I guess because the Go creators didn't study much (or cared about) the last decades of language research. Heck, the major language they were involved and associated with, C, wasn't that "current" even at the time when it appeared, just pragmatic in a "worse is better" way and readily available. They were every influencial systems guys, with C, Unix, Plan 9, etc, but not really big PL/Compiler guys.
Now, Swift is pragmatic too, and far from cutting edge, but still ways ahead of Go in that area, and its creator is a very strong player in the field of compilers (he also did LLVM).
Here's a link that shows how much hasn't changed:
Robe Pike seems to like to lament the fact that systems research is irrelevant. Well, PL research sure seems to be irrelevant - either something he is largely unaware of, or doesn't feel he needs - to him.
Except for its concurrency model (and it's an open question whether having a concurrency model built right into a language is even a good thing; it's certainly not a no-caveats win), Go is really, really conservative in design. Swift, by contrast, has adopted concepts left right and centre; there's nothing particularly novel, but it's certainly far more adventurous.
Interesting idea. Can you post a link that explains why that is?