November 3, 2015

Inside Apple's perfectionism machine, adnotated

Filed under: Uncategorized — @ 12:00 a.m.
Inside Apple's perfectionism machine, adnotated

Original article may be found here. Images have been stripped.

CUPERTINO, California — In retrospect, it was easy to miss — a bit of combined technology never really seen before in a laptop. Everyone missed it, even those who tore down the ultra-portable MacBook, even those who looked right at it1.

The little strip of black along the two back edges of the MacBook’s twin speakers could easily have been mistaken for a bit of shielding or a vibration dampener. Except, that's not what it is.2

Some like to call it the "Speaktenna."3 The black strip along the back edge of the MacBook speakers is a never-before-tried combination of speakers and antennas for Wi-Fi and Bluetooth. It's a fairly ingenious bit of space-saving technology that teardown artists ignored4.

But that's not really the story. It's about what "speaktenna" represents: the Apple way of conceiving, designing and building its products.

The Speaktenna

This MacBook speaker module has a thin strip of black along the outer edge, which is actually the system’s antenna. The wiring for it, is the silver strip of wire that runs in a channel on top of the gold speaker chassis.

Phil Schiller, Apple's longtime senior vice president of worldwide marketing5, is waiting for me when I arrive at Building 3 at the company's Infinite Loop headquarters in Cupertino, California on an unusually warm fall day. He's wearing his trademark gray button-down shirt and, incongruously, a pair of glasses that he quickly removes.

Schiller doesn't give a lot of interviews. A quick Google search tends to bring up quotes from keynotes and one lengthy chat with Daring Fireball's John Gruber.

Schiller is, for most people, a cipher: A highly knowledgeable Apple spokesperson who can speak as passionately about its products as his late boss, Steve Jobs. In fact, to listen to Schiller speak is to hear echoes of the visionary leader who died in 20116. It occurs to me more than once that, perhaps, many of turns of phrase we often attributed to Jobs may have actually started with Schiller7.

Schiller, who joined Apple in 1987 as the company was launching the Mac SE and Mac II (and, as Jobs did, left the company only to return in 1997), has seen a lot of change in the last two decades. He described those changes in a somewhat Jobsian way: "A couple of things that I observe are different and, I think, incredibly great compared to the past."8

"From the beginning, the Mac has been about Apple taking responsibility for the whole thing: hardware, software, how applications can work and, increasingly, Internet services9. But that means something different today than it did 20 years ago," Schiller said.

"Today, those teams are not only integrated and designing something together, they’re actually thinking of features that could only exist because of that integration and solving problems that could only be solved because of that unique advantage."10


It's a recurrent theme in my conversations with Schiller and John Ternus, vice president of Mac and iPad engineering. As they see it, much of the innovation — certainly something like the speaktenna — inside the 2-pound, 0.5-inch-thick laptop would not have been possible without disparate teams working together.

In the case of the speaktenna, Apple engineers did everything in their power to fit the maximum amount of technology possible into the tiny anodized aluminum chassis. This included creating new battery chemistry and forms to support a terraced battery design that marries perfectly with matching cutouts in the chassis. There's even a deeper level of terracing cutouts in the body that aren't for more battery power, but to cut down on the overall system weight.

"We realized we could not create a great antennae and a great speaker because we'd be compromising," Ternus said. "Both of those elements need space. Antenna elements are small, but they need cavity; they need space to resonate."

MacBook casing

The MacBook’s anodized aluminum case interior is a bas-relief map of the battery topology. There is, though, a little more extra space removed from the chassis to save on weight.

The answer was to make the speaker and antenna teams collaborate to create something new.

"We absolutely married them together to make this happen," said Ternus, who laid out before me a disemboweled MacBook. I saw what looked like a regular pair of gold-plated speaker housings that sit at the back of the MacBook, near the hinge. Along both rear speaker edges was a millimeter or so of black material, which turned out to be the antennas.

"We ended up with a group of antenna engineers who know more about speakers than any other antenna engineers and a group of speaker engineers who knew more about antenna design than just about anyone else in the world," Ternus said with a laugh.

That level of collaboration is part of the company's DNA today. But this wasn't always the case.

"Before Jony [Ive] and Steve came back, there was a relationship between the industrial design team and the engineering teams, but not an integrated one," Schiller recalled. "The industrial design team might create some models and designs and those may not reflect the product that was ultimately made or they may not result in any product at all."

It was, put simply, much less of a clear path from product conception to production. There was, instead, a sort of back and forth between groups, where one would come up with an idea, lob it over the wall to the other group — design or engineering — which would then throw back over its concepts.

This is in stark contrast to how things work today.

"One of the amazing things for me is how stable things stay from first models that we make to the products that we ship,” Ternus said.

Schiller recalled that turning point in development and design, which led to products like the reinvented iMacs of the turn of the century and categories new to Apple — like the iPod, iPhone and iPad.

"Ever since Steve came back [in 1997] and worked with Jony on redefining the entire process, the industrial design teams, the engineering teams are joined at the hip in the work they do," Schiller said. "They think up solutions to problems together as the disciplines are merged into a seamless process."11

Beyond collaboration, Apple relies heavily on institutional memory to build upon past achievements.12

"It is very deeply ingrained in how we work, and it's also something that is amazing to watch because it's a process that is building on itself each year," Schiller said. "These learnings all help to create the next product we’re making, regardless of what part of the company it comes from."

It's also evident in the MacBook parts laid out before me. The tiny logic board (aka the motherboard) fits in one of my hands. It's literally packed on both sides with chips of varying sizes and includes everything from memory to storage and even the display drivers. It's also deeply informed by everything Apple learned from building circuit boards for handheld technologies like the iPhone. As I examine it, Ternus tells me the board is 67 percent smaller than the one found in the 11-inch MacBook Air.

Creating something new

Apple may have mastered coming up with new solutions for tough technology problems, but it doesn't always have the answers or resources in house.

"There are times when we have to add resources to Apple, grow talent to do something that we weren't able to do before," Schiller said.

And then there are times when Apple has to make others change what they do to meet its expectations.

Apple has a habit of leaning on suppliers in ways few other companies would do13. Schiller said it's not uncommon for Apple to refine third-party part specification and to help suppliers deliver exactly what the company needs to achieve its original product vision.

Sometimes "we're asking for technologies and features they've never created before or brought to market, things that [our partners] wished they could do, but were never able to do," he said.

“Almost every new product has a new material, process or technology that we’ve never done before,” said Ternus, “and, in many cases, no one has ever done before.”14

Of course, Apple works with many suppliers because, ultimately, they do not build the individuals parts. Apple specs out what it needs and then finds someone who can build it for them. In the case of Corning, which supplied Gorilla Glass for the original iPhone, Corning had the glass technology, but, in a world filled with mostly plastic phones, cellphone manufacturers weren’t willing to buy and use it.

“We had a higher expectation of what the experience should be and we had a partner who was amazing at delivering on that,” said Schiller, offering rare, third-party-vendor-specific praise.

Of course, Apple’s relationship with these partners is more complicated than that. Direct questions about who supplies Apple’s signature glass screens and, sometimes, backs, has often been met with silence or obfuscation. So it came as a surprise when, at the unveiling of the iPhone 6S and 6S Plus, Apple casually shared that they both featured Corning Glass (not necessarily Corning Gorilla glass).15

Why can't Apple consistently share where it gets its parts?

In part, it's because Apple doesn't always source parts from a single supplier, Schiller said. He added that those partnerships can change, meaning the parts change.

Take a look the recent news about the two simultaneous A9 chip suppliers (Samsung and TSMC) and possible battery performance differences — a claim Apple quickly disputed. In general, Apple simply doesn’t want to constantly explain the potential minuscule differences between part suppliers or when they choose a completely different supplier.16

There is a larger reason, though.

"The most common scenario is simply that what we got from a supplier basically has been created so uniquely for Apple that implying it's an off-the-shelf part like others may get would be really misleading," Schiller said. "So it's best not to even talk about the source because that implies things that aren't true."

Ternus said it was, for the MacBook, a pattern repeated again and again. The full-sized keyboard that essentially defined the overall size of the MacBook could not simply be sourced from a company adept at building laptop keyboard modules. Apple found a keyboard supplier, and then redesigned the keyboard, including the mechanisms under each key, from the ground up.

Staring in the mirror

Inside the Apple bubble, a giant campus with more than 10,000 employees, it's easy to lose introspection. Yet, when I ask Schiller if Apple does everything well, his answer surprises me.

"No, of course not, of course not," he said. "And we don’t want to sound like we’re perfect. We never are, we always have to get better and always have to listen to where we’re not doing well."17

Apple, Schiller said, is also always up for re-examining preconceived notions, whether that means selling a large-screen iPad with an optional stylus (the Apple Pencil) or creating something entirely new.

“When we look at creating each new product, one of the great things about the process is the product teams are free to re-imagine a product any way that would make it better. They’re not tied to a schedule, to force it into a specific time window, not tied to past definitions of the product,” said Schiller.

Radically different products put pressure on Apple and its suppliers who are sometime asked to change their factories to support Apples plans. I wondered if anyone within Apple ever vetoed one of these crazy ideas.

Ternus, though, told me “no” isn’t a very popular term at Apple. “There’s ‘No’s’ in some way,” he explained, adding, “There’s ‘No’s’ about what we do, but in terms of how we do it, going and making the best product there really aren’t a lot of ‘No’s.’ That’s one of the great joys of working here. [For] something really compelling. We can afford to make it happen.”

Building the new certainly gets you noticed. We were all surprised when Apple unveiled the new MacBook, a laptop shorn of the "Air" moniker as well as all its ports save one: USB-C.18

That kind of risk-taking can appear both bold and dangerous. It's a delicate balance not lost on Schiller, who said Apple puts a lot of thought into meeting expectations while advancing the state of the art.

"We found that the best thing we could do is push boundaries beyond what's expected, beyond what's comfortable sometimes," he said.

Schiller contends that people expect a little bit of daring when it comes to the MacBook. "It's a little counterintuitive to people, that doing what people don't expect, ends up being what people do expect. But that is true of Mac."

The spaceship approaches

Radical innovation, though, isn't possible without the level of intense collaboration Apple currently enjoys, according to Schiller and Ternus.

As I walked through the sprawling Cupertino campus with Schiller, noting how Apple employees hurried past, lost in their own product conundrum thoughts, perhaps not even noticing Schiller, I marveled at the sheer number of people. During the lunch hour, employees jockey for table space in the lunchroom. That room faces a large quad into which all of Infinite Loop's building faces.

As a result, Apple employees are often spotted walking between buildings, passing by each other, and perhaps, stopping to chat. To some extent, the design may be a reflection of Jobs' desire for serendipitous interaction (he reportedly helped design Pixar's headquarters for just that effect).

However, if you were to go to the third floor of Building 3 or 4 and stare east you would see, rising in the distance, Apple's new "Spaceship" campus, which may eventually house almost all the employees from One Infinite Loop.

The 2.8-million-square-foot building is a circle with a giant quad in the center. It's hard to imagine how employees will get around it, let alone collaborate as they've been doing for decades at One Infinite Loop.19

What, I wondered, would happen if one team was seated on one side of the circle and another all the way on the other side? Would collaboration suffer?

"Quite the opposite," Schiller said. "The design of the new campus has been all about fostering collaboration."

Apple, Schiller said, is well aware of the collaborative platform that the current building offers. The open quad means people can always grab a chair for impromptu meetings, though Schiller says its standalone buildings aren't actually as conducive to continual collaboration.

"Whereas, with the new campus, everything about it is designed for that. From the fact that, on the ring, the internal and external surface of the ring are the hallways, and they completely traverse the space," he said. "So you can walk through the entire space, both on the inside and outside perimeter and go from section to section."

"It has, obviously, the biggest, open quad-like space you can imagine and it's being designed to encourage us all to mingle and travel through it with pathways and even running trails through the campus. It has very large, open stairways and spaces vertically through the place both for its airflow and light, but also it creates a floor between the sections. We're creating an environment with large, open spaces between everyone’s seating areas to foster much more work together in communal space," Schiller said.

By this time, Schiller had gotten pretty excited. He obviously spends a lot of time thinking about Apple's new headquarters.

However, he caught himself and reminded me that he's not the architecture expert.

"But, absolutely, this is going to be the most incredible collaborative space that's been created."

Surely a phrase that Jobs would be proud of.

In-house collaboration and outside partnership on new designs and technologies only takes Apple partway to an obvious goal.

The company does not exist in a vacuum. It's part of a vast, competitive landscape where Apple is far from the leader in desktop and laptop PCs. But it's bucking the current downward trend where PC market leaders are slipping quarter over quarter.

Apple just reported a record 5.7 million Macs sold in the last quarter; over the past 12 months, it's sold more than 20 million Macs — more than any other year. Late last year, IDC reported Apple's U.S. PC market share had grown to more than 13 percent.

Schiller, however, told me that estimate was conservative.

"There are endless numbers of stats out there, you can quote any of them to tell any story, but the number that closest reflects what we all experience and see as we travel around is actually a U.S. market share approaching 25 percent," he said. "That closer reflects what we see when we go to the airport, when we go to the coffee shop, when we go to schools. You increasingly are seeing more Macs than PCs."

The design of Apple's campus encourages impromptu meetings, but its new "spaceship" facility will have collaboration in mind throughout.

Schiller was referring to the perception-versus-reality problem that dogs Apple.

Even as Mac sales numbers grow, Apple is consistently reported well behind Lenovo and HP, which, as of late 2014, had between 18 and 20 percent worldwide market share, respectively, compared to Apple’s 6 percent global numbers.20

Standing in a startup office, at a college lecture or on an airplane, it may look as if every other person (or more) is using a Mac. So why don’t the numbers match the optics?21

"I think what offsets that is maybe IT purchases or system integration projects or other things," Schiller said. "Why you don't see over 50 percent, because it sure feels like it in the real world. I feel like more people like Macs than other things and they’re not equal."

Schiller’s perception, though, is accurate. NPD analyst Stephen Baker confirmed those market-share numbers, which represent retail and consumer sales (including those in Apple Stores). It does not include sales to enterprise customers, through resellers and large OEMs. In the NPD data, which tracks sales and not shipments, the company is still behind HP, but Apple has been "slowly gaining share on Windows and the Windows brands," according to Baker.

For Schiller and Apple, that trajectory matters.

"The trajectory has been a market where the majority of PC vendors are not doing well; they're shrinking and we're growing," Schiller said. "That's reflected in an increasing market share and the number I quoted is the highest we’ve had since the start of the Mac. Our highest market share."

Apple, however, is growing at a time of sea change in the PC industry, where devotion to a piece of hardware may be less important than connection to files, content and services. As
the cloud looms larger, will the hardware we use still matter? Schiller rejects this notion.

"No. 1, the importance and value of great hardware has not diminished in any way," he said. "Across the board, our goal is to make the best in the categories we choose to compete in.
It's what we're doing and it's reflected in customers choosing our products over anyone else's. So I do think people are showing with their choice that they do value quality and beauty of the hardware and that is not diminishing."22

"I have never heard anyone say, ‘Because I like to keep my stuff in the cloud, I will take a cheap piece of hardware and I want it to be ugly.' All things being equal, of course,
nobody wants that," Schiller said.

He also rejects the idea that there's a growing market for hybrids, or, to be more specific, laptops with touch screens that also happen to be tablets.23

"There certainly are more offerings today, more people trying to create a market. But based on all the data that I've been able to see, it is still incredibly small and niche and may
not be growing to anything significant. Time will tell," he told me.

Schiller didn’t mention Microsoft by name, but the implication was clear. He doesn’t believe products like the swiftly selling Surface Pro or new Surface Book quite represent a trend.

In particular, Schiller doesn't just believe that people are choosing products like the Mac because they appreciate the innovation and craftsmanship. He contends that they love them.

"It's important to us and something we're lucky to have and it’s never lost on us that we have customers who really deeply love these products."

Customers' adoration for their Macs is not a new phenomenon, according to Schiller.

"This is something that goes back to the first generations of the Macs that people felt an attachment and a bond, and cared about their Macs in a way that you would normally not
expect from inanimate objects," he said.24

"Truth be told, knowing that that can be the case, we certainly try to support and enhance that, not only when delivering amazing products, but identifying with them ourselves.”

What's love got to do with it?

If you know — or at least believe — that your customers have an affinity for your products that goes beyond the norm, you may do unusual things.

As we looked over the pieces of MacBook, examining the tiny logic board and usual speaktenna, Ternus encouraged me to lift the lid of the completed MacBook he'd placed before me.

I placed the edge of one finger on the base to hold it in place and started to lift the lid. Ternus told me to stop holding the bottom and just lift the top. I did as I was told and noted how the bottom did not move as opened the MacBook. Ternus said that it's no accident.25

It's also an indication of the care Apple put into tolerances on the ultraportable.

"Every single unit gets measured on line for force required to open it, and we actually adjust every single unit," Ternus said.26

In fact, Apple is apparently taking the time to custom-fit all sorts of pieces in the MacBook through a process it calls "binning." Since there can be minuscule variances that might
make, for instance, the Force Touch trackpad not a perfect fit for the body or the super-thin Retina display not exactly a match for the top of the case, Apple finds matching parts
from the production line. Even the thickness of the stainless steel Apple Logo, which replaced the backlit logo on previous MacBook models, can vary by a micron or so, meaning Apple
needs to find a top with the right cutout depth.27

Pretty much all of Apple's manufacturing partners use robots and laser-vision systems.

Apple's Apple store

Inside Apple’s brand new campus store. In addition to all the typical Apple products, the campus store is the only one where you can buy Apple swag like Apple logo T-Shirts.

The result is that every MacBook is, in a way, special and imperceptibly different. I joked that every MacBook is like a Cabbage Patch Kid. "Every one is unique," I said. Ternus
finished the thought: "all in an effort to make them the same."

It's an almost unprecedented attention to detail. And with each successive generation of Mac, Apple is getting better at it.

"You see a Mac today and its level of technology integration, the specifications for how tight the tolerances have to be and on and on are higher this year than they were a year ago,
and a year before that, and a year before that," Schiller said. "As you learn to do something better, you can then build on that. Next year you now assume you can do that and try to figure out where else you can push the boundaries."

Talking to Schiller and Ternus, it occurs to me that the concept of Mac fanboys and fangirls is inside-out. The so-called cult of Apple doesn't begin and end outside its headquarters and Chinese factories. It is a part of Apple itself, a sort of cult of perfectionism that has been driving product development for 20 years.

For those struggling to understand what Apple is up to, it might be best to imagine the Apple logo as a giant, rose gold-colored apple sculpture that's being polished beyond
perfection, to some sort of ideal, a level of quality that is so undeniable that no competitor dares forget it.

Before I left, Schiller walked me over to the brand-new on-campus Apple Store. Compared to other Apple Stores around the world, it's relatively small. But like most others, it was
bustling. Here I saw the 27-inc Mac with its redesigned keyboard and minimalist Magic Trackpad.

Schiller smiled as he touched the devices. He loves these products and fully expects you will, too.



It's always lost on people who take things at their surface impressions that they are that thing and that thing only. This guy is a psychiatrist, not an alcoholic. That guy is a CEO, not a father. That guy is a father, not a programmer. That woman is a mother, not a whore. That tire iron is for popping wheels off the bike and stripping tires effectively, not opening bottles.


One must actually understand how things are made in order to look at a novel thing and understand how its parts fit together to achieve the design goals. The best thing that you can do for the mechanically-inclined kid is take him out to the junkyard for an afternoon or three and wander around pointing out all of the things that he'll never see during the day-to-day business of life, hiding as they are underneath the branded wrappers that insulate Americans from chancely encountering the mechanisms that drive life.

This is one of the grand tragedies of starting a family in 2015: it's nigh impossible (I'll have to assemble the components myself, apparently) to get a chemistry set that can be used to blow things up, and similarly difficult to get something like the Amiga (this I cannot assemble from parts for my children, unfortunately) that the curious child can crack open the programmatic guts of and begin to learn how computers work.

Not that I'd wish a knowledge of how computers work on anyone, they'll be useless come the dawn of Stan's computronium fabric.


The author, probably, and he won't stick his name on the name because it's so patently stupid a name.


Ignored, or didn't have the knowledge to understand it for what it was? The "teardown artists" in question are probably iFixit (but remain unnamed likely to avoid pissing off the subject of the puff piece), and "artists" because if they were engineers they wouldn't be running a repair shop, they'd be working with the objects of their adoration! Fandom, it's been said, only ever goes in one direction: extraction from the fans, to the adored.


This piece was written by a "journalist" who came up with PCMag and now writes Apple puff pieces for Mashable. The job of these people is not to write hard-hitting investigative pieces, lambasting crimes of the powerful (its own bag of worms), but to take leads from PR folk and turn those into articles inflating the perceived market value of the companies who pay PR's salary. The journalist follows the SVP of Worldwide Marketing around learning about all sorts of things, but fails to tease out that the marketroid's job is to sell that which engineering and design have decided to produce.

"If you're reading it, it's for you" << I find more and more applications for this Ballas line every single day. Today, the messages "for you" are that a) the marketing people matter b) Mashable matters and c) Apple's customer's love for Apple products trumps all other concerns, like "does this machine do what I tell it to?

Jumping off from there, I recently discovered that MACOS 10.10.2 categorically refuses to sleep if a video is playing in Chrome. I haven't tested this in other browser because I can't be arsed to get this machine to behave the way it should, but isn't it funny that the thing purports to be an actual computer up until you ask it to do things like "lock after two minutes", to which it salutes and responds "sure, boss!", right up until you do something consumer-y like…play a video in a web browser, at which point it starts acting like a dumb butler: "oh I thought you might like to watch that video, so I left the front door agape."


George Bush was one of these people as well. An intimate of his machine once described his public behavior to me, the long and short of which is that the man is excellent at regurgitating that which had been said to him or that he'd heard, and to stitch together a pastiche that can pass for creative conversation, but largely unable to cook up his own output.


Like "one more thing", the emails from, and "Rip. Mix. Burn." could all have come from anyone in the organization. Apple is a corporation, Steve was a well-crafted illusion, and only the men who work the sausage plant day in and day out could ever tell us what's in the links. Not that they'd know how many fingers made it in, but that's immaterial.


Ah, the vacous and meaningless statement from PR, over which the 'journalists' swoon. Of course the HMFIC of marketing thinks that things have changed and that they've changed for the better. Twenty years ago Macintoshes didn't run on stolen Unix, and Apple's still around as a company. Things are different and wonderful, you see.


True, and disingenous. Truly disingenous. They absolutely are "taking responsibility" for the whole stack, but in the opposite direction of that Moldbuggian notion of the modern bureaucracy dissolving both authority and responsibility across the whole organization. Apple's working quite well to gather the mantle of authority around themselves while avoiding any responsibility at all for the work that they produce.

Nobody, and I mean nobody, is holding their toes to the fire on any front. The software quality is continually decreasing (their hilariously inadequate office stack, Pages and Numbers, completely rejects the notion of backwards compatibility, and corrupts files willy-nilly), the build quality is on a similarly constant decline (the current Macbook that I'm using is chipping along the right front edge, and my iPhone's screen doesn't sit flat in its housing, leading to the dreaded 'creak' every time I drag my finger across the screen). The development tools are similarly universally broken, with Apple shipping entirely non-functional Xcode updates on a regular basis, Swift not having an "either" type, and the idiotic "spaces" debacle in place of an end-user programmable window manager are further examples in this vein.


Sort of like when piston heads developed into convective heat transfer devices back in the day, reducing piston mass and cost for only a minimal strength drop but also radical cooling characteristic improvements. Cross-training and spreading of the institutional knowledge across the institution is clearly a thing that Apple came up with, and not a standard management practice in all long-lifed manufacturing operations.


I suspect that Jobs wasn't necessarily a brilliant manager so much as he was a driven tyrant (in the sense of tyranny being a prerequisite for quality) who was willing to cut through swathes of bureaucracy to play with new organizational models.

Communication friction between working groups is very, very expensive. In a healthy organization, the separate groups respect each other (on the surface) and strive to present quality, finished work to each other in order to preserve the group and individual reputations. This polishing of potential turds is very expensive, especially when the teams are iterating on things. Eliminating some of this friction is at the root of the "agile movement": the notion being that an "agile team" can avoid wasting lots of time and energy polishing features for presentation to the business or design stakeholders.

The process can go entirely off the rails when the design and product stakeholders treat the development teams as prototyping engines, and fail to involve us in high-level application design: "here are some designs for this screen and that screen and the other screen—please go and implement them." "Mnah, I changed my mind about pretty much everything here, and possibly even the whole flow through the application. Please rip everything from the past two weeks out and replace it with the following designs."

"Agile" (to the extent that it's actually a thing, and not a construction of frustrated nerds) really only works when the tech team is involved in designing applications and their user interfaces (I know, it's a sad world we live in wherein the user interfaces are written by programmers and cannot be diddled programatically by people using them. I however, live and breathe and eat in this world, and not some hypothetical alternative universe where Symbolics didn't come to depend on USG largesse and so survived the AI Winter). In sanely-designed "modern applications", the UI is itself a projection of the application state. For the programmers to put that projection together and codify with the paradigms of the targeted platform's user interface toolkit, first we must have a robust understanding of what the application is meant to do. Then and only then can we put together designs for marshalling application state through its lifecycle and projecting that state through the native UI widgetry to the user.

If you just throw designs over the wall in an ad-hoc fashion and ask us to write software around them, you are going to get software that was designed as a result of attempting to introspect what all the application does from the limited data contained in the mocks. Such software won't look terrifically expensive to develop at first, but due to the ad-hoc nature of its growth and evolution it will rapidly mount in the costs to maintain and extend.

I ascribe these dynamics in software development to (in at least some small part) a cultural refusal of programmers and business stakeholders to draft and rewrite. Refactoring can be done, sometimes, under some circumstances as bizfolk can be made to understand that specific subsets of application code are crazy and need redoing. Raising the spectre of rewriting the whole application is almost always a non-starter, and for good reason: no business team wants to see progress come to screaming halt while the development team rewrites the world. One solution I've pondered with my peers in the past is to let the team grinding on feature work continue to do so while the A-team takes a step back and re-designs the whole application. The costs of this approach can be astronomical (adding somewhere between a tenth and a half of the development costs to the project, depending on scale and velocity), and so are hard to justify outside of companies that have everyone on salary and not desperate to get a product to market and making money.

All of the above is why I can get a dope mobile app built for under 80K, and you'll have trouble getting a prototype for 100K. You don't have the wisdom to articulate your vision, and the strength of character to trust my team to execute. I, on the other hand, have a well-seasoned team who understand intimately how to write quality software, into whom I can install a product vision and get results rapidly.


Old hat, if you've been following the "institutional memory is the only way to survive as a long-term manufacturing concern" thread around here. It's also old hat from a purely biological perspective, as evolutionary design ratchets forward against the relentless pull of entropy. Multicellular organisms depend on the existince of unicellular organisms, and the randomly +ev specialization of individuals in those colonies.


This is such complete and utter hogwash, but I have a hard time faulting Ulanoff for buying the Apple party line on this, unfamiliar as he is with how the world works.

Every company with the power to change their suppliers' processes does so. Honda and Toyota, for example, take this so far as to negotiate profit margins with their suppliers. One manufacturing company I worked with spent several labor-months with a casting supplier improving the mold design and casting process in order to drive reject part numbers down, an investment that the casting supplier wasn't about to make on their own preferring instead to take the hit on both internal and delivered quality.

To digress (because I can't do footnotes in footnotes yet), the parallels and contrasts between software consulting dynamics and contract parts manufacture dynamics is really interesting, especially with regard to tooling and quality.

The typical contract part fabrication shop will quote you part costs given tolerances within their ability to machine. The shop knows that it can readily hit certain tolerances (± 0.005 inches, or what have you), and will happily quote within that range. For tighter tolerances, sales will have to talk to the operators about how much ongoing tweaking of machine parameters the operator's will need to engage in to hit the tighter tolerances, which machines the parts will have to be run on, etc. Cutting faces wear, and modern machines have some amount of prediction built in, but cutting lots of parts at the bottom of your normal tolerance envelope is going to be expensive in: keeping machines in true to hit the numbers, inspecting the parts, and accounting for rejected parts.

Software's reflection of the machine shop's tooling setup is…murky.

Deploying new iOS builds to Apple is a nightmare of error-prone manual steps, compounded by their constant tweaking of how iTunes Connect works and where the buttons are put. What is this new save button for? The (post-TF acquisition) semantics for shipping builds to testers is very similar to the old TestFlight routine, but complexificated with a myriad of new steps whose role and effects aren't immediately apparent. So all of that's a complete nightmare to automate the deployment of. Nevertheless, we have some tricks up our sleeve, in the same way that part fixture designers have a bunch of tricks up their sleeves for cutting reusable or multipart fixtures.

Deploying Android builds is much less complex, and so much more easily automatible that it's almost not worth spilling the ink to talk about—just configure Jenkins to crap compiled builds out and email them to everyone, or link them into your VCS system or something. Doesn't matter, as all of your mobile application deployment costs are going to be on iOS.

Single-page JS applications? We've got piles of tooling and knowledge of how to build quality into those as well (step one: never try to retrofit an ad-hoc collection of HTML, CSS and JS into a SPA—while it might work, you can easily burn 10x what you would have had you simply rewritten the thing with a sane design up-front. Plus, maintenance costs will be significantly lower).

Server provisioning and deployments—we've got that handled. Conveniently, none of our clients use bare metal servers, which gave me the freedom to write automation that neglects the entire server state mutation problem, with all of its concomittant rollback problems. Nevertheless, I chose to take a dependency on Docker so that my automation stack can make those assumptions: provisioning involves installing Docker and then building the relevant images (although Docker's surface area is far too great, and still reeks of a company attempting to ensnare a new generation of developers in a toolchain whose complexity will ensure that some fraction of them will need "professional" support).

All of our institutional experience and in-house tooling around modern mobile and web application development boils off a whole shitton of complexity and uncertainty in writing software for other people. While I can't make claims like the parts producers do, of ability to produce so many widgets within thus and such a variance in size on these critical dimensions, I sell on other fronts: extant tooling saves time and money, institutional knowledge of best practices across the various software stacks produces higher quality and more cheaply maintained code; and (if you can bring yourself to buy it) holistic design and development a la Apple upon Steve's return lets us move at vastly faster paces than organizations who segregate these "functional areas" from each other.


In this regard, at least, Apple is no different from any other startup building mission-critical systems atop MongoDB and NodeJS. What saves Apple from the startups' fates is that (like all manufacturing operations of any quality) they test their parts before building production lines around them, and the nature of production lines and software development pipelines are so wildly divergent as to defy simple comparison.

A production line will have umpteen CNC mills (for example), of by-and-large the same make so that the same fixtures can fit into all of them. The product that they make may be a complex one, requiring parts from all fifteen of the mills (and who knows how many other lathes, stamps and presses), all of the machines are interchangeable with regard to their ability to produce parts for inclusion in the final work product. A well-designed production line will have a clear flow of parts into assembly, and a clear flow of parts through production steps. Iterating on this business construct is far simpler than iterating on businesses built on software: the clean interfacing between steps in the process lends itself readily to root cause analysis, and the constrainedness of each process or assembly step lends itself similarly readily to incremental improvements, or even step eliminations when appropriate.

A business that runs on software (and frequently the software that drives businesses whose core operations are not even software) is more akin to bootstrapping that entire production line from the mills on up oneself. Except in the sense of the request/response HTTP cycle and analagous technical pipelines, software companies do not produce widgets in the same way that fabrication companies do. Instead, they build whole systems that…do something. Perhaps the system is an aircraft carrier that flies. Or a submarine stapled to a Warthog, because it was easier to do that than build a flying submarine from scratch. Or, one simply ends up with a flying submarine because that's what the kids who hacked on the first draft accidentally crufted together out of other open-source systems they found lying around.

Actually, I take it back. Apple's web software has always been complete shit, and their desktop and mobile software has trended in the same direction for years now. The parts of their business that are vulnerable to bad software exhibit all of the hallmarks of unstable systems: unpredictable processing times, random changing of UIs, inconsistent data across applications (I think that I've updated my name 3 times in iTunes without it taking, and iTunes Connect at least twice now).


Sure, either they "casually shared" the source or "reluctantly complied with their contractual obligations". How could we ever know? Well, we can make a guess by looking at what their puff pieces say and assuming the polar opposite.


This story is a total hoot that somehow slipped past my radar. Okay, "somehow slipped" implies that I a) give a shit about Mac rumors and b) have radar tuned for this shit, neither of which is true.

The hoot:

According to the company, battery life between two otherwise identical iPhone 6S or 6S Plus models may vary, but only by 2-3%.

2% might be the margin of error in mechanical and civil engineering, but that's only because the engineers in those trades use 2-3x safety ratios more than compensating for 2% errors in computation, even if those errors stack up across multiple elements and interfaces.

For industrial products, performance numbers matter. Furthermore, the methods by which benchmarks are constructed also matter. Yes yes, chips are designed to beat benchmarks, and this is old hat in computerland, but Apple's brazen "fuck you, you don't get to see the numbers", or more accurately, "nono, these metrics are the real metrics, pay no attention to the reproducible experiments" attitude is endlessly hilarious, and utterly unsurprising. They build the boxes, and you buy them. Oh, did you want to upgrade anything in that laptop? Get fucked, not even the hard drive is available for an upgrade—nominally to serve their gods of design but once again look for what they say and assume the opposite: the real reason those hard drives aren't replaceable is so that you fill them up with (what, a blockchain? You can't stick a blockchain into an Apple laptop and still use it for things, idiot!) your photos (NB: the iPhone shoots in RAW by default. Why? Clearly not for the quality of the photos (again, ask why and assume the opposite), but because by setting one bit in software, Apple can chew through your iPhone's storage, your iCloud storage, and your local storage as well), music, Dropbox's duplicated space, the 3 15GB backups of your iPhone and then come to the (reasonable, if you're the sort of moron that believes it when Schiller's stooges take the stage) conclusion that you need a new device with more storage!


Schiller then goes on to talk about anything except for things that Apple sucks at, like writing their own software, or cracking the whip on Chinese fabricators, or stealing Unix (okay, okay. Buying NeXT) and pretending like they've added value atop it. Ulanoff completely fails to examine this "oh yeah, we suck in some areas claim". Why? The guy's a "car reporter". No more, no less. If he says negative things about the companies that pay his salary, he gets fired and the publication for whom he works never gets another scoop/lead/exclusive.


Because a giant dongle hanging off the side of your laptop so that you can charge your laptop is "good design". Someone at Apple's praying that everyone forgot about FireWire. Perhaps Apple has even forgotten FireWire…


I gather that Ulanoff has never heard of the Panopticon.


This doesn't even begin to account for the number of USG-mandated Windows licenses in the world (good luck getting a Linux system through the FDA's approval process), and beyond that buys the fallacious argument that the important thing here is the hardware and not the operating system. Yes, Apple sells the only portable computers that even sort of work. No, that doesn't matter so long as you're splitting out Lenovo and HP, blessed resellers of Windows licenses and treating them as "hardware sellers".

Look, the point is complex. Apple sells hardware, and the operating system is ancillary. Everyone else sells Windows machines (if you're chosing between Lenovo and HP and plan to put some breed of Linux on it, you don't count in these numbers either), and the hardware is ancillary. Apple only exists and can compete with Microsoft because they bought NeXTSTEP (including all of the hard work underpinning a Unix, stolen from a generation of nameless open-source developers) and made a couple of well-placed and well-timed media buys.

I too once loved their operating systems and user interfaces, but as I've come to want as complete control over my computing hardware as is possible, I've also learned that the tyrannical quality Steve drove Apple towards naturally precludes ordinary mortals from tinkering in the guts of the machine.

It's actually rather impressive—his tyranny extends beyond the grave and reaches slowly, surely, inexorably for my throat. I hear that "root" may not even be "root" in upcoming Apple OS updates. I'm not surprised (I ain't even mad, bro!), but I'm still driving myself towards a Linux toolchain I can cut over to in a heartbeat. At the moment the setup is Emacs in X11 with urxvt, but I'm still hobbled by the complete absence of an X11-compatible browser.

Perhaps I'll install Masamune in a VM and see if I can run with an emulated system. Le sigh.


Because corporate IT departments want fungible workers, and if you're all using Apple machines the temptation to start using Linux on the backend is just too great. Once you start using Linux for your line-of-business applications, people will start using other open-source tools and before you know it you'll have to change your hiring standards for the IT department to include people with blue mohawks, and you'll buy yourself a whole mess of organizational politics that are honestly more easily avoided by hiring derps who can tinker with Microsoft systems. Worst case, you have to hire some really expensive consultants to do the hard stuff your endogenous teams can't handle. That's what all the money you saved on salaries is for, I suppose.

That said, if you don't want to give away industrial secrets to China and Russia, you gotta stop using Windows. Linux—it's the patriotic choice!


Some of us will settle for vaguely nixy portables that also Just Work (tm) (r).

NB: this is the exact same sales pitch used to push Louis Vuitton and other high end brands that target the aspirational 14%. "Show your class and discerning taste by buying this commodity product that's classy, elegant, and 2x as expensive as other things that also get the job done but not nearly as prettily nor as ego-strokingly."


Sometimes a bad idea is obvious to everyone. Sometimes technological convergence is impossible to see coming.


Douglas Adams has a hilarious piece on the topic that explains the relationship between the early Macintoshes and the users thereof: Frank the Vandal. It's a real pity the man appears to have never encountered the Symbolics machines. To quote:

Why, when I'm working in a document in one word processor, do I keep on finding that if I want to do something else to the document I have virtually to dismantle the document and ship it over to another word processor which has a feature I need that the first one doesn't? (Why don't I just use the second word processor? Well, because it doesn't have other features that the first one does, of course). Or, if I want to put a picture in it, why do I have to go off to another program entirely and do the picture there, and then go through all mind numbing palaver of discovering that for some reason the WP I'm using doesn't know how to handle graphics in that particular format, or claims that it does but then just goes all black and sulky or makes the machine go bing when I actually ask it to. In the end I have to paste all the various bits into PageMaker which then refuses to print for some reason. I know that MultiFinder has made all this a bit easier, but it's really just the equivalent of making Birmingham easier to get to, if you follow me.


Nor was the coefficient of static friction of the surface upon which the device was placed an accident.


We? Not Foxconn?


This is either insane or awesome, and I cannot tell which. I was trained to make parts that fit together and to perform stackup analysis of the tolerances on assemblies to identify potential intersection zones. Apple's fabrication partners appear to be taking the opposite approach, which is to select parts whose tolerance errors cancel each other out. Sounds great in theory, but I wonder about statistical pileup of some classes of parts from some subcontractors that are routinely of the wrong size in the same dimension, and cannot be used in assemblies. How, again, is an outsider to know? Nifty technique, though, and kudos if it works for them.

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URL


« The American consumer, in two photos --- Perceived vs. actual barriers to homeownership for young adults »