The comment about the last 20 years also makes me twitch a bit. That sort of thing always makes me wonder if it's supposed to be about the state of actual usabl

The comment about the last 20 years also makes me twitch a bit. That sort of thing always makes me wonder if it's supposed to be about the state of actual usable programming languages or of recent PL research. Either way, the first two things I think of when asked about recent advances are substructural types (maybe stretching the 20-year limit) and higher-order contracts, and these are both available in things intended as general-purpose languages (Rust and Racket).