He envisions a bunch of languages in the future, all of which are similar enough to use the exact same data model? Surely, if we really need different language

He envisions a bunch of languages in the future, all of which are similar enough to use the exact same data model? Surely, if we really need different languages, they will be different enough to have different data models, and an encoding of the data from one language will be required for interoperability. For example, a dynamic language can be encoded into a more static language by passing along extra meta-data. And BTW, this has performance benefits over his idea of making the underlying VM as dynamic as the lowest common denominator language. Good luck with that idea if you have IO in your language stack. I also disagree with the apparently widely held belief that we really need to have many different full languages in use. Such a state of affairs may be useful for competition, but the overhead of interoperability hurts productivity. It's analogous to the situation with Buffalo. Here, let me switch to Navajo to make the analogy, since it has better words for Buffalo. BIH TA-ZI-KWA-I-BE-KA-DI COH-NAI-ALI-BAHN-SI ALTH-AH-A-TEH TKAL-CLA-YI-YAH.