Erasure is just as type safe as reification.

Erasure gives you a little less flexibility to express certain constructs than reification allows, but from a type safety standpoint, the two approaches are equivalent.

If you want widen the meaning of "type safe" a bit, I would argue that reification is "less" type safe in the sense that it allows you to perform additional runtime type checks / type casts, which basically means you are second guessing the compiler and invalidating all the type soundness that it has provided you by accepting to compile your code.

It would not have allowed one to seal the loopholes demonstrated in the article?

How is reification any more type safe than erasure?

Would it not have allowed one to seal the loopholes demonstrated in the article?