Month: May 2016

Remember yesterday? Remember that weird rumour that said that Apple was planning on killing iTunes music sales within a couple of years? Michael Rockwell looked into the source:

I did some digging on a few Apple rumor sites — MacRumors, AppleInsider, and 9 to 5 Mac — to see how accurate Digital Music News’ reporting has been on companies’ future plans. I was only able to find three instances of original reporting from the site. The published rumors pertained to Spotify and Apple, but none of these stories were proven true.

Digital Music News can sit conveniently next to Digitimes both on an alphabetic scale, and a reliability scale.

Facebook is having a bit of a rough week. On Monday, Michael Nunez of Gizmodo published a report in which a former Facebook employee alleged that the company actively suppressed conservative topics:

This individual says that workers prevented stories about the right-wing CPAC gathering, Mitt Romney, Rand Paul, and other conservative topics from appearing in the highly-influential section, even though they were organically trending among the site’s users.

Several former Facebook “news curators,” as they were known internally, also told Gizmodo that they were instructed to artificially “inject” selected stories into the trending news module, even if they weren’t popular enough to warrant inclusion—or in some cases weren’t trending at all.

This has made some waves, to put it mildly. There are two different stories here: that news stories of interest to conservatives are deliberately being omitted, and that non-trending stories are manually being included. Happily, they can largely be addressed together thanks to a great scoop for the Guardian’s Sam Thielman:

Leaked documents show how Facebook, now the biggest news distributor on the planet, relies on old-fashioned news values on top of its algorithms to determine what the hottest stories will be for the 1 billion people who visit the social network every day. […]

[The] documents show that the company relies heavily on the intervention of a small editorial team to determine what makes its “trending module” headlines – the list of news topics that shows up on the side of the browser window on Facebook’s desktop version. The company backed away from a pure-algorithm approach in 2014 after criticism that it had not included enough coverage of unrest in Ferguson, Missouri, in users’ feeds.

Facebook’s statement that they “do not insert stories artificially into trending topics, and do not instruct our reviewers to do so” is, therefore, wrong. According to this document, there is near-constant manual human intervention to blend similar topics together, add topics to the list, remove topics when they become stale, and make all kinds of adjustments.

But are they suppressing conservative news? The document obtained by the Guardian doesn’t say that, but it does clarify a topic’s eligibility for newsworthiness:

You should mark a topic as ‘National Story’ importance if it is among the 1-3 top stories of the day […] We measure this by checking if it is leading at least 5 of the following 10 news websites: BBC News, CNN, Fox News, The Guardian, NBC News, The New York Times, USA Today, The Wall Street Journal, Washington Post, Yahoo News or Yahoo.

That list of websites is pretty old-guard, and there’s nothing to suggest that it’s deliberately omitting conservative viewpoints or news. As Fusion’s Kashmir Hill says, it might simply be a quality barrier:

Regarding these particular topics being omitted by curators, New York Magazine’s Brian Feldman writes, “Given that list of overlooked topics, which range from IRS conspiracy theories, to an unreliable news aggregator, to a brutally unfunny conservative comedian, can you blame them?” […]

[What] that suggests is that Facebook preferred that news come from non-biased sources. Which is not a crazy thing to do. And it suggests that the bias might exist for news from the other side of the aisle as well, but it seems that Gizmodo didn’t rigorously assess whether liberal news and news sources were ignored by curators.

This story has understandably riled up reliably conservative media personalities, but it’s only newsworthy if Nunez did his due-dilligence to examine whether conservative stories were singled out.

Whatever the case, it’s clear that what remains unanswered is Facebook’s responsibility for its new status as one of the top referrers to any major news site. Brian Stelter for CNN:

Facebook has a unique ability to turn on a firehose of traffic — and the ability to turn it off. Publishers may not live or die by Facebook alone, but they certainly thrive or struggle based on the company’s decisions.

So Gizmodo’s recent reports about the production of Facebook’s ‘trending” stories have gained a ton of attention. Journalists, academics and some average users want to understand how and why Facebook does what it does.

This isn’t exclusive to Facebook; Google, Twitter, and Apple all have their own ways of displaying ostensibly algorithmically-determined trending news stories. All of them also lack transparency in how these stories are determined.

Update: Facebook has officially published the 2016 version of the document leaked to the Guardian. There are some redactions, but they appear to be nothing more than internal points of contact. Meanwhile, Sarah Emerson of Vice ponders whether Facebook lied in its earlier press statement.

Jennifer Kennard (via Coudal Partners):

The Book Club of California has announced the release of a definitive account of Herman Zapf’s enduring typeface Palatino, written and designed by Canadian poet, historian, linguist, typographer, and book designer, Robert Bringhurst. Published nearly a year after Zapf’s death at age 96, Palatino, The Natural History of a Typeface, explores the evolution of his most ambitious design project; from his original sketches in 1948, to the first trial cutting of Palatino roman at the Stempel Foundry in 1949, and to the last authentic digitally drawn members of the family which Zapf himself carefully supervised just ten years ago.

Palatino is one of my favourite type families; I even used it on this very website at one point. The book isn’t inexpensive, as you can imagine, but it looks absolutely tremendous. It’s going on my wish list.

There’s a chain of articles that I stumbled through today with all the grace of someone falling down a flight of stairs, angrily cursing with each bounce.

An article by Andrew McGill of the Atlantic was the originator:

When they were first launched in 2015, emoji skin tones corrected an obvious wrong. Previously, if a black man or a Latino woman wanted to text a friend the thumbs-up emoji on an iPhone, a white hand would show up. The Unicode Consortium’s solution made the new default color a Simpsons yellow and allowed users to tint certain emoji with one of five skin tones, ranging from “pale white” to “darkest brown.” […]

But as emoji with skin tones spread to Twitter, Facebook, and workplace chat applications like Slack, I noticed something I hadn’t expected: While I saw plenty of Thumbs up and Thumbs up, almost no one I knew used the lightest skin tone, or even the second-lightest. Indeed, as a white man who tends to be either pale or sunburnt, I had never considered using it myself. When I did switch briefly to the lightest tone at work, it felt… weird. […]

Perhaps the squeamishness on the part of whites has more to do with the acknowledgement that only white people hold this special privilege; to use the white emoji is to express a solidarity with people of color that does not exist.

So far, this makes sense: depending on the context, using white emoji can either feel like a way to represent your own skin tone, or it can feel oppressive for the reasons elucidated by McGill.

Here’s where this whole thing breaks down:

When white people opt out of racemoji in favor of the “default” yellow, those symbols become even more closely associated with whiteness — and the notion that white is the only raceless color.

Eli Schiff adds:

It follows that in using a yellow emoji, a white person is presuming whiteness as a default, and therefore reinforcing racist prejudice by preferring whiteness to be the default signifier.

And John Gruber, too:

I’ve been wondering about the decision to use yellow ever since iOS started supporting the skin tone variants. It still seems “sort of white”, in a way that a Smurf-y blue or Hulk-y green would not.

I’ve never questioned the choice of yellow as the default emoji colour primarily because it’s the generic colour assigned to little smiley faces since we called them “emoticons” or “smileys” back in the days of iChat, AIM, MSN Messenger, and Yahoo Messenger.

Andrew Yang:

That white people think yellow emoji are white seems more a problem of entitlement and appropriation than of yellow emoji.

Perfectly said. In no sense is the colour yellow presumed to be representative of whiteness, or even race as a whole. It is cartoonishly unrealistic, serving only to enhance the emotion, not to convey any kind of racial characteristics.

Where Schiff’s post goes off the rails — and why I’m surprised Gruber linked to it — is where he paints white people as being oppressed by all this:

It is therefore quite strange that yellow (white) emoji were set as the default, given that not assuming all users to be white was the entire premise behind making the new diverse set of emoji. In this way, the Unicode Consortium’s efforts to achieve a more inclusive solution only served to doubly reinforce a racism of defaults. […]

It is not simply that it is problematic for whites to use the white emoji, but so too is it racist for them to use the brown shades and the yellow default. In sum, it is racist for whites to use any emoji.

There are two choices going forward: either white users should refrain from using emoji, or an alternative default must be drawn. Perhaps green, blue or purple would be an ideal choice as they don’t have racial connotations.

What utter horse shit.

The initial emoji spec was white-only, and the change to yellow as a default and the addition of other skin tones was an inclusive move. The idea that white people are, for some reason, racially persecuted by emoji and are left with no choice but to not use them is nonsense.

Furthermore, the universal choice of yellow is not by accident and is not defined by the spec. Microsoft’s race-neutral colour of choice was grey, but that was ugly. It could be green or blue, but that would just look ridiculous, wouldn’t it?

Use whatever you like while being cognizant of its context. And don’t be a racist dick. It’s not hard.


Update: More grey Microsoft emoji, which they’ve since updated to be — big surprise — yellow by default.

Update: The screenshot that Schiff includes purportedly showing that the default yellow left-facing hand is considered “white” by Unicode was taken out of context.

Update: Per Ken Weingold in a Slack chat, the yellow smiley face originates with Harvey Ball in 1963. I’d say that the yellow instant messenger emoticons and today’s yellow emoji simply reference that, and nothing more.

Update: Fine, I’ll tackle one more thing that’s been nagging me since publishing this. Schiff closes his “Racemoji” post so:

This is an example of the meaningful political and cultural dilemmas that designers would not have had to consider 40 years ago. Back then, we had more homogenous societies, so inclusive representation was a less pressing concern. Now, designers must be more considerate and take seriously the power of defaults.

This has been biting at me since I read the initial post, but I’ve been having a hard time coming up with an adequate response.

Forty years ago, people of other races still existed; they were, however, ignored and persecuted. We should have treated everyone equally then, but we didn’t. We still don’t, as evidenced by the first white-only rollout of emoji.

It’s not “considerate” to recognize that there are non-white people in the world. It’s not even basic decency. It should be an expectation.

Schiff has updated his post to include Google’s proposal (PDF) for more gender-equal emoji, particularly in regards to the suite of characters that reflect specific professions. Again, this should have been a given. However, as tech companies remain dominated by white men, consideration of different ethnicities and genders remains an afterthought.

Cabel Sasser is rightfully proud of Panic’s new(-ish) sign. I wish the tenant on the opposite side of the street would install a webcam, though, because it would be cool to see the colours change.

Now there are three words I didn’t expect to be writing before we’re all using holographic interfaces or something.

Armin Vit:

The ensuing shitstorm on the internet today will be epic. About 75% of the negative reaction will be simply to the fact that it has changed and the other 25% will be to the not-quite-fact that there is a generic aesthetic to the new icon where it could be a “camera” icon for the upcoming smart microwave from Apple or whatever other user interface you would imagine. This is not to say it’s a bad-looking icon, no… as far as camera icons go, this is quite lovely and has the minimal amount of elements necessary to be recognized as a camera BUT not the minimal amount of elements necessary to be recognized as Instagram.

I think Vit nailed it with that last sentence. While I like the icon with its insane gradient orgy,1 I’m not seeing it as Instagram — yet. It lacks any sort of tie or bond to the old icon in shape, colour, or — blessedly — style. And this is something they solved while they were working on this refresh; I grabbed this still of one of the rejected icons from Instagram’s announcement video. It feels less like an update to bring it in alignment with the rest of the icons on my home screen and more like an entire rebrand. To my eye, though, the “Instagram” script running across the top of the app remains untouched.

The app itself received a black-and-white refresh, too, and I’m a fan. After seeing that bright blue bar every day for the past five-and-a-half years, I didn’t notice how much it distracted from the photo feed until it was removed. The adjustment tool icons are particularly well-drawn.

Update: Do yourself a favour and check your push notification preferences. I found a bunch of things that I previously had switched off were back on again.

Update: If you’re wondering what the icon sounds like, I got the Yams to make me a playlist.

  1. The gradient’s palette might be a bit much. ↥︎

If you shop on the web as much as I do and you use Safari, you’ve probably come across this notice sliding into the bottom of your browser window:

Your browser blocks some 3rd party cookies by default. By clicking on this page you allow our partner Criteo to place its cookies and serve personalized ads. You can read more or disable Criteo ads here. This notice only appears once.

I’ve seen this banner on the websites of retailers ranging in size from boutiques to major brands, and it seemed a little bit fishy to me with its super careful wording. So I did a bit of digging.

Let’s unpack that statement a little bit at a time. First, who is “[their] partner Criteo”? To get a complete picture of exactly who Criteo is and what they do, I started where anyone would — on their about page:

Unlike the vast majority of the market, we use a transparent cost-per-click model and we measure value purely on post-click sales. This demanding model is supported by ongoing, automated learning of each shopper wherever they are in mobile apps or online, along with your campaign’s performance against thousands of variations of dynamically-created ads.

Criteo are, in short, an advertising technology firm with a retargeting product. That’s nothing particularly special; there are dozens of companies that do that, as amply demonstrated by this aneurysm-inducing chart.

But Criteo has another trick up their proverbial sleeves. At the end of 2013, they got some press for a feature dubbed their “mobile web solution”. Judith Aquino of Ad Exchanger explains:

Similar to their desktop counterparts, mobile browsers handle first- and third-party cookies in different ways. Google’s mobile Chrome browser, for example, allows all cookies by default and allows users to switch to more restrictive options. Apple’s mobile Safari browser accepts cookies from sites users have visited (i.e., first-party cookies) and blocks third-party cookies by default, though users can change their privacy settings.

And while Mozilla announced early this year that it was experimenting with blocking third-party cookies from its mobile Firefox browser, the company has since delayed this plan.

[Chief product officer Jonathan Wolf] declined to discuss in detail how Criteo navigates the browsers’ various criteria but pointed out that after releasing its mobile web solution to a handful of customers in late September, the company claims to have delivered “at least two billion” mobile ads among 20 countries.

This, rather conveniently, brings me to the first sentence in that notice you see when visiting a Criteo-enabled website: “Your browser blocks some 3rd party cookies by default”. As Aquino explained, different browsers treat cookies differently by default. Safari and Mobile Safari behave a little differently than Chrome or Firefox by default, in that they disallow third-party cookies other than those from sites navigated to by the user.

For example, if you’ve never visited in Safari but you read a news site that has a Facebook “Like” button, Facebook will be unable to set its cookies. If you have visited at some point, it will be able to set cookies via that news site. Other browsers behave differently by default, insomuch as they will allow Facebook to set cookies via that news site, even if you’ve never visited Facebook.

This default setting presents a problem for Criteo. While an average user can be expected to have visited Facebook at some time in their browsing history, almost nobody will knowingly visit or their ad server addresses. Furthermore, while desktop Safari does not represent a significant share of the global web browser market, Mobile Safari does. For Criteo’s ad targeting platform to work, they need to be able to set a cookie, and Safari is configured by default to disallow this behaviour — unless Criteo somehow forces users to visit their site.

And that’s the next part of that notice:

By clicking on this page you allow our partner Criteo to place its cookies and serve personalized ads.

Criteo relies upon a couple of small pieces of JavaScript to work. As with most scripts on the web — particularly those from companies that collect and process visitor data — it’s compacted and obfuscated, which makes it a little tricky to read.

Here’s what happens: when visiting a site that includes Criteo’s scripts, a bit of browser sniffing happens. If it’s a Safari variant — and only Safari — Criteo rewrites the internal links on the page to redirect through their domain, and it looks something like this:

The user is then sent to their intended destination page, and Criteo’s cookies are allowed to be set. All that’s needed is that split-second redirect for the first link clicked on the site.

Before I continue, I’d like to be perfectly clear: there’s nothing inherently nefarious about cookies, even those from third parties. They can be used to set language settings, store shopping cart data, or provide other services on a user-specific level.

But what Criteo is doing here is in direct violation of Safari users’ browser settings, whether they’re set explicitly or by default.

If this feels like deja vu to you, you’re not wrong. Between 2011 and 2012, Google did its best to work around Safari’s default privacy settings, too, according to the Wall Street Journal:

Last year, Google added a feature to put the +1 button in ads placed across the Web using Google’s DoubleClick ad technology. The idea: If people like the ad, they could click “+1” and post their approval to their Google social-networking profile.

But Google faced a problem: Safari blocks most tracking by default. So Google couldn’t use the most common technique—installation of a small file known as a “cookie”—to check if Safari users were logged in to Google.

To get around Safari’s default blocking, Google exploited a loophole in the browser’s privacy settings. While Safari does block most tracking, it makes an exception for websites with which a person interacts in some way—for instance, by filling out a form. So Google added coding to some of its ads that made Safari think that a person was submitting an invisible form to Google. Safari would then let Google install a cookie on the phone or computer.

Google settled this case last year for $17 million and increased oversight, without admitting any wrongdoing.

Criteo is not the only company circumventing Safari’s default restrictions — indeed, the Journal report cites a few other advertisers that used similar methods as Google. The comparison is particularly strong; Criteo also appears to explicitly target Safari users. I disabled third-party cookies in Chrome and Firefox, then visited sites known to incorporate Criteo. I was not presented with the banner notice.

All of the above is applicable to AdRoll’s retargeting scripts as well; they have a comparable method for evading Safari’s restrictions on third-party cookies. (AdRoll competes with Criteo in the retargeting space. Though lots of companies might have similar behaviour, I’m picking on these two because they’re pervasive and growing.)

The difference between what Google was caught and fined for doing and what Criteo and AdRoll are doing today is that the latter present users with a notice. However, I don’t think it’s as forthcoming as they make it seem.

For one, it’s hard to say whether the technical knowledge of most users is adequate enough to make an informed decision about first- and third-party cookies. In most cases, I do not believe users have enough information to make a decision.

Second, the opt-out setup that both ad tech companies have created — like other Ad Choices opt-outs — requires a cookie from those providers. When it expires or you clear your cookies, it will be assumed that you’ve opted back into their tracking. Update: Peter Clark at AdRoll has asked that I clarify that the banner will appear again, rather than immediately opting you in.

To make matters worse, a service both companies promote is their cross-device targeting feature “across browsers, devices, and apps”. The message is simple: Criteo will follow you everywhere, so be sure to opt out on every device and browser you regularly use.

And then there’s the implied message of the banner notice: you can either opt into retargeted advertising, or you can fuck off. Oh, and — as the banner says — the notice only appears once. Blink, and you’ll miss it.

Happily, there is a solution to this and other deceptive cookie practices. In your browser’s cookie settings, choose to allow cookies from the current website only. Additionally, you can block and using the JavaScript Blacklist extension for Safari, or the built-in JavaScript blocker in Chrome.

Alas, these are solutions that require a reasonably high level of foreknowledge. Most users will never know about these scripts and never understand what they do. And, as a result, two companies that most people have never heard of will amass a large amount of browsing data that most people would never elect to provide if given an honest choice.

Igor Bonifacic, MobileSyrup:

Early this morning, Apple significantly expanded the number of payment options Apple Pay supports in Canada, and announced that it plans to add even more Canadian banks in the coming months.


Support for Bank of Montreal, TD Canada Trust and Scotiabank debit and credit cards will be added in the coming months, according to an Apple spokesperson. […]

While TD Canada Trust declined to provide a specific launch date for Apple Pay support, the financial institution said “we mean weeks away” in an email sent to MobileSyrup.

Shit. So close, yet still so far. (I use TD.)

Launching without the initial support of TD is a little weird — so many Canadians use TD that it would be like launching without support for Chase in the United States.

Even so, this represents a huge step forward for Apple Pay in Canada, where we already have loads of contactless terminals. It’s also significant in that Apple Pay will be supporting Interac cards, our bank consortium-led debit payment system. We use Interac all the time; not supporting it would have been a notable omission.

Caitlin McGarry, Macworld:

Right now, Viv is an iOS app, though it won’t always be. You open the app and ask the assistant questions or issue commands. These can range from basic queries like what the weather forecast looks like, which Viv quickly answers with Weather Underground data, or as complex as, “Send Adam $20 for last night,” and Venmo handles the transaction. You don’t have to have these apps installed for Viv to work on your behalf. […]

While most voice assistants or bots have to be programmed with preselected responses to specific queries and pull from one domain at a time, Viv’s software figures out the intent of your question and writes itself. In response to a question like, “Will it be warmer than 70 degrees at the Golden Gate Bridge after 5 p.m. the day after tomorrow?” Viv write a 44-point program to answer with Weather Underground data, but it can combine data from several different sources to respond to a question in seconds.

Last week, Elizabeth Dwoskin of the Washington Post got an early preview of Viv:

The two [founders Dag Kittlaus and Adam Cheyer] faced a similar choice six years ago, when Jobs offered to buy their little-known app and distribute it to millions of people. Jobs took them to his home in Palo Alto, and the group talked for three hours by the fireplace. They left his home convinced that they shared a vision. It didn’t turn out quite that way.

Today Kittlaus and Cheyer find themselves in a similar position: Do they sell to a giant or go at it alone?

If this works even half as well in real life as it does in this demo, that’s going to set a very high benchmark for other companies’ virtual assistants. But, if Viv remains a third-party app, there’s no way it will have meaningful usage when compared to first-party assistants like Google Now and Siri.

According to Dwoskin’s article, Google has already made an acquisition offer. If Apple’s not trying to get Siri to these levels on their own, they’d be foolish not to at least try to woo Viv.

Update: Apple just happened to debut a Siri-focused ad today, which opens with Neil Patrick Harris requesting that Siri read him a note in a particularly stilted voice, and closes with Harris saying “I’ll be a lot more natural”. Make of that what you will.

Maya Kosoff of Vanity Fair:

Venture capitalists can probably see themselves purchasing a Juicero and keeping it on their countertops, just another gadget in their toy chest. A single, working-class mom in the Midwest wouldn’t see the point. The median American household income is about $53,657; if you’re buying a Juicero for yourself and using it to make one $8 green juice seven times a week, you’re spending about 7 percent of your annual income on a juicer. The $700 Juicero does exactly one thing with its proprietary bagged fresh produce: juice those specific blends of fruits and vegetables. A $50 food processor does a number of tasks at a fraction of the cost. Starry-eyed venture capitalists may think they’re revolutionizing the agricultural business, but in reality, they’re providing luxury services to a sliver of the top 10 percent of people in a handful of cities.

According to Gallup’s polling in the past ten years, between 15–20% of American adults struggled to afford food. The groups disproportionately most affected were black, Hispanic, and women.

See Also: Keurig’s struggle to make an environmentally-friendly version of their gross single-serving coffee pods.

Evan Malmgren, writing for the Awl:

As the longtime poster child and one-time presumptive standard-bearer of small-scale “additive manufacturing” — the technical name for the process of 3D printing, which adds rather than strips away material — MakerBot’s rapid rise and equally blistering crash has mapped closely onto the public’s expectations of the technology. The desktop 3D industry is far from dead, but MakerBot’s difficulties are rooted in a broader contraction of the consumer market. The gatekeepers of viral tech-hype have largely stopped trumpeting consumer 3D printers as a revolutionary technology. Now that the dust is beginning to settle on the MakerBot saga, it seems like a good time to ask: what was all the hype about?

In some science fiction utopia, we’d create the physical objects we need from materials that were obtained in ethically-sound and environmentally-responsible ways, putting an end to rampant consumption and greatly increasing our ability to do things ourselves.

Of course, the word “utopia” has its roots in the Ancient Greek words for “no” and “place” — that is, a utopia cannot exist.

I’ve always loved music. From a very young age, creating and listening to music has been an integral part of my life, and it remains so. So, you’ll understand if this is a bit long-winded — music matters a great deal to me.

I’ve carefully built an iTunes library from the time that I bought my first iPod in 2005 with music discovered in lots of different ways, from automated services like to personal recommendations from friends and family.

One of the services I’ve long used is Apple’s Genius feature, built into iTunes since 2008. Since its introduction, I’ve provided Apple with the entire contents of my library on a weekly basis.

So it will come as no surprise to you that I have not been entirely enamoured with the recommendations provided in the “For You” section of Apple Music. For the uninitiated, Apple Music functions entirely separately from its iTunes precursor. It uses a different library of music — though they’re almost identical — with a separate cloud storage mechanism. Even its recommendation engine is detached from iTunes, which means the past seven years of data about my library is completely forgotten.

And that, in a nut, explains why I keep seeing “Intro To…” playlists for Nine Inch Nails, Kendrick Lamar, Kanye West, Modest Mouse, Burial, Deftones, and other artists for which I have every record they’ve made. iTunes, by way of Genius, knows this. While I have listened to every song from those records at least once, iTunes knows which songs I skip regularly and which songs I’ll listen to time and time again. But Genius doesn’t talk to Apple Music, which is why my recommendations are inconsistent and frustrating, despite tapping the little heart icon on everything I enjoy.

Other algorithm-based suggestion services have also found me wanting. Perhaps it’s because computers interpret patterns differently than you or I might, making it difficult for them to be programmed to suggest music we might like, but not necessarily in a way that’s directly related to something we already enjoy.

So I’ve been hunting for a new way to find new music, but it can be hard. Everyone has different tastes and approaches new music with different expectations. You may have a multifaceted aural palette, but you may also be exploring a genre or artist for the first time. Other people may prefer very specific sub-genres of music.

From the time I got my first iPod, I’ve been reading industry publications, scrobbling my played songs, jumped between a bunch of grey-area music blogs, and juggling a bunch of other ways of surfacing new music. But all of these things require a considerable time investment that I don’t necessarily have any more. I don’t want to have to train anything either. I just want to listen to great tunes.

There is not a universal solution, I don’t think, but I’ve found something that gets pretty close to perfection. It’s called the Yams. I’ve been using it for the past few months, I think it’s terrific, and I think you should know about it. And that worries me a little bit, for reasons I’ll get into.

I’ve said some critical things about algorithmic playlist builders, but I’ll add one more to the pile: they feel pretty anodyne. By contrast, personal recommendations from a friend or a knowledgable record store clerk feel, well, personal. These are people who know their shit, they know what you like, and they’re giving you an earnest opinion. And I love that.

Along the range of almost entirely algorithmic recommendation engines — iTunes Genius or, for instance — to completely personalized — like that one friend of yours who always knows the best new stuff — the Yams feels entirely new because of how close it gets to the latter. The user experience is what defines this; the Yams exists entirely through text message conversations with a real person.

I was introduced to one of the service operators in a welcome text. His name is Miguel, he makes music as the Miserable Chillers, and he writes for Ad Hoc.1 Good start. Here’s the first conversation I had:

What kind of music would you like to hear?

I have a varied taste in music, though I typically stay out of the top 40. I’d like to listen to more interesting rock (of all genres, with the exception of chugging post-grunge and anything much louder than Deftones or Mastodon), and more interesting hip-hop and rap (I love Kendrick, Kanye, RTJ, and Odd Future). Plus, anything experimental and weird you can think of.

Cool. We can definitely dish out some weird stuff. Do you use any streaming services?

Apple Music and Spotify.

I kind of lied here — I have a Spotify account, but I haven’t used it since Apple Music launched, and I discontinued my premium subscription as soon as I received my first bill from Apple.

Perfect. We’ll get back to you soon, Nick.

A little over half an hour later, I was sent a link to a personalized Spotify playlist: “Y.O.D. – N. Heer”.2 As it was an introductory playlist and I provided a wide swathe of genres and artists, I received ten tracks that span the gamut. A mix of bluesy rock, electronic bleep-bloop stuff, some great indie hip-hop, and a little fuzzy psychedelia played for a little over forty minutes, and it was wonderful.

Every so often since, I’ve sent Miguel a text with a request for a new playlist. Last month, I was working a little late and I asked for a playlist based on Burial. A couple hours later, I received nine tracks that were completely in the same vein. Yes, there were some gimmes in there — Four Tet and Zomby — but they also popped in some XXXY and FaltyDL, both of which I’d never heard of before and thoroughly enjoyed.

Then, a couple of weeks ago, I wanted a playlist to match the weather, based on my previous requests. I got back a truly excellent melancholic playlist that was blessedly free of Bon Iver and Death Cab. Don’t get me wrong — I dig both of those artists, but I’ve heard enough “rainy day” playlists that feature them.

I think that gets to the crux what I’ve enjoyed most about the recommendations from the Yams. On other services, especially more algorithmic ones, songs and artists that are popular will tend to be recommended more often, thus becoming even more popular. The playlists I’ve received from the Yams, on the other hand, have been chock full of interesting, off-the-wall tracks that I haven’t heard before. It’s truly music discovery.

The other thing that makes it so great is the user experience. The conversational aspect of it is what helps it feel so much like asking for new music suggestions from a friend.

So why am I worried about the Yams? I have a hard time seeing how the text messaging and personalized service will scale to work with a very large user base. I want to tell everyone about it because it is truly great and innovative, but I worry that this thing that I use and love will become saturated and lose that personalized quality. In some ways, that’s already happening a little — I have occasionally felt the need to text a reminder when a playlist I requested was seemingly forgotten about.

I wanted to know more about this, so I asked Miguel who I could talk to. He suggested I email the founder and CEO of the Yams, Shannon Connolly; yesterday, I had the opportunity to speak with her by phone.

She told me that they’re structured to deal with scale. Front-line operators like Miguel, Connor Hanwick, and herself dispatch playlist requests to contributors who know what they’re talking about. Every contributor has a generally great level of music knowledge, but they all have their own areas of specialty: some are into funk and soul, while others are metalheads.

Not only does this help with scalability, it also allows for a future personalized recommendation service for pretty much anything. Want to know what book to read next or movie to watch? At some point the future, you’ll be able to text them about those kinds of media as well.

One of the things Shannon and I talked about at length was the advantage of the personalized user experience. While the Yams uses bots to automatically reply for generic messages — such as when a request is made outside of business hours — all of the interactions that occur while making a request are with an actual human being. That means that your time is not spent fighting with a bot.

The humanity of the service also creates a deeper level of interaction. Shannon told me that they’ve had users text them about specific moments in their life, like breakups, getting married, or being fired from their job. I can see another advantage to this if you want a playlist that specifically omits some songs or artists.

Aside from my concerns about scalability, the only downside I can think of is that it’s not an always-available service, as I alluded to above. It relies upon human operators, so it only really works within business hours, and it does take minutes-to-hours for a playlist to be created. So if it’s 11 PM on a Sunday, there’s little chance you’re getting that emergency study session playlist. (You can add “URGENT” to a request, but there’s no guarantee.) Their organizational structure allows them the flexibility of adding more contributors and operators around the world, so this should be alleviated.

But I think that’s okay. The Yams doesn’t feel like it’s set up to be the sole place where you find music to listen to. It’s the thing you turn to when you need something new and different. Miguel is the guy who recommends new music to me when I ask. I’ve never met him, but he knows what I like to listen to. That’s pretty cool.

The Yams is currently in a private beta, but they’re adding new users all the time. You can register on their website.

  1. The Yams has lots of contributors who actually build the playlists. Among them are: Vlad Sepetov, who has worked with a ton of great artists and designed the album covers for “To Pimp a Butterfly”, “untitled unmastered”, and “Oxymoron”; Jim Macnie, who has worked for Downbeat Magazine, VH1, and Billboard; and Peggy Wang, a founding editor of Buzzfeed. ↥︎

  2. I tapped the ‘Open Spotify’ button from the short link and nothing happened. I realized that I had deleted the Spotify app, so I redownloaded it and opened up the playlist.

    Because I no longer had a premium account, I was shown a minute-long video ad before I could play anything. Video seems a little heavy-handed, but I’m okay with ads on the free version. It’s free, after all.

    I tried tapping a song to start the playlist and then remembered that Spotify Free runs in shuffle mode only. That’s fine, I guess — I’ll just listen to the playlist on shuffle.

    Imagine my surprise when the first song that started playing didn’t seem like it fit my tastes, and I didn’t remember seeing it in the playlist. It turns out that the free version of Spotify mixes ‘suggested’ tracks with those from the playlist, even if they’re not entirely compatible.

    Imagine if you were a first-time Spotify listener, and you were exploring the various streaming music services available before committing to one. And imagine that the first song you heard on Spotify was not one in the playlist you selected, but was seemingly chosen at random. Not a good impression.

    So I added premium to my Spotify account, and tried again.3 And I was very, very happy. ↥︎

  3. Is it redundant to pay for two streaming music services? Yes, very. And I’d love to drop at least one of them. But Apple Music is integrated into iOS and if I turn it off, I’ll start seeing an invasive interstitial ad. It’s almost worth paying $10 every month simply so I don’t have to see an ad in a core app in my OS. I see that ad as one of the most inelegant and inept decisions Apple made in the past year.

    Meanwhile, I use Spotify very rarely, but as I don’t have iCloud Music Library turned on, it’s the only way I can save playlists and songs for offline listening. ↥︎

Alex Webb, Lucas Shaw, and Adam Santarino broke the news of an impending revamp of Apple Music for Bloomberg:

Apple is altering the user interface of Apple Music to make it more intuitive to use, according to people familiar with the product who asked not to be identified because the plans aren’t public. Apple also plans to better integrate its streaming and download businesses and expand its online radio service, the people said.

Mark Gurman added his own reporting:

While the most of the Apple Music service will be redesigned, much of the emphasis is on the “For You” feature, the tab that recommends songs, albums, artists, and music videos. This section will be simplified and better promoted to increase usage of the feature. While the interface will change, the functionality will use algorithms similar to today’s Apple Music recommendations engine, according to sources. The new service will also discontinue the “New” tab, which is a jumbled list of top charts, genres, featured music launches, and curated playlists. It will be replaced with a section called “Browse” which better organizes the aforementioned content. The Beats 1 Radio service is unlikely to see notable interface changes this year.

While the organization of the Music app’s interface is in need of refinements, I love the existing large artist photos and colour-matched album screens.

John Gruber:

A big aspect of iOS’s success, from day one in June 2007, is that it emphasized smaller focused apps that do less over larger monolithic apps that do more. The monolithic style leads to desktop iTunes — a single app for managing your personal music collection, buying music from the iTunes Store, buying and playing TV and movies, podcasts, iOS app purchases, and device syncing and backups. The iOS style leads to dedicated separate apps for music playback, video playback, podcasts, and store purchases. Maybe there’s a way to design “all your music in one app” that is completely clear, convenient, and obvious. But the bottom line is that a music app shouldn’t be confusing. I think that’s held Apple Music back.

Kirk McElhearn:

I’m not sure that’s the most confusing aspect of Apple Music. There are two sections, My Music – which contains music in the iTunes library connected to your device, and music you’ve added to your library from Apple Music – and the New and For You sections, which contain only music from Apple Music. You check My Music when you want to listen to music you own, or music you’ve saved, and you check the others when you want something new.


One of the problems with Apple Music is the existence of the iTunes Store. Apple can’t fully commit to Apple Music because they still need to sell music. If they didn’t have the iTunes Store, then Apple Music would have one less layer of complication.

At its core, as a pure “queue up a song I don’t own” streaming service, I have found Apple Music totally competent. However, there are lots of ways in which the service is failing to live up to the promises of last year’s introductory keynote.1

Recommendations are notoriously poor, ignoring years of Genius history and in favour of requiring the user to completely re-train its engine. iCloud Music Library is required for basic functionality like saving songs or playlists for later listening, but it’s unreliable.

Connect has not been more successful than Ping — even Drake has barely used it. This is the same Drake who appeared onstage at WWDC last year to launch Apple Music, who made his new album an exclusive to the platform at launch, and who hosts a show on Beats 1. He hasn’t posted song snippets, nor has he debuted videos of the recording process, as the keynote suggested artists might.2 Instead, his Connect account is largely a repository for music videos available elsewhere, and auto-posted recordings of his Beats 1 show. He doesn’t have time to sit around and reply to comments.

Beats 1, meanwhile, hasn’t done too badly. I and a few people I know still tune in from time-to-time, but I don’t know the schedule by heart. I’m more fond of listening to recordings of past shows on-demand.3

Despite all this, Apple Music has been a resounding success story when measured by user numbers: 13 million, as of their second quarter results. But you’ll notice that user satisfaction levels — one of Apple’s favourite metrics — have not been touted.

After reading this series of somewhat disjointed thoughts, you might reasonably conclude that Apple Music is a failed product. I do not believe this is the case. But I do think that it is unfinished, it was rushed, and it does not behave in the real world as Apple imagined it might. And, for those reasons, it is in desperate need of recalibration for how we actually listen to, discover, and share music.

Update: Mark Gurman is now reporting that the Connect elements of Apple Music will be “demoted” in iOS 10.

  1. I struggled through the Apple Music part of the WWDC keynote again so you don’t have to. ↥︎

  2. I’m not sure why this was such a focus of the marketing around Connect. I’m an artist; I’ve spent much of my life around artists. Most artists hate showing a work in progress. Trent Reznor, who narrated the introductory video, doesn’t reveal stuff that isn’t finished. Jony Ive and the rest of Apple aren’t fond of showing things they’re still working on. ↥︎

  3. As the rest of the media world moves closer to an on-demand model, the introduction of a live and time-dependent service is somewhat perplexing. ↥︎

James Pinkstone found that his subscription to Apple Music and iCloud Music Library caused his local files to be replaced:

When I signed up for Apple Music, iTunes evaluated my massive collection of Mp3s and WAV files, scanned Apple’s database for what it considered matches, then removed the original files from my internal hard drive. REMOVED them. Deleted. If Apple Music saw a file it didn’t recognize—which came up often, since I’m a freelance composer and have many music files that I created myself—it would then download it to Apple’s database, delete it from my hard drive, and serve it back to me when I wanted to listen, just like it would with my other music files it had deleted.

Lucikly, he had a backup. But Serenity Caldwell at iMore disputes Pinkstone’s version of events:

In an ideal world, iCloud Music Library would work like Dropbox, or even iCloud Photo Library — whatever you upload is yours, it doesn’t get “matched” to anything, and as long as you re-download everything before you cancel your subscription, you’re fine. But DRM and downloading streaming tracks you don’t have ownership rights to mucks things up. iCloud Music Library is always going to be complicated, and people are going to make mistakes because of it. And if they don’t have backups, those mistakes might be costly.

Jim Dalrymple, writing in an “open letter” format:

I don’t mind a public beta of Apple Music that is being worked on, but don’t walk on that stage and tell me it’s a finished, working product if it isn’t.

The amount of trust and loyalty you’ll lose with another round of broken Apple Music will be mind boggling.

This is what happens when there’s a not-quite-consistent duplication of services and poorly written dialogs. Apple Music is a rental version of the iTunes Store, but its library is slightly different. iCloud Music Library and iTunes Match are effectively the same thing, except the former enables some Apple Music-specific features while the latter matches songs differently. All of these services can be combined in various ways, but it’s hard to see why this should be the case.

A simpler music offering from Apple would roll as much of this functionality into as few options as possible.

Sebastiaan de With used — and I do mean used — his Leica M for two and a half years before writing this review. Much like Craig Mod’s review of the Leica Q, this is a dangerous article because it will no doubt send you to your favourite camera store, tempting you to drop a huge amount of money you might not have on a camera you don’t need because, hey, it’s a Leica.

MetaFilter founder Matt Haughey was interviewed by There’s lots here to love, but this is important:

It’s weird that there’s a time where we envisioned MetaFilter would run like Reddit, up and down votes and we wouldn’t have to have a human component to it, but that would kill the personality and the humanity of the site. It would just be mob rule and it would have all the problems Reddit has: terrible people controlling the conversation. It’s just really hard to run a human-curated anything at scale. I think that’s why I never wanted to do anything at scale.

Many of Ole Begemann thoughts mirror my own, but I wanted to elaborate on one in particular:

I continue to not be a fan of voice interaction on any device. Siri is too slow and unreliable. The ability to dictate in text fields (added in tvOS 9.2) is welcome, but only semi-useful if you want to use it with multiple languages. Siri only understands one language at a time, and there is no convenient way to change the dictation language when you’re in a text field. I estimate my search queries are split about 50–50 between German and English, so it sucks either way.

I wasn’t aware of Siri’s language limitations, but so far, the Apple TV has been the best voice control and dictation experience for me. It seems more consistent and reliable than iOS or watchOS, and many of the baffling early-day omissions — lack of App Store or Music search, for instance — have been remedied.

Begemann’s annoyances with the slow system animations — particularly when resuming play from the screen saver, for which I have an open radar, #25008323 — and remote design are spot on, however.

Apple’s vision of the iPhone as a wallet is bold and ambitious, but it’s exactly the kind of thing that they’re good at tackling. There is, of course, no reliable predictor of guaranteed future success, but there are few companies who can do something like this.

And, of course, there’s the question of which company — if any — customers would trust with the contents of their wallet.

Dave Howell, as quoted by Michael Tsai:

9/14/14: Sold 500 edu copies of Air Display 2.

4/25/16: All 500 refunded.

No explanation, contact, or even customer name.

I’m guessing some school district returned its trial iPads and went Android but who knows.

Basically Apple let them use my app for free for two whole school years without even asking my permission.

A lack of decent, consistent communication is the root of so many of the issues plaguing the store, whether it’s apps being rejected for ridiculous and inconsistent reasons, or poor advice from the App Review team. Open, reliable communication between Apple and its developers has never been a strong suit for the company, but it ought to be a priority.

Another year begets another media blitz purportedly revealing the identity of Bitcoin’s pseudonymous creator Satoshi Nakamoto. But there’s a surprising twist this time: it’s the same guy as last time. The other weird thing? Instead of reporters chasing the creator down or someone hacking into his email account, Craig Wright — Nakamoto, allegedly — himself is doing the unmasking. And that’s raised a few eyebrows amongst those, as they say, familiar with the matter.

Wright’s “proof”, as demonstrated to the BBC and Economist, is that he cryptographically signed a Sartre passage with a private key known to belong to Nakamoto. However, as Dan Kaminsky explains, this proof isn’t proof at all:

Wright is pretending he has Satoshi’s signature on Sartre’s writing. That would mean he has the private key, and is likely to be Satoshi. What he actually has is Satoshi’s signature on parts of the public Blockchain, which of course means he doesn’t need the private key and he doesn’t need to be Satoshi. He just needs to make you think Satoshi signed something else besides the Blockchain — like Sartre. He doesn’t publish Sartre. He publishes 14% of one document. He then shows you a hash that’s supposed to summarize the entire document. This is a lie. It’s a hash extracted from the Blockchain itself.

Robert Graham of Errata Security summarizes it in a way that’s a little more digestible:

Craig Wright magically appears to have proven he knows Satoshi’s private-key, when in fact he’s copied the inputs/outputs and made us think we calculcated them.

There’s more, too. Adam Goucher found several problems with the code and mathematical explanation that Wright posted, many of which would have been obvious to undergraduates in a cryptography or security-related field. And then there’s the lingering question of why Wright — if he is Nakamoto — chose such a byzantine method of proving his identity. He could have just spent a coin from Nakamoto’s wallet instead while the press watched.

For his part, Wright says that’s what he’s going to do to. But I’m doubtful. This feels less like a way to reveal the identity of Nakamoto and more like a misguided but so-far successful marketing strategy for Wright’s own interests.

Or, who knows — perhaps aliens invented Bitcoin.