As the software and hardware teams scrambled to make a working phone, the negotiations with the carriers—first with Verizon, and then with Cingular (soon to become AT&T)—had been going on in the background. They concluded in 2006, with AT&T winning out, albeit with some important concessions. “I was involved to the extent that what I wanted to argue for was a transparent data pipe,” Williamson says. “Which was something the carriers had never given anybody. Up until that point, WAP was predominant.”
Carriers were favoring the conditions of the network over the device, trying to ensure speed over quality. “The carriers used to actively filter content, so they would do things like, if you were trying to show an image, they would transcode the image to lower the resolution so it’d be smaller so it’d be a little faster to the device,” Williamson says. “So we went through a lot of negotiations with AT&T to get them to agree to a clear pipe. And now it’s become de facto in all contracts.” They also had to negotiate with AT&T to get persistent connections. “Without a persistent connection, you can’t do things like notifications, and things like iMessage become a lot more difficult. They said, ‘No, we can’t do persistent connections! We have millions of devices! No, no, we can’t do that!’”
They did that. “That went into the contract too. We wanted to bring them out of the mentality where, you know, they were in, which was, ‘We want you to pay for ringtones, pay for text messages,’ and into this reality of ‘It’s just a computer that needs an IP connection, and we want that and everything you can get with an IP connection.’ So that, that’s huge, in terms of enabling the device to actually be capable as a modern smartphone.”
It’s hard to overstate how crucial a development this was. If you can remember using pre-iPhone cells for anything besides making calls, you also remember cascading phone bills filled with text-message charges and ringtone and game-download fees. AT&T was forward-thinking, but they were also very concerned about their business model, Williamson says. “In fact we eviscerated them. Nobody wants to pay for text messages or ringtones anymore. But they also did really well as the exclusive carrier for the iPhone. So it was win-win.”
Meanwhile, Jobs’s fears of the carriers sending them big books filled with specifications had come to pass. Williamson remembers interminable meetings with AT&T and their technical folks about issues. “They would come with these specification books, and they were like, ‘We have to do it this way, we have to support that.’ And we were like, ‘No. No. No.’ And we triumphed in the end; it was just hard to get there.”
For an idea of how central the negotiations were to the project, consider that Apple hired a project manager specifically to oversee carrier relations. “At one point, his team was the same size as the software team,” Williamson says.
Hardwired for Touching
And then, of course, there was the hardware. Tony Fadell started hiring hands from around the company and, because the extreme-secrecy clause applied mostly to the user-interface and the industrial-design teams, hiring new engineers and third-party suppliers.
“We had to get all kinds of experts involved,” he says. “Third-party suppliers to help. We had to basically make a touchscreen company.” Apple hired dozens of people to execute the multitouch hardware alone. “The team itself was forty, fifty people just to do touch,” Fadell says. The touch sensors they needed to manufacture were not widely available yet. TPK, the small Taiwanese firm they found to mass-manufacture them, would boom into a multibillion-dollar company, largely on the strength of that one contract. And that was just touch—they were going to need Wi-Fi modules, multiple sensors, a tailor-made CPU, a suitable screen, and more. The list was exhausting.
“Any one of those was very difficult,” Fadell says. “All of it together was a moonshot. It was like the Apollo project.”
Thanks to the ENRI team and the FingerWorks crew that Apple acquired in 2005, it had the know-how. “We had the basic science. It was about having the right technologists for chips. The right technology about manufacturing.… The question was, could we scale it and make it work with all the different environments,” Fadell says. For instance: “We had a real problem with sweat; that would make it fail. And we switched from plastic to glass at the very last minute, which was a curveball.”
Because they were trying to run Mac-caliber software on a tiny device, the hardware constraints were considerable. “Everything had to be really well fucking optimized,” Grignon says. “So we built our own chip.”
And Grignon, who was the senior engineer in charge of radios, had something of an outsize job. “Introducing radio into a handheld device was something we had never really done,” he says. They had experimented with the iPod phone, but they’d never attempted to scale into something that would be ready for the mass market. “We had to, because of the enclosure materials, we had to engineer our own antennas, which are its own set of art and magic.” It would take an enormous effort to build and test those antennas to make sure that they’d work.
Put it all together, Grignon says, and you’ve got a recipe for madness: “At the fundamental hardware layer, everything is new.”
A brand-new CPU running a brand-new operating system running brand-new apps interfacing with brand-new hardware. “Imagine you’re a developer or a tester, and you’d have a crash,” Grignon says. “‘Oh, shit, the app crashed. Why’d the app crash?’ Well, it could be at any layer in that entire stack, all the way down to the silicon. I mean, think about that. Imagine an entire piece of silicon that can shit the bed, because it’s new. We would have actual CPU bugs, or compiler bugs, because we were building the operating system for a different instruction set. Or we could have an actual legit bug in the coding, a logic error in the app. It was just a fucking nightmare.”
Key Us In
For a moment, there was a chance that the iPhone would kill QWERTY.
“The radical idea was that we had no physical keyboard,” Williamson says. “In hindsight, that was obviously the right thing to do. But then, we were all very concerned.” BlackBerry was finding success with its hard-button keyboard. And fear of another Newton-style input misfire was thick in the air. “We all had this fear of, you know, the Newton disaster,” Ganatra says.
In the 1990s, the Newton’s glitchy handwriting software had been so widely derided that an episode of The Simpsons took aim at the device. One bully tells the other: “Take a memo on your Newton: ‘Beat up Martin.’” The device can’t read the input and spits out, “Eat up Martha.” So “Eat up Martha” became a cautionary mantra among the engineers, repeated often in the Purple Dorm.
To ensure that people using the device would be able to interact with the objects on the screen accurately, the engineers determined a “minimum-region hit size,” or how small something could be while still reliably responding to clumsy fingers. “Anything that you touched on the screen had to be as big as that size, otherwise it was too hard to use, and you’d make a lot of mistakes,” Williamson says. But given the size of the phone’s screen they were working with, a QWERTY-style virtual keyboard was out of the question—the key buttons would be too small. “So we had this big conundrum. In fact, the early prototypes we had were terrible.”
It kept triggering wrong key presses. “Initially it didn’t work that well,” Ording says. “So it was a good time to rethink how you enter text. Because QWERTY is based on old typewriters, and it’s a little strange. On the other hand, people know how to use it. So that’s why there were other experiments, lots of exploring there.”
QWERTY, the keyboard layout named for the order of letters from the top left over, was literally built for inefficiency. It was designed to keep nineteenth-century typists from hitting the keys too fast and jamming up early typing machines. It’s persisted for over a hundred years because of its familiarity—people who knew how to type on typewriters could easily transition to computer keyboards, and so they did. The layout has lingered on despite the suggestion of more efficient configurations, li
ke DVORAK, that have all but proven to increase typing speed. The prospect of a new, key-free touch surface opened the possibility of reimagining how we input text and offered a chance to break away from a centuries-old layout.
A good keyboard, of course, was make-or-break for a device that relied on text input for the vast majority of its functions. So the designers and engineers had to get creative.
“We kind of took a hiatus from general development,” Williamson says. “And we encouraged anyone who wanted to to write a keyboard. It was kind of a fun time; we’d all been super stressed out, and getting the freedom to do anything you wanted, knowing it wasn’t going to have to ship, was a good diversion. There must have been a couple weeks we did that, which doesn’t sound like a lot of time, but for this group it was.”
Some engineers proposed chord keyboards, which would divide the screen into a 3x3 grid, and users could select a letter by touching two of those regions. “We had bubble keyboards where you could click and slide,” Williamson says. “You’d click on the screen, and you’d get a pop-up with four letters, and then you’d slide onto the letter you wanted.”
New algorithms were tested and altogether new layouts were examined. Radical rethinkings of text input were floated.
“A lot of people were thinking, ‘We can do whatever,’” Ording recalls. “Things where you can swipe on the keys, or double tap, or there were a whole bunch of variations. There were different orders based on frequency of key use and letters.” New layouts that might take a while to learn but would ultimately prove more efficient. “We tried all kinds of stuff to come up with all kinds of variants to make keys appear bigger or have a multitap that you could use to cycle through the letters.”
“The chord keyboards were probably the most crazy,” Williamson says. “One of them was like a piano keyboard, and you could kind of play letters on the keyboard.” Another keyboard closely resembled Graffiti, the much-maligned input technology on the ill-fated Newton. “We had a Graffiti-esque keyboard that got shot down very early.”
The team put a website together to compile the keyboard designs. An engineer named Ken Kocienda “won” the contest and ended up leading the keyboard project.
By Williamson’s estimation, they developed around half a dozen alternative entry methods. They went so far as to design ways to introduce the brand-new keyboard layouts—which, naturally, would be totally alien to users—through simple learning games. Williamson says, “We went down this path of thinking we’ll ship the phone with this keyboard game. That will teach you how to use the keyboard. Some of the games were, like, you had to type the letters in a certain time frame, it counted down, or you could blast letters in a word—fun little games.” Jobs, however, was not amused.
“We showed Steve all of these things and he shot them all down. Steve wanted something that people could understand right away,” Williamson says. They stuck with the suboptimal key configuration for the same reason it had migrated to computers half a decade ago—familiarity. “When people pick up this phone in the store, it has to be something that’s instantly recognizable, that they can use immediately. And that’s why we stuck with the QWERTY keyboard, and we added a whole bunch of smarts in there.”
Those smarts would be crucial. “People thought that the keyboard we delivered wasn’t sophisticated, but in reality it was super-sophisticated,” Williamson says. “Because the touch region of each key was smaller than the minimum hit size. We had to write a bunch of predictive algorithms technology to think about the words you could possibly be typing, artificially increase the hit area of the next few keys that would correspond to those words.”
When you hit a letter, the predictive software guesses what you’re going to hit next, and it enlarges your minimum-region hit size. So if you tap H, the hit size around I and E widens, rendering the keyboard more forgiving. With the help of those algorithms, the keyboard improved.
There’s another way to gauge the importance of the keyboard: it’s the only part of the entire project that was user-tested outside of the core team.
“We were so worried about the accuracy of the keyboard that we had everybody that was disclosed on the phone but not working on the software do usability testing,” Williamson says. The once-abandoned user-testing lab where the sapling of the iPhone was planted would finally serve its intended function.
The First Rule of Fight Club
The locked-down Purple Dorm was bustling. “There was never enough time, never enough people,” Lamiraux says. “People worked very, very hard.”
They were adding people to the team, but slowly, largely because of the secrecy demanded by Jobs and Forstall. The UI was their crown jewel. No one was allowed to see it unless they were on the P2 team or had received explicit approval from Jobs. Initially, that was only a small number of Purplers. “Less than fifteen or twenty people tops could see the UI, including the UI designers that were drawing the pixels,” Grignon says. As a member of the hardware crew, he was initially barred from seeing it too.
If Jobs wasn’t around, P2 couldn’t add any engineers, even from inside Apple, even if they wanted to. Management referred to those who’d been approved as “UI-disclosed.”
Inside the Purple Dorm, the engineers were too busy to think much about the security measures.
Outside, however, the obviously cordoned-off building broadcast a certain exclusionary vibe to the rest of the company. “It was literally locked down with a metal door, which is bizarre and unsettling,” one iPhone team member says. “Steve loved this stuff,” Grignon says. “He loved to set up division. But it was a big ‘fuck you’ to the people who couldn’t get in. Everyone knows who the rock stars are in a company, and when you start to see them all slowly get plucked out of your area and put in a big room behind glass doors that you don’t have access to, it feels bad.”
If engineers outside the Purple team were called in to debug technical issues, black cloth would be draped over any screen that might be displaying the user interface. “When you have engineers who are separated with a cloth between them debugging a problem, that’s dumb,” Grignon says. Taking that to its logical conclusion, the AT&T people could never see the phone either, and they didn’t, he says. “Never. They saw it onstage when we announced it along with everybody else.”
Then there was Steve Jobs.
“The fear of Steve was great,” Evan Doll tells me. “He was feared more than anything else by the rank and file, and even the middle-management layer at Apple. It was like a cult of personality. He would come walking down the hallway, and I would shut the fuck up,” Doll says. “People were more concerned about the downside of a Steve interaction than the potential benefit. I don’t want to make it seem like it was this gulag environment, but there was definitely a strong undercurrent of fear, paranoia, that definitely was a part of any interaction that a team might have with Steve, for sure.”
Jobs did look for ideas in parts of the company outside the Purple Dorm—he just didn’t tell anyone what they’d be working on. Abigail Brody was the creative director heading up the Pro Apps group. She was handed down a request to work on a mysterious project, something called “P2.”
“They told me, ‘You have to work on a multitouch project,’” Brody says. “And they gave me a multitouch prototype that was a little bit smaller than an iPad, but it was bigger than a conventional phone. It was a very crudely put together thing. If I remember correctly it was taped, so I could experience the gestures.” They wanted her to design a user interface and a health management app, among other things. But her team was told little else. “We had no clue whatsoever,” she says. “We were just told, there’s a list view and a main menu, we need a gallery and this and this—it was pretty vague. The only thing they did not mention was a phone.”
Meanwhile, third-party suppliers tapped to work on various iPhone components would be given false schematics so they’d think the project was just another iteration of the iPod. Member of the iPhone team would pose
as representatives from other companies when meeting with vendors to avoid starting rumors. And everyone had to sign strict nondisclosure agreements that stipulated they could be fired if they leaked information about the phone.
“That whole experience was, like, you’re a ninja, you don’t exist,” one iPhone designer says. “It was weird, samurai-type shit.”
Sometimes, new recruits had to sign a preliminary NDA first, agreeing that they would never discuss the existence of the next NDA they were about to sign, in case they didn’t want to sign that one. “It’s always weird if you have to try to be all secretive about things,” Ording says. In the beginning, he was told to keep the project secret even from the rest of his own HI team.
The closed doors were demoralizing some employees and agitating others. Especially those on the iPod team who were, essentially, tasked with building the hardware for a device whose software they weren’t allowed to see. They had to make a fake operating system so people could actually test it out.
“That’s when we invented skankphone. That was Apple’s kind of exclusivity that was a mix of paranoia and politics at its worst,” Grignon says. “Skankphone was just like a clown vomited all over the screen. It was the worst-looking dialer, it texted, it had all the functionality, but that’s what the quality assurance people could use. It still was built on iOS at its core, but none of the actual UI widgetry was there. It allowed the AT&T people to test it, our own QA people, but, you know, eighty percent of the people on the program couldn’t see the actual UI that it was going to ship with.”
The One Device Page 36