Why does alpha equivalence not apply to free variables?

56 Views Asked by At

It's mentioned in the book "Haskell programming from first principles" in the first chapter that one free variables don't have equivalence that is:

$$ \lambda x.xy \overset{\alpha-\text{eq.}}{\neq} \lambda x.xb$$

Could someone explain why this is so? What is the problem if we allow equivalence for free variables just like bound variables?