The program model is the Def’s to move; the user model almost never is — that’s Joel Spolsky’s rule from 2000. Ten years later, Apple ran the experiment in reverse, at the scale of an entire product line, and got the result Joel’s essay predicted.

Hold It Differently

The iPhone 4 launched on June 24, 2010, with its antenna built into the metal band running around the phone’s edge — an elegant piece of industrial design that also happened to short out when a bare hand bridged the gap between two antenna segments, dropping calls and tanking signal bars. Asked about it directly, Steve Jobs’s response was to defend the design and redirect the fix onto the user: “All phones have sensitive areas. Just avoid holding it that way.”

That’s a company choosing to move the user model instead of the program model — the exact inversion of Joel’s rule. The program model (an antenna that shorts out under a normal grip) was left in place. The fix on offer was behavioural: hold the phone differently than you already do, than every phone before it had trained you to.

The User Model Didn’t Move

It didn’t work, for precisely the reason Joel’s essay gives: a user model, once formed, is “remarkably hard” to change by explanation. Tens of millions of people already knew how to hold a phone. Consumer Reports pulled its recommendation. The story became national news within days, and no amount of Apple explaining the physics moved the grip habit of a single existing owner.

Three weeks later, Apple reversed course — not by writing a better explanation, but by moving the program model instead. A press conference on July 16 announced free bumper cases for every iPhone 4 owner, a rubber shim that physically prevented a bare hand from bridging the antenna gap. Jobs’s own words at that event were closer to Joel’s rule than his earlier ones had been: “We’re not perfect… but we want all our users to be happy.” The following year’s iPhone 4S shipped with a redesigned, split antenna — the program model actually changed at the hardware level, not just patched with an accessory — and Consumer Reports restored the phone to its recommended list.

The Grip Was Never the Bug

Apple spent three weeks trying to move fifty million grips before conceding the antenna needed to move instead. That’s the expensive way to relearn Joel’s rule: the user model is the side that’s already paid for — formed, sunk, distributed across everyone who already owns the product — and the program model is the only side still sitting on the Def’s desk, free to change. A company with fifty million users insisting they’re all holding it wrong is a company about to redesign the antenna anyway. The only question was ever how long the explaining phase lasted before the actual fix showed up.

This post is part of the Gutenberg/Semantic series. Related: The Program Model Is the Def Model on Joel Spolsky’s original argument, and Architecture Astronauts on Joel Spolsky elsewhere in this series.

Sources: Poor iPhone 4 Signal? You’re Holding It Wrong, Jobs Says, June 25, 2010; Looking Back at 2010’s Antennagate, PhoneArena.