The more times and ways that players can play and replay the game, the more devtime-efficient they are. This means the most efficient situation is systemically supporting combinatoric results - subsequent player decisions apply multiplicatively, ultimately resulting in some A x B x C x D … and so on possible results. This kind of system-based gameplay is the core of games like Minecraft and Breath of the Wild - these materials and items have properties that interact with each other in established ways, so you can end up with multiplicative results from the established inputs.
The ultimate result of this kind of system-driven combinatoric gameplay is generally in one of two core gameplay paradigms:
Procedural content generation, i.e. the game systems themselves create new content through a bunch of randomized decisions resulting in novel game content for each playthrough. Continued dev work would be spent on adding new options for the randomized content generator to create even more interesting possible combinatoric results.
Player-facing content creation tools (i.e. a “maker” game) alongside a curation system to allow players to create content for and distribute said content to other players. Continued dev work would be adding new assets and new tools for players to create new and interesting content from those tools, as well as better ways of curating and distributing popular user-submitted game content.
To answer your second question, the inverse of this - the least devtime-efficient content - is game content that is primarily single-use. This usually means content like cinematics, plotted narrative, voiced lines, action set pieces, puzzles, quick time events, scripted boss fights, that sort of thing. They’re pretty fun the first time most players play them, but significantly less so by the fiftieth, and they’re very expensive to turn into a procedural thing that can be handled systemically.
Got a burning question you want answered?