Super Nintendo games were the flavor of the decade when I was younger, and there’s no better example of building incredible things within comparably meager constraints. Developers on SNES titles were limited to, among other things:
16-bit color.
8 channel stereo output.
Cartridges with storage capacities measured in megabits, not megabytes.
Limited 3D rendering capabilities on select titles which embedded a special chip in the cartridge.
Despite these constraints, game developers cranked out incredible and memorable titles that will endure beyond our lifetimes. Yet, the constraints SNES developers faced were static. You had a single platform with a single set of capabilities. If you could stay within those capabilities and maximize their potential, your game could be played—and adored—by anyone with an SNES console.
PC games, on the other hand, had to be developed within a more flexible set of constraints. I remember one of my first PC games had its range of system requirements displayed on the side of the box:
Have at least a 386 processor—but Pentium is preferred.
Ad Lib or PC speaker supported—but Sound Blaster is best.
Show up to the party with at least 4 megabytes of RAM—but more is better.
Apparently architecture does the same job as set design, "it creates units of environment, atmosphere, or events"—whatever you wish to call them—but with more weight, carrying more material, slower. This is why it can raise the curtain more times and repeat "there it is again" for longer. Perhaps for this and other reasons there are periods in the history of architecture in which stage design or the folly, for example, have been an effective field of experimentation for serious architects.