate minutiae, but the most plainly fundamental aspects of Facebook’s business and privacy policies. Rather than the congressional beatdown many had expected, the most striking aspect of Zuckerberg’s testimony wasn’t his painful apologias or excuse-spinning, but his ability to spend nearly 10 hours saying almost nothing. The hearings may prove to be a sea change moment for Facebook and the greater data-mining industrial complex, but it would be hard to say the public learned much of anything.

Alex Kantrowitz, Buzzfeed:

During his two-day marathon testimony in Washington this week, Facebook CEO Mark Zuckerberg looked particularly uncomfortable answering basic questions about how Facebook tracks people when they’re not using Facebook. In case you hadn’t already heard, yes, it’s true: Facebook can track your online activity even if you aren’t signed in to Facebook.

Paris Martineau, the Outline:

Facebook claims that you can download a copy of everything it has on you here. Mark Zuckerberg said the same during his testimony to the U.S. House of Representatives yesterday (“Congressman, I believe that all of your information is in that — that file.”). However, according to Facebook’s own Privacy Operations Team, both of these statements are wrong. Even better, Facebook has told users it cannot give out this information because it’s too difficult to access and package into a readable format.

Alexis C. Madrigal, of the Atlantic, expands on the same topic:

This apparent contradiction relies on the company’s distinction between the content someone has intentionally shared — which Facebook mines for valuable targeting information — and the data that Facebook quietly collects around the web, gathers from physical locations, and infers about users based on people who have a similar digital profile. As the journalist Rob Horning put it, that second set of data is something of a “product” that Facebook makes, a “synthetic” mix of actual data gathered, data purchased from outsiders, and data inferred by machine intelligence.

With Facebook, the concept of owning your data begins to verge on meaningless if it doesn’t include that second, more holistic concept: not just the data users create and upload explicitly, but all the other information that has become attached to their profiles by other means.

Gennie Gebhart, for the EFF:

Facebook’s ethos of connection and growth at all costs cannot coexist with users’ privacy rights. Facebook operates by collecting, storing, and making it easy to find unprecedented amounts of user data. Until that changes in a meaningful way, the privacy concerns that spurred these hearings are here to stay.

Andrew Ross Sorkin, New York Times:

When Google first introduced Gmail in 2004, this newspaper raised questions about the prospect of users objecting to a service that displayed advertising to them based on the content of their email: “For many, the bottom line appears to be that sifting through personal email with an eye toward making a sale is beyond the pale.”

Well, now more than 1.2 billion people have active accounts with Gmail, a service that until the end of last year sifted through your private messages. Apparently, it wasn’t beyond the pale.

For consumers, the transaction has always been pretty clear: The convenience of free service in exchange for information that allowed advertisers to specifically target us. The distinction in that equation was motivation; we figured our data was being used by benign companies seeking to sell us that pair of sneakers we wanted, not by bad actors trying to influence our political votes — or incite violence in places like in Myanmar.

These are all very good points made by astute writers in publications that I trust. Yet, most of these web properties — the EFF’s and the Intercept excluded — use some form of Facebook’s tracking scripts, whether that’s a Like or Share button, Beacon, or Pixel. That means they’re part of the problem; in a way, I am, too, by linking to them but, in my defence, Facebook’s scripts are among the web’s most popular, as are — surprise, surprise — Google’s.

The Outline uses Facebook’s Custom Audience and Pixel tracking, according to Ghostery. However, you may not be aware of that because it isn’t exactly something they advertise. To find it, you’d have to open the hamburger menu on the above linked page, click the small “Legal” link at the bottom, and look under the second section of their privacy policy where they link to a help article about the Facebook Pixel without outright stating that they’re using it.

They’re not alone; many websites don’t fully disclose their use of these trackers, and most do so only in a byzantine and buried privacy policy. The New York Times, for example, has a tiny grey “Privacy” link nestled in their footer. Then, you must scroll about halfway down the page to the “Third Parties” section, where you may “click here” to view a list of third-parties that “may be using cookies”. However, Facebook does not appear on this list despite the Times absolutely using their scripts. I was unable to find a reference on their website to the Times’ use of Facebook’s advertising and targeting scripts.

What’s absolutely clear here is that websites need to stop using Facebook’s tracking scripts — and Google’s too, while they’re at it.

For what it’s worth, users can and should make it harder for advertising companies to collect their browsing data. In iOS, under Privacy in Settings, you can switch on the Limit Ad Tracking option, and turn on Prevent Cross-Site Tracking under Safari settings. The latter option is also available for Safari on MacOS. Zack Whittaker at ZDNet has more information on opting out. You can also use a script or ad blocker to prevent tracking scripts from loading.

I completely understand that these scripts provide many advantages from a marketing and advertising perspective. I also get that the realities of the news business mean that publishers feel forced to make hard choices that increase revenue despite potentially compromising on principle. But websites that embed these scripts are contributing to these privacy-violating platforms. All web property owners — but especially highly-trafficked properties — have a responsibility to their visitors. Participating in a web-wide tracking scheme betrays that trust. It must be stopped.

Gurman: HomePod Sales Are Middling So Far

Mark Gurman, Bloomberg:

At first, it looked like the HomePod might be a hit. Pre-orders were strong, and in the last week of January the device grabbed about a third of the U.S. smart speaker market in unit sales, according to data provided to Bloomberg by Slice Intelligence. But by the time HomePods arrived in stores, sales were tanking, says Slice principal analyst Ken Cassar. “Even when people had the ability to hear these things,” he says, “it still didn’t give Apple another spike.”

During the HomePod’s first 10 weeks of sales, it eked out 10 percent of the smart speaker market, compared with 73 percent for Amazon’s Echo devices and 14 percent for the Google Home, according to Slice Intelligence. Three weeks after the launch, weekly HomePod sales slipped to about 4 percent of the smart speaker category on average, the market research firm says. Inventory is piling up, according to Apple store workers, who say some locations are selling fewer than 10 HomePods a day. Apple declined to comment. The shares gained 1.4 percent to $173.83 in early trading.

This doesn’t surprise me. I mean that in the sense that the HomePod isn’t, as far as I’m concerned, a very good product yet, but also because it’s a version one Apple product that doesn’t have a wide rollout. Some analysts were disappointed with initial Apple Watch sales, too, and it launched in nine major markets instead of the three that the HomePod is currently available in. That’s not to say that miserable sales are good or that the HomePod’s launch has been all rosy; I just wouldn’t read too much into this report.

Meanwhile, Juli Clover at MacRumors is reporting that Siri has been updated with — and you’re not going to believe this — more jokes. Truly, what we have all been waiting for.

Behind the Music

After I linked to Kirk McElhearn’s piece about Apple Music’s limited search capabilities, Erin “Syd” Sidney pointed me to a three-year-old post he wrote about the lack of detailed creator information available on the platform:

Liner notes are how my friends became walking encyclopedias who could draw a line, no matter how thin, between records that spanned genres and generations.


Songwriters whose work we could admire and follow as they provided the musical framework for artists to develop. Producers. Engineers. Humans.

Each one of these people represents an industry, one being bulldozed over by what appears to be simply a lack of attention to detail.

Music purchased via the iTunes Store has long included a PDF version of the album booklet, and went even further in 2009 with the introduction of the interactive iTunes LP format. In an amazing coincidence, Apple just recently stopped accepting new iTunes LPs in the Store.

While I don’t think the full experience of the iTunes LP format was successful, I wish elements of that could be brought into Apple Music. Hip-hop producers and a handful of rock producers are well-represented in Apple Music playlists, but imagine if you could get detailed information about any track. Lyric support, introduced in iOS 10, is a great start for listeners to begin to explore music in greater depth,1 but songwriters, engineers, musicians, and non-superstar producers regularly go uncredited.2

Note that the absence of this information isn’t necessarily a technical issue. My understanding is that major artists submit directly to streaming platforms with track metadata set according to the ID3 spec and album metadata added separately; indie artists submit this information via intermediaries like CD Baby and Tunecore. If you’ve ever edited track or album information in iTunes, you’re familiar with several of the fields ID3 supports. However, there are several fields not shown in iTunes that are also supported, including the “TIPL” field, which stands for the “involved people list”.

It would certainly be a Herculean effort to add this information to all of the tens of millions of tracks in Apple Music — an effort that, in my fantasy world, would be totally worth it. For starters, many producers and songwriters are known for particular styles; adding more of this information could make for more accurate suggestions. But, along the lines that Sidney writes, it could also encourage deeper user discovery. There’s nothing like working your way through a songwriter’s catalogue, or understanding the widely-varied engineering career of someone like Steve Albini, or grasping the scope of every album Bob Ludwig has mastered.

  1. Not only does Spotify display lyrics, they went one step further and built in Genius support to help explain the lyrics. ↩︎

  2. I wonder if this is one reason why producers like Metro Boomin and Murda Beatz frequently tag the tracks they produced with a unique vocal signature. ↩︎

The Joys of Data Hygiene

The Economist describes Europe’s new data privacy law, GDPR, set to go into effect on May 25:

The new law was mostly written by privacy-conscious Germans. Consent to collect and process personal data now has to be “unambiguous” and for “specific” purposes, meaning that catch-all clauses hidden in seldom-read terms and conditions, such as “your data will be used to improve our services”, will no longer be sufficient. “Data subjects” can demand a copy of the data held on them (“data portability”), ask for information to be corrected (“right to rectification”), and also request it to be deleted (“right to be forgotten”).


As a result the GDPR ensures that all organisations which collect and keep data will take their use (and abuse) much more seriously. Take the fines. Under the GDPR’s predecessor, an EU directive dating from 1995, fines were negligible. The upshot was that firms gave data protection little attention and few resources. But the risk of hefty penalties has raised privacy to a board-level matter. “We have support from the top down,” says Susan Bandi, who is in charge of data security and privacy at Monsanto, an agrochemicals company.

There has never been a more consumer- and person-friendly data privacy law than GDPR. We can all hope for a ripple effect where adhering to GDPR’s rules becomes the easiest solution for companies worldwide; unfortunately, that’s not likely for giants like Facebook and Google. But it is a huge step forward for Europeans, and a model of what a good personal data protection law looks like.

Updates From Indonesia

I returned recently from another trip throughout Indonesia and want to share some observations and updates since the first time I visited two and a half years ago.

When I visited in 2015, 3G service was the norm, and even in densely-populated areas of Java and Bali, it was typical to see only two or three bars. Now, strong LTE service blankets much of both islands. That’s important: the only internet connection many people have is through their smartphone.

It’s also great for battery life. The iPhone X I used on this trip has a much bigger battery than the iPhone 6S I used last time, which obviously contributes to longer battery life, but so does the quality of the cellular signal.

One of the more notable changes is that most people are now carrying one smartphone, as opposed to the two or more per person that I saw previously. There are, I suspect, several reasons for this — phones are better, there’s better balance between performance and battery now, and high-end smartphones are more expensive — but, based on what I’ve been told and what I can figure out with the limited online reporting on this, it appears that Indonesian law now requires cellular plans with voice and SMS capabilities to be associated with a national ID number when they are registered. I don’t think this means that someone can’t have two or more cell plans, but my understanding is that it’s discouraged.

For that reason, I had a data-only plan purchased for me, with no voice or SMS capabilities. This time, I didn’t need to power-cycle my phone for my Telkomsel SIM to be recognized, but a weird thing happened where, because iMessage couldn’t send and receive its authentication text messages, it was unable to complete its setup on the Indonesian number.

The pre-paid SIM offer I got is no longer available, but it cost about $10 for 11GB. Instead of being in a single bucket of bytes, my data allotment was split: 7GB of general data, 2GB for WhatsApp and BBM, and 2GB for “VideoMax”. The general 7GB bucket was also split into different amounts for 3G and LTE data, and roaming within the country — the SIM card was purchased in Surabaya, but I travelled to Lombok, Bali, and Semarang as well. It wasn’t quite clear how this data was split up; all I know is that, after two weeks, I got a text message from Telkomsel that I had zeroed out my data allotment after using just 2GB, according to iOS’ cellular settings. Data continued to flow, however, without topping up the card.

I find this plan’s separation of data into different buckets confusing and ultimately unhelpful. If I don’t use WhatsApp or BBM, I forfeit 2GB of my plan; conversely, if I were a heavy user of these services, I would have to stop after 2GB was used, even if I had a lot of data available in the general bucket.

In addition to my phone, I also brought my MacBook Air on this trip: it’s a great — and legal — backup battery, and it means that I can offload photos from my SD card every evening and back them up for safety.

Unfortunately, MacOS tends to be quite aggressive about its internet use when given the opportunity, and there are limited controls to restrict it. For example, I have automatic software updates enabled, which means that hundreds of megabytes-to-gigabytes download in the background, even on lower-bandwidth connections. This is good for my computer’s security, but it can be a bit rude when using someone else’s internet connection with a monthly bandwidth cap, or a portable wireless hotspot. Furthermore, I use iCloud Photo Library, which tends to monopolize bandwidth while it uploads all those RAW photos.

There are controls to switch these functions off individually — though the button to pause iCloud Photo Library uploads did not reliably appear for me — but I feel like there should be some sort of global option to restrict the bandwidth consumption of these system service. MacOS could also do a better job managing this automatically. A third-party app called TripMode appears to work well for this — I just didn’t discover it in time for this trip.

Last time I visited, Samsung and LG phones were everywhere, but so, too, were BlackBerrys. Still. Now, the BlackBerrys are gone and, while the two giant Korean companies remain popular, newer brands from China are on the ascendance. Everywhere I went, I saw loads of people using phones from Vivo and Oppo. It was impossible to miss the giant green Oppo banners hung outside seemingly every phone vendor’s store. Both companies make shameless iPhone clones with iOS-styled versions of Android. iPhones remain very expensive in Indonesia: a 64 GB 4.7-inch iPhone 8 is Rp 12,599,000 — about $920 USD or nearly four months of minimum wage earnings in Jakarta.

Uber wasn’t able to make inroads in Southeast Asia, but two other companies have taken Indonesia by storm: Grab, which acquired Uber’s Southeast Asian business, and Go-Jek. Both operate platforms for multiple services. Go-Jek, appropriately, offers rides for a single person on a motorbike, but they also have car drivers and a partnership with Bluebird taxis for fixed-rate fares. In addition, they provide food delivery and even have a payment service built in. Based on what I’ve read, Grab is similar, but I only used Go-Jek.

On a non-technical note, leaving Indonesia for a second time was even harder than the first. It was and remains a beautiful country full of exceptionally generous people, delicious food, beautiful weather, and a depth and breadth of culture. I can’t wait until I get to go back.

Rethinking the Apple Watch Platform

Manton Reece, reacting to stats that David Smith posted of the adoption rate of the Series 3 Apple Watch compared to prior versions:

The big difference between the Apple Watch and the original iPhone or iPad is that many people (perhaps most) do not run third-party apps on the watch. Those people are not even counted in David Smith’s numbers. Unlike the iPhone and iPad, which are significantly improved with new apps, the Apple Watch is pretty good with only the built-in Apple features.

The Apple Watch, as a product, has become very successful, and I know I use mine regularly throughout the day. But it has not been a good platform for third parties. Opening iOS up to third-party developers was instrumental in the success of the iPhone and iPad, but I’ve seen no evidence of a correlating effect between WatchOS and the Apple Watch.

An easy answer to this is that third-party WatchOS apps simply don’t need to exist, but I don’t think that’s the case either. There are plenty of instances where you might want to run a non-Apple app on your watch. I know that I would love to use Transit to know, at a glance or with a voice command, when the next train will arrive. But the platform simply isn’t there yet, and likely won’t be until Apple starts using the same tools as third-party developers.

Mark Zuckerberg’s Fourteen-Year Apology Tour

Zeynep Tufekci, writing in Wired:

Facebook’s 2 billion users are not Facebook’s “community.” They are its user base, and they have been repeatedly carried along by the decisions of the one person who controls the platform. These users have invested time and money in building their social networks on Facebook, yet they have no means to port the connectivity elsewhere. Whenever a serious competitor to Facebook has arisen, the company has quickly copied it (Snapchat) or purchased it (WhatsApp, Instagram), often at a mind-boggling price that only a behemoth with massive cash reserves could afford. Nor do people have any means to completely stop being tracked by Facebook. The surveillance follows them not just on the platform, but elsewhere on the internet — some of them apparently can’t even text their friends without Facebook trying to snoop in on the conversation. Facebook doesn’t just collect data itself; it has purchased external data from data brokers; it creates “shadow profiles” of nonusers and is now attempting to match offline data to its online profiles.

Again, this isn’t a community; this is a regime of one-sided, highly profitable surveillance, carried out on a scale that has made Facebook one of the largest companies in the world by market capitalization.

As is often the case with one of Tufekci’s pieces, this is a must-read in full. I pulled the above quote because I think it illustrates the depth and breadth of Facebook’s business model and its intrusiveness in the public sphere, even among those who are not registered users. I don’t think it’s possible to grasp the scale of their power and influence, but Tufekci comes close.

This Could Revolutionize That

Kieran Dahl, writing for the Baffler:

Hyperbolic language is nothing new in Silicon Valley, of course. But could revolutionize presents the tech media at its worst. The phrase’s juxtaposition of two contrasting words — could implies a distinct possibility of something not happening, while revolutionize means the strongest possible version of a change to something’s fundamental nature — is manipulative. No one clicks a headline that reads, “X might make an impact on Y,” no matter how intriguing the X or culturally relevant the Y. But could revolutionize is an enabler, a gateway drug into the world of false hope, hedging, and bright-eyed optimism that cyclically drive Silicon Valley into a frenzy. When could revolutionize is used in a headline, the article automatically falls Connect Four-style into one of two categories: a tepid argument for X’s tenable but ultimately minor effect on Y, or a fawning quasi-press release.

Now that privacy and security concerns and a general wariness of Silicon Valley have reached a more mainstream audience, I have to think that could revolutionize is now cause for alarm, as in: this hyped startup might carpet bomb an industry for its own short-lived success before fizzling out in the wake of a massive controversy.

Remember the Mac Mini, Too

Joe Rossignol, MacRumors:

Mac mini is three-and-a-half years old: Apple today confirmed that its revamped Mac Pro will be released in 2019. It’s an opportune time for a reminder that the Mac mini hasn’t been refreshed in three-and-a-half years as of April 16. 1,267 days ago as of today, according to our MacRumors Buyer’s Guide. We asked Apple for a comment, but it’s unlikely they’ll break silence.

Last year, the Mac Mini was upgraded from “a product in [Apple’s] lineup” to “an important part of [Apple’s] product line going forward”; Panzarino made no mention of any status change indicated during his Mac Pro briefing.

And that’s weird. Half of the Mac models Apple ships are stale. It isn’t just me who finds that strange, right? We now know that a new Mac Pro is coming next year, and we know that the likely fate of the MacBook Air is that it will be replaced by the MacBook when the latter’s price point allows it. But what’s the likely roadmap for the Mac Mini? I’m kind of intrigued that it’s unclear — maybe it will be something more like an Intel NUC, or perhaps something even smaller — but I’m also worried that it’s another product Apple won’t make minor updates to because they’re too busy reinventing it. I don’t think it’s unreasonable to expect them to do both.

The Story of Windows’ Decline

Ben Thompson:

The story of Windows’ decline is relatively straightforward and a classic case of disruption:


What is more interesting, though, is the story of Windows’ decline in Redmond, culminating with last week’s reorganization that, for the first time since 1980, left the company without a division devoted to personal computer operating systems (Windows was split, with the core engineering group placed under Azure, and the rest of the organization effectively under Office 365; there will still be Windows releases, but it is no longer a standalone business). Such a move didn’t seem possible a mere five years ago, when, in the context of another reorganization, former-CEO Steve Ballmer wrote a memo insisting that Windows was the future […]

It’s like seeing the Kübler-Ross stages of grief turned into a corporate strategy.

Apple’s Next Mac Pro Will Arrive Next Year

The headline is the bad news. While that was the implication at last year’s press briefings, we now have a year. We don’t yet have a timeframe for when in 2019, but I sincerely doubt we’ll see it before WWDC 2019.

Of course, that will mark six years since the current iteration of the product was introduced, never to be upgraded in its history. Six years without an upgrade — not just without a major upgrade, but without an upgrade at all — is an embarrassing black mark on Apple’s history of pro products. The only change made to the current product was to slide the mid-leve configuration down to the entry-level slot; that change was made last year, which is far too late.

Where things get more frustrating, from my perspective, is that it feels like the Pro is stuck in a position of not being released until it is a perfect rethinking of what a professional Mac should be.1 Matthew Panzarino of TechCrunch got to interview core members of the Mac Pro team recently, in a quasi-sequel to last year’s press briefings:

Now, it’s a year later and Apple has created a team inside the building that houses its pro products group. It’s called the Pro Workflow Team, and they haven’t talked about it publicly before today. The group is under John Ternus and works closely with the engineering organization. The bays that I’m taken to later to chat about Final Cut Pro, for instance, are a few doors away from the engineers tasked with making it run great on Apple hardware.

“We said in the meeting last year that the pro community isn’t one thing,” says Ternus. “It’s very diverse. There’s many different types of pros and obviously they go really deep into the hardware and software and are pushing everything to its limit. So one thing you have to do is we need to be engaging with the customers to really understand their needs. Because we want to provide complete pro solutions, not just deliver big hardware, which we’re doing and we did it with iMac Pro. But look at everything holistically.”

This sounds great. Apple is taking the time to really understand where professional users’ sticking points are and address them — whether in improving hardware design, fixing software bugs, or addressing incompatibilities with system components — in current products and using that understanding to guide the future Mac Pro. None of this is bad news, and Ternus even suggests that this research will also influence MacBook Pro updates as well:

“Well, it’s a need for some of them,” adds Ternus. “I want to be clear that the work that we’re doing as a part of the workflow team is across everything. It’s super relevant for MacBook Pros, it’s super relevant for iMacs and iMac Pros and in the end I think it helps us in dialogue with customers to figure out what are the right systems for you. There is absolutely a need in certain places for modularity. But it’s also really clear that the iMac form factor or the MacBook Pros can be exceptionally good tools.”

Where I think this whole saga gets very frustrating for a lot of current and potential Mac Pro customers is that Apple is describing a product — a powerful, professional-grade, modular desktop computer — that already exists: it’s the tower-style “cheese grater” Mac Pro. While Apple is working away to reinvent one of the most critical components of a professional user’s workflow, those users are stuck with product choices that may not quite fit.

Though last year’s mea culpa acknowledged the weaknesses of the current Mac Pro, I think Apple should have taken it a step further and taken the PR black eye by pulling that product from the market, replaced it with the old cheese grater in a more current configuration, and kept iterating in it while developing the new Pro. I have to think there was something technically fraught with doing so; and, now, it’s probably too late.

As it is, Pro customers that need a modular product are once again left in limbo as they await a reinvented high-end Mac. I hope it’s worth the wait, but several professional users have indicated that they don’t trust Apple to get it right.

Or, see Dr. Drang’s more succinct version of this argument:

Apple will be taking an extra year to design the only product in its lineup whose buyers don’t care about its design.

I’m optimistic that there’s a good reason to take several years to build and ship a rethought product worthy of the Mac Pro badge. But I’m also realistic: that’s a very long time to ship a revolution, when what many users want today is an evolution.

  1. There are plenty of users — yours truly included — who have expressed a desire for Apple to slow down and get things right. I don’t think you’ll find anyone who thinks that Apple is moving too quickly with the Mac Pro. Six years between updates is a lot↩︎

Facebook Says Data on Most of Its Two Billion Users Vulnerable

Sarah Frier, Bloomberg:

Facebook Inc. said data on most of its 2 billion users could have been accessed improperly, giving fresh evidence of the ways the social-media giant failed to protect people’s privacy while generating billions of dollars in revenue from the information.

The company said it removed a feature that let users enter phone numbers or email addresses into Facebook’s search tool to find other people. That was being used by malicious actors to scrape public profile information, it said.

Well, yes, of course it was. Facebook is a website that centralizes the conversion of abstract, individual pieces of personal data for over two billion people; that’s a golden opportunity for any data miner.

That’s not just me saying that with hindsight, either. Jeremy Kirk reported on this capability for PC World over four years ago. A couple of years ago, Slawomir Tulski built a proof-of-concept way to match Facebook and LinkedIn profiles using, in part, Facebook search. A quick web search will return dozens of discussions about the possibilities of using Facebook search to scrape profiles. This shouldn’t be the first time the company has realized that creating a powerful search engine for a third of the world’s population could be misused.


Facebook also said data on as many as 87 million people, most of them in the U.S., may have been improperly shared with research firm Cambridge Analytica. This is Facebook’s first official confirmation of the possible scope of the data leak, which was previously estimated at roughly 50 million. It has resulted in calls from legislators and policymakers for greater regulation of social media, helping to shave billion of dollars from the company’s market value.

Like Equifax’s massive breach last year, I’d bet good money that Facebook’s value will return to its previous high within a year or so. There’s simply no lasting consequence for not adequately containing the data of millions or billions of people when the company responsible is as entrenched and as powerful as these giants are.

Apple Hires John Giannandrea

Jack Nicas and Cade Metz, New York Times:

Apple has hired Google’s chief of search and artificial intelligence, John Giannandrea, a major coup in its bid to catch up to the artificial intelligence technology of its rivals.

Apple said on Tuesday that Mr. Giannandrea will run Apple’s “machine learning and A.I. strategy,” and become one of 16 executives who report directly to Apple’s chief executive, Timothy D. Cook.

This is a big get for Apple; Giannandrea‘s name is on a bunch of the patents that form the groundwork for Google’s “Knowledge Graph” search infrastructure. My interpretation of this is that it means that he’s familiar with making assistant-type software more understanding of user intent; however, it should also be noted that an extension of the Knowledge Graph has been a source of pain for Google as well.

As this Times story explains, Apple has been making big hires in the machine learning space, and MacStories’ John Voorhees points to over a hundred Siri-related job postings on Apple’s website. What I’ve long wondered is how much of Siri’s lacklustre qualities can be blamed on a lack of staff, how much is the fault of Siri’s managers, and how much is a product of the way the company operates. If it’s primarily the first and second, then hiring renowned leaders and additional staff across Siri’s various teams could make a positive contribution. But if it’s the third, it’s going to require more than a few key hires. What’s most interesting, I think, about Giannandrea’s hire is that he’s reporting directly to Tim Cook. Hopefully, that means a clear direction for what Siri’s capabilities ought to be, and how we should expect it to function.

Drivers Report That Google Maps Isn’t Reliable in Indonesia

Resty Woro Yuniar, writing in the South China Morning Post:

“Google Maps has often guided me farther away from my destinations. I’ve had to call my customers, which means I have to keep buying phone credit just so I can pick-up passengers or deliver their packages,” Nurani says. “My income is not that big and I can’t keep wasting my money on phone credits.”

Nurani isn’t alone. Many drivers with ride-hailing companies in the Southeast Asian nation are less likely to rely on digital navigation tools than their Western counterparts, as online maps can be full of glitches and lack short cuts for two-wheelers – a common mode of transport in the region. Weak internet connectivity also causes headaches for drivers, something that Agus Saputra, another Go-Jek driver in Jakarta, is all too familiar with.

“One time I was following Google Maps, and suddenly it just stopped because I lost the signal, I was confused because I didn’t know where I was,” Saputra said.

I’ve been travelling around Indonesia for the past couple of weeks — though not to Jakarta — and I’ve seen similar issues. Google Maps frequently doesn’t display local businesses, and it hasn’t always shown roads accurately: sometimes they exist in reality but aren’t shown on the map, while others aren’t correctly shown as one-way streets. And that’s Google Maps, in well-populated areas of a country where Android has a market share of over 90%. I’ve barely opened Apple Maps on this trip because it makes the country’s second-largest city look like a ghost town.

But, as Yuniar explains, maybe there’s an advantage to the biggest Silicon Valley firms so far failing to accurately map Southeast Asia:

The secret to Grab’s success can be traced back to two years ago when the company deployed resources to improve mapping data.

This effort resulted in more than 3,000 new, precise pick-up points across Southeast Asia.

These in-house, localised data complement existing data provided by commercial maps that Grab uses such as Google Maps, Foursquare, and Nokia’s HERE, among others. Grab also has created algorithms to help drivers obey traffic laws, for example, like an odd-even car licence plate rule in Jakarta that sees vehicles take to the road only on alternate days and a regulation in Hanoi that bars contract cars with fewer than nine seats from 11 roads during peak hours, says Ajay Bulusu, regional head of map operations at Grab.

A recurring point I’ve made on this website for a couple of years now is that giant American companies often have too much influence over other countries’ communications and web infrastructure. I still think that’s the case, but their blind spots can help encourage local development. Given the size of the largest American tech companies, though, that also makes these smaller businesses prime acquisition targets; I wouldn’t be surprised to see an offer made for Grab.

Apple Music Doesn’t Let Users Search for Composers

Kirk McElhearn:

I know, classical music is a small share of the overall music market. But it’s still an important part of the overall music landscape, and if there are currently 36 million Apple Music subscribers, that means there are at least a couple of million people who listen to classical music.

Yet you cannot search for composers.

You see some composers listed as “artists,” you see their names in the titles of albums (for a number of years, many if not most classical albums that feature music from a single composer have that composer’s name at the beginning of the title), you may see playlists with a composer’s music, you even see “songs,” but you cannot see all the music by a composer.

It’s not just composers excluded from Apple Music’s search function, either: searching Apple Music seems to be limited to title, artist, album, and genre fields. Even something as basic as the year of release cannot be searched even amongst local tracks on iOS, and I find that completely absurd. I would love nothing more than to see a modernized version of the column browser better tailored for Apple Music’s vast library.