Nakamoto consensus (the combination of longest chain rule for consensus and PoW for sybil resistance) was a minor innovation in distributed computing, specifically state machine replication. That had previously been solved in the late 1990s in the permissioned setting with BFT-type protocols (such as PBFT).

What's new in crypto is permissionlessness which enables decentralisation [1]. That has

a) basically no benefits that cannot be achieved through (permissioned) distributed computing, except censorship resistance, or, as normal people call it, regulation dodging, and

b) it comes at enormous costs (in terms of inefficiency, cumbersomeness, etc.)

Crypto is systematically and inherently designed for and suitable for crime.

[1] You can claim that LCR also expanded the design space in the partially synchronous model from always consistent and eventually available to always available and eventually consistent, but again, not a huge benefit.