The intended way to support **unpacking is the _asdict() method:

    t(**a._asdict())

It doesn't really make sense to add direct support for **unpacking because named tuples are sequences and not mappings.  To support **unpacking, we would have to add a keys() method and modify __getitem__() to handle both integer indexing and string key lookup.  Once we have **a and a[k] and a.keys(), people would also want a.values() a.items() etc.  The world gets murky when both sequence and mapping behaviors become commingled.