You want me to compete with a theory that made up an imaginary language?
Also, the notion that the PIE reconstructions describe a "language" in the sense that we usually understand it (a vernacular tongue of a group of people in a single time and place) is a false one: even though the reconstructions with very high likelihood represent individual IE language features in an old form thereof, it would be a folly to try to slap them together and say "this is a language that someone spoke", for the same reason you can't reconstruct individual genes from the tree of mammals, slap them together and say "this was the first mammal". It doesn't work like that. What you get is a constrained probability function over possible languages, not a single solution with 100% probability.