Once we know something, it becomes hard to imagine what it feels like not to know it.
More quotesCustomer: “Shuffle is broken.”
Engineer: “It is random.”
Customer: “It played tracks one, two and three.”
Engineer: “Yes. Randomness allows that.”
Customer: “Then fix random.”
The customer did not buy a probability lecture. They bought variety. True randomness can repeat a song, create ugly streaks and reproduce part of an album’s order. Over enough plays, the distribution may behave beautifully. Human patience expires long before the law of large numbers clears its throat.
So iTunes made shuffle less random and the experience felt more random.
Purists may call that dishonest. In entertainment, the label describes an expected experience as much as a mathematical procedure. A random rota, lottery or selection process carries promises about equal odds and must not secretly massage outcomes for aesthetic comfort. A music player carries a weaker promise: surprise me without becoming irritating.
Design is full of these collisions. Technically instant work can look careless. Perfectly equal spacing can look uneven. A neutral default can feel like a recommendation. Correct implementation does not rescue the wrong interpretation.
The useful question is which truth the product owes: the internal rule, the experienced result, or both.
After the third consecutive track, the listener presses skip. That click is not a misunderstanding to be corrected by documentation. It is evidence about the experience they expected. The engineer can keep the distribution honest while limiting irritating streaks, then state what the feature does. The argument ends in the next listening session, not in the probability lecture.