Comment by pornel

I can believe that it’s just overconfidence and optimism gone too far.

The problems they claim to solve look deceptively easy on the surface. Something like escape analysis (required for automatic borrowing without GC or refcounting) has many easy cases, but also incredibly hard or literally unsolvable edge cases.

They may have been encouraged by progress on the easy cases, and assumed the rest is just a matter of a few bug fixes, rather than hitting the halting problem.