Under schedule pressure that's certainly true, but subsequent products will often start by taking a skeleton of a previous one. In this way tech debt could live on for many additional games unless it's addressed.
Is this actually true? The engine lives for multiple games, and the tooling is often reused as-is, but do games start with existing projects, or do we add new technical debt through the cleansing baptism of copy-paste?
In 1982: all games were written from scratch. You might copy the good parts from a previous game, but there generally wasn't much: poor abstractions meant the game logic and UI was mixed together (in their defense, on the limited CPUs of the day you couldn't have made proper code separation work, and of course the limited CPUs also meant you couldn't create such complex systems where it mattered.)
As games entered the mid 90s games started to get releases, and game engines came out. However each release was treated like an all new project even if some code was reused. Games would customize the engine in ways that wouldn't apply to the next game. Bugs in the engine would be worked around by ensuring that part of the game could not be reached (thus resulting in games that you could never get the 100% complete stat as the engine knew you didn't go through some locked door without a key). Some of the core engine was reusable, but for the most part the engine and game were tested as one.
Sports games tended to be the first to figure this out: every year was a new release of essentially the same features, but different player stats. As such it is becomes obvious quickly the game engine maintenance is a cost, there were few new features (until the next console offered better graphics). Most other games were latter because part of the new game was engine changes to support it, so it was hard to find a stable core to the engine.
Gradually games have seen more and more value from separating the engine from the game. As this has happened engine changes are no longer for one game, but for the possible use by all future games. Part of this is game designers have figured out the right abstractions, things that previously would have looked unrelated and thus change different areas of the engine are now seen as variations on a theme if you squint just right - and we know how to squint
Of course there have always been, and probably always will be one-off games written like 1982. When your game is small this is okay.
Different game companies have reached this at different times. Infocom had mostly figured out their reusable engine in 1982.
