We almost had a smartphone in the 90s. Why did it fail?

What It Covers

General Magic, a Silicon Valley startup founded in 1991, built a functional smartphone prototype nearly two decades before the iPhone. Despite backing from Apple, AT&T, Sony, and Motorola, the device sold fewer than 3,000 units. Journalist David Epstein and Apple/iPod engineer Tony Fadell analyze why abundant resources destroyed the project.

Key Questions Answered

• •The Constraint Paradox: Excess resources kill creative focus. General Magic had funding from Sony, Motorola, AT&T, and others, yet shipped only one product after four years. Tony Fadell later built the iPod in under eight months at Apple, which was $500 million in debt, because financial pressure forced the team to prioritize ruthlessly and eliminate scope creep.
• •Define One Specific Customer Problem: General Magic built for a fictional customer called "Joe Sixpack" without identifying a concrete problem to solve. When Tony's mother user-tested the device, she couldn't articulate why she needed it. Before building anything, identify one real person with one specific unmet need — "1,000 songs in my pocket" is the model to follow.
• •Brooks's Law on Team Size: Adding people to a late project makes it later, not faster. General Magic kept hiring after delays mounted, compounding inefficiency. When building the iPod, Fadell kept the team lean and set a hard Christmas deadline because Sony was expected to release a competing product that same season, forcing disciplined execution over expansive staffing.
• •Borrow Before You Build: General Magic reinvented existing technologies from scratch, including infrared remote technology that had existed for years. Fadell's approach at Apple was to source existing components globally — processors, batteries, screens, hard drives — then assemble and differentiate through software and design, cutting development time and cost while reducing unnecessary engineering complexity.
• •Additive Bias Undermines Prioritization: Humans are cognitively wired to add rather than subtract, a bias rooted in evolutionary scarcity. General Magic's calendar function ballooned from four lines of code covering 1904–2096 to spanning the Big Bang to the future. Counteract this bias by explicitly defining what a product will NOT do before development begins, creating a written exclusion list.