The explanation acquired a sequel
The usual account begins with Microsoft engineer Raymond Chen’s 2012 post about Pinball’s disappearance from Windows Vista. During early work on 64-bit Windows, he encountered a version in which the ball passed through the objects it was supposed to hit. The result was a game whose basic collision behavior no longer worked. With a much larger body of Windows code to port and limited time to investigate, he described deciding to drop it.
That account also rejected a popular explanation involving legal trouble. It offered a specific engineering memory instead: a broken component, unfamiliar old code and competing demands on the people doing the port. It did not mean that every possible 64-bit build was incapable of running Pinball.
The machine makes a difference
In January 2022, Chen filled in an important part of the setting. The severe bug he remembered was on Alpha AXP hardware used before Itanium machines were available. The 64-bit Alpha version of Windows was never released. That distinction changes the chronology: this was an early development port, not a demonstration that the game could never work on later 64-bit Windows.
Working Pinball versions did appear in 64-bit XP and Vista previews. As The Register’s contemporary report explains, those builds complicate any simple line from Chen’s broken port to the final absence in Vista. They are evidence of a working game after the failure he encountered.
A guess is still a guess
Chen suggested possible ways it might have recovered: a change to the C runtime’s rounding behavior or a compiler fix could have removed the problem, allowing someone to test and restore the game. He explicitly called that reconstruction a guess. Nobody had told him it was working again.
The defensible conclusion is narrower than the familiar legend. Chen documented a real porting failure and an early decision to drop Pinball, then clarified the hardware and the limits of his knowledge. The later working builds mean that his memory does not, on its own, establish the complete reason for the final Vista omission. Keeping that gap visible makes the history more accurate than assigning the entire disappearance to one permanently unsolved bug.