[zer01]

Right now smart contracts are very crappy computer programs, but people are still building amazing things with them (like DAOs or decentralized exchanges). It’s hard to say what people will be building 20 years from now or what they’ll look like in the same way that the idea of end to end encrypted video calls through Signal was not even something people could dream up in a practical sense without the groundwork invention/innovation of things like public key cryptography, and a minimal level of ubiquitous computing power in the form of cell phones to make it usable and practical.

Quantum computing is going through this too right now where we have the ability to compute with a handful of qbits, but the implications on a years time horizon are profound.

You do bring up a valid point in that it’s unclear exactly how the “legacy software” story fits in here. Migrations to new blockchains and the blockchains themselves upgrading without contention (Tezos comes to mind with this as a core feature) seem like the natural response to this problem. Not unlike anything else though it’ll be hard to predict until it slaps us in the face lol.