The company that almost built the iPhone died because it didn't believe it had any constraints
General Magic had the biggest industry alliance in history, more money than it could spend, and the best engineers alive — and sold 3,000 devices in six months. It lacked neither money nor talent. What it lacked was any sense that it had constraints.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
The iPhone was imagined in 1989
While at Apple's Advanced Technology Group, Peratt used a large red notebook to describe a future information and communication product: it should bring personal satisfaction like jewelry, have perceived value even when unused, offer the comfort of a touchstone, the feel of a shell, the magic of crystal. What he drew was a smooth rectangle, no buttons, only a touchscreen, able to serve as a phone, a messaging device, a portable TV and a game console, and able to book plane tickets. It was 1989: most cell phones were the size of a brick, Apple's new portable computer weighed 16 pounds, and the World Wide Web did not yet exist. He called the idea a pocket crystal, imagining people carrying it everywhere — "if your house caught fire, you'd think, first save the family, then grab my pocket crystal." He told Bill Atkinson, and Atkinson brought in Andy Hertzfeld, two of the core programmers behind the original Macintosh.
— Tim HarfordA company with no product can still make Wall Street line up
General Magic had no profits, no revenue, not even a product — only an idea — and it issued stock to ordinary retail investors. Goldman Sachs bankers shoved each other to grab allocations; one banker asked, "What's it going to take? Would it help if I took off my shirt?" and then actually did, standing half-naked in the conference room. The stock doubled immediately after listing. Each partner in the alliance had already put in several million dollars. As one vice president in the book told David Epstein, they had so much money they never had to worry about making payroll — what startup never has to worry about making payroll? That is the key mechanism: enough money that you never have to make a trade-off.
— Tim HarfordThe tech was so good it made a partner's department obsolete on the spot
Two young engineers flew to Japan to show a hardware prototype to Mitsubishi. The demo had barely started when the host suddenly called in a group of employees in work clothes, apologized, and asked them to start over — because this thing had no modem, it was all implemented in software. A groan came from the audience, and a manager named Kato slumped forward and put his head on the table. The host apologized again: "You just made Kato's department obsolete." This technology made answering machines, arcades, books, keyboards and mice obsolete. At the same time they built a precursor to USB, a new programming language called Telescript, called portable networking the cloud, put skeuomorphic icons on a touchscreen, and built a store where you could buy applications — back when nobody called it an app store.
— Tim HarfordThey built the future but not a product
The first device running the General Magic operating system was Sony's Magic Link in 1994. Behind the gorgeous graphics it was hard to use, shipping with a 200-page manual; it was the size of a hardcover book, its grayscale monochrome touchscreen only the size of a postcard, and it ran on old-fashioned AA batteries that drained alarmingly fast. No Wi-Fi, no Bluetooth — the body had only an old-style telephone jack, so sending messages and making calls both meant sitting at a desk and plugging into a phone line. It didn't even have a built-in phone. In today's purchasing power, about $1,600. Sony sold 3,000 units in the first six months, mostly to friends and family of General Magic employees. A few years later, Apple sold roughly 300,000 iPhones on day one. In 1995 Motorola's Envoy had wireless communication, but cost nearly twice as much as the Magic Link.
— Tim HarfordeBay's idea was dismissed by a colleague as the dumbest thing he'd ever heard
Pierre, a technical support engineer at General Magic, built a webpage on the side: list the things you have sitting unused at home, let other people bid on them. He called the site AuctionWeb, showed it to colleagues, and asked whether he could spin it out as a company, or whether General Magic would invest. One colleague's response was typical: that's the dumbest idea I've ever heard — you want strangers to trust each other on the internet? Good luck, we're not interested. This quiet man was Pierre Omidyar, and that website was eBay, which later made him a billionaire. It is also a miniature of General Magic missing the Web — people inside reminded management daily, and management always found a reason to ignore it.
— Tim HarfordMissing the Web was embarrassing, but it wasn't the cause of death
In the documentary, former employees talk about missing the World Wide Web with laughter and sighs mixed together, but Tim Harford argues this failure is a distraction. When the Magic Link launched in 1994, the Web was still a curiosity; even if it had been Web-compatible, that wouldn't have been enough to make someone spend the equivalent of $3,000 on a Motorola Envoy. The second explanation is "too early": by the time the iPhone launched in 2007, memory prices had fallen 1,000-fold since 1994, a typical microprocessor had 500 times the computing power, and there was Wi-Fi, 3G and a Web with real content. That technology gap is real, but one technology commentator in the documentary says General Magic could never have succeeded no matter what it did, and Tim says: that sounds plausible, but he's wrong.
— Tim HarfordThey had time to extend the calendar back to the start of astronomical time
Epstein's argument is that General Magic had no constraints at all: the best hardware and software engineers, excellent design and marketing, a charismatic prophet-like leader, plus Apple's investment, several million from each alliance partner, and retail investors' money. With both talent and time, why compromise — just go build the pocket crystal, even if early-1990s technology couldn't possibly build it. The consequence was that nobody was forced to figure out what they were actually doing: when they described the target market they said Joe Sixpack, meaning everyone. So the product got stuffed with an instrument panel, a big walking-lemon emoji, a coin-flip animation. The coin flip came from co-founder Andy Hertzfeld, a man who at that moment clearly felt he had all the time in the world. The calendar originally covered only 1904 to 2096, a few lines of code, and it was first changed to the year 1 AD, then to the start of astronomical time.
— Tim HarfordTony Fadell asked only two questions
Tony Fadell used to call General Magic every day begging for a job, and when the company collapsed he was heartbroken — but he learned something. At his next company he focused on just two things: given today's technology, what is new, exciting and possible? The answer was MP3s and ever-cheaper hard drives, and combining them could produce a digital music player that made Walkmans and CD players obsolete. The second question was what deadline to set himself. The answer: it's January 2001, and we ship before Christmas. The company he joined was Apple, and what shipped on time a few weeks before Christmas was the iPod — one thing done beautifully. That restraint paved the way for the next project, the pocket crystal that was actually built: the iPhone. His colleague Andy Rubin later founded Android, and others co-founded LinkedIn, ran Google Search, and led Apple's AI, Safari and Apple Watch.
— Tim HarfordIn their own words · checked verbatim
They say, my God, this guy's right. I can go to the roof of this building and if I jump, I will fly.
Tim Harford2:18
This is the dumbest idea I've ever heard. You're going to get strangers to trust each other on the internet? Lots of luck. We're not interested. Have fun.
Tim Harford20:00
Missing the web? Embarrassing. But not the reason General Magic failed.
Tim Harford22:03
We had too much money, one of the company's vice presidents told David Epstein. We never had to worry about payroll. What startup in the world never has to worry about payroll?
Tim Harford25:10
The coin flip, by the way, was programmed by General Magic's co-founder Andy Hertzfeld, one of the greatest software engineers ever born... and a man who at that moment clearly felt that he had all the time in the world.
Tim Harford29:19
General Magic's problem wasn't that they actually had no constraints. They did have constraints. Every person and every organisation does. Their problem was subtly different. They didn't think they had any constraints.
Tim Harford30:20
When Peratt told you that you could fly, you believed it. And he believed it too. But gravity is real. If you persuade yourself that it isn't, you're going to hit the ground hard.
Tim Harford36:40
Figures
| Weight of Apple's new portable computer at the time | 16 pounds | 4:28 |
| Each alliance member's investment | several million dollars | 8:36 |
| Page count of the Sony Magic Link manual | 200 pages | 16:53 |
| Sony Magic Link sales in first six months | 3,000 units | 17:55 |
| Magic Link price (in today's purchasing power) | about $1,600 | 17:55 |
| Increase in typical microprocessor computing power | 500-fold | 23:04 |
| Calendar function's original year range | 1904 to 2096 | 29:19 |
Glossary
- Pocket Crystal
- Peratt's 1989 vision of a buttonless touchscreen portable device, the prototype of the iPhone.
- Telescript
- A new programming language developed by General Magic to let portable devices communicate with each other over a network.
- skeuomorphic
- Icon design that represents functions with the appearance of real objects, such as tapping a folder to see files.
- Joe Sixpack
- American slang for the ordinary man on the street; General Magic used it to describe its target market.
How to listen
Founders and product managers interested in product trade-offs, startup financing, and how too many resources can be worse than too few; also suited to team leads running long projects who need to set constraints.
If you know 1990s tech history well, fast-forward the first 12 minutes of character and alliance setup.