On November 20, 1985, Microsoft shipped Windows 1.0. It had menus, icons, dialog boxes, a mouse, Paint, Notepad, Calculator and Reversi. It also lacked the behavior that now seems to define a desktop: application windows could not overlap.
Open another program and Windows divided the screen again. The programs sat edge to edge like anxious roommates marking out a studio apartment. You could resize the divisions or minimize a program into an icon strip at the bottom, but you could not toss one app casually over another.
At a glance, this looks like an early-system limitation, as though Microsoft had not yet discovered the third dimension. That explanation is too convenient.
Tandy Trower, the product manager who finally shipped Windows 1.0, wrote that the project began with overlapping windows. Scott McGregor, its former development lead and a Xerox PARC veteran, changed it to a tiled model before Trower arrived. When Trower took over in January 1985, Steve Ballmer told him to finish the badly delayed product rather than redefine it again.
The system could overlap things, too. Dialog boxes, menus and other pop-ups floated above tiled applications. The restriction applied to ordinary top-level program windows. This was interface policy, not a graphics engine discovering that one rectangle could sit in front of another.
Tiling solves a real problem
Overlapping windows give each program room to breathe, but they make the user the window manager. Someone has to drag, resize, uncover and rearrange them. A tiled desktop does that work automatically. Every open program remains visible, and every pixel belongs to something.
The cost arrives with the third or fourth program. Because nothing can hide behind anything else, each new window takes space from all the others. A calculator and a file manager can coexist nicely. Add a text editor, a clock and Reversi, and the desktop begins to resemble a spreadsheet that has opinions about leisure.
The argument was not settled by obvious usability law. In a 1986 experiment, Xerox researchers Sara Bly and Jarrett Rosenberg found tasks and users for which tiled windows produced faster performance. Their conclusion was not that tiling won, but that the tradeoff deserved closer study. Tiling saves arrangement work and keeps context visible; overlapping preserves size and lets position express priority.
Windows 2.0 allowed application windows to overlap in 1987. The familiar floating desktop won the default, but it never eliminated the other idea. Modern desktops keep rediscovering halves, quarters, snap layouts and dedicated tiling window managers because arranging rectangles by hand remains a surprisingly durable clerical job.
Windows 1.0 looks primitive now, but its tiled screen was not merely an unfinished version of the interface we eventually received. It was a coherent answer to a question desktop software still has not settled: should the computer protect every open window from being hidden, or should it let us make a useful mess?