Limited Technology Can Yield Better Products

Doom II, the game from 1994 is best in class, in my humble opinion, even when compared against its successors. The reasons I think so include:

1) improved technology over predecessor Wolfenstein 3D

  • multiple altitudes and staircases
  • heat seeking projectiles

2) creative innovations

  • monster infighting
  • no “points” or treasure for more realism

but the one that we can see only now, in retrospect, is the one I’m interested in:

3) simplicity forced by limited technology

  • true 3d is not possible because you cannot walk under a bridge
  • monster AIs are easy to understand and predictable given the whole game must fit on three floppy disks

It seems to me that the limited technology of the early 90s allowed a single person to have a deep understanding of the whole code base. The building blocks were mature, stable, and well documented. In this case, the Windows OS.

Technologists since then, by and large have done a bad job boxing things up. Docker is huge for this, but sometimes I feel like we need to do better when we publish open source. There is so much published half working code, that it’s become difficult to see the signal.

When building a system one person needs to be able to understand enough of it to make progress. Since Doom II came out, a single human’s available working memory has remained constant. But the amount that we need to hold in our brains has increased. This is a recipe for bad software.

So try to answer this question: what technology of today is nascent enough to be “fully” understood but ready to be improved to be hugely impactful?