Alan C. Kay wrote the paper in 1993, at Apple Computer, to document the ideas behind Smalltalk and personal computing. He argues that the system grew from research into interactive graphics, time sharing, languages, pointing devices and education. Its central design move was to represent computation as objects that hide state and behavior and communicate through messages.

Kay identifies early clues in the Burroughs 220 file system and the B5000. The 220 stored data alongside procedures and references to those procedures. The B5000 added segmented storage, byte-coded execution, protection and procedural interfaces. Work with the CDC 6600 and Gordon Moore’s 1965 prediction about improving chip density widened the context.

At the University of Utah in 1966, Dave Evans introduced Kay to Ivan Sutherland’s Sketchpad and the Simula language developed by Ole-Johan Dahl and Kristen Nygaard. Sketchpad used master drawings, instances and constraints, and included clipping and zooming. Simula created independent objects from descriptions and used processes to control them. Kay saw a shared pattern: complex systems could be built from components with much of the power of the whole.

From 1967 to 1969, Kay and Ed Cheadle developed the FLEX machine as a small personal computer. Its design drew on JOSS, EULER, Simula and byte-code execution, with a budget of 16K 16-bit words. FLEX explored object references, coroutines, persistent objects, generalized assignment and event-driven expressions called when. Many ideas were too ambitious for the hardware, but they fed into later Smalltalk designs.

Doug Engelbart’s NLS supplied ideas for interactive work, multiple panes, graphics, navigation and user-defined interaction. Kay also studied GRAIL at RAND and LOGO with Seymour Papert, Wally Feurzig and Cynthia Solomon. These encounters pushed his vision beyond a personal computer as a tool. He imagined a personal medium that children could use for programming and learning. The Dynabook concept combined a notebook-sized computer, a keyboard, a stylus, a high-quality display and wireless networking.

During 1969, Ned Irons’s IMP influenced Kay’s thinking about extensible syntax and direct interpretation. Carl Hewitt’s PLANNER, Pat Winston’s concept-formation work, Butler Lampson’s CAL-TSS and the LISP ideas of John McCarthy added further pieces. Kay developed the view that objects could act like servers, accepting messages and deciding how to respond.

Xerox established PARC in 1970. Kay worked on the KiddiKomp and proposed SLOGO, a compact system intended for a SONY tummy trinitron display. PARC recruited researchers from Berkeley Computer Corporation and Doug Engelbart’s group. The laboratory built MAXC, a microcoded PDP-10 emulator using integrated-chip memory, and developed the environment in which Smalltalk-71 and later versions could mature.

The paper follows the evolution through Smalltalk-72, Smalltalk-74, Smalltalk-76 and Smalltalk-80. It covers overlapping windows, byte codes, graphical editing, animation, music, inheritance, storage management and the role of children as users. Kay credits Dan Ingalls as the central implementer and designer during much of the project, with Adele Goldberg and the rest of the Xerox PARC Learning Research Group helping turn the ideas into working systems. The history’s punchline is simple: the famous object model arrived after an unusually crowded queue of predecessors.