Look at row 2 of the second matrix; it contains 15 and 4. After any sequence of column swaps, any row that contains 15 and 4 will still contain these two numbers. A similar statement goes for any sequence of row-swaps: since there's a row containing 15 and 4 before the swaps, there'll always be some row containing 15 and 4 after the swaps. But the first matrix doesn't have 15 and 4 in the same row.
A proper proof of this would involve induction, with the inductive hypothesis being that after $k$ swaps, the matrix will have a row containing both $4$ and $15$, and the overall claim being that for any sequence of $n$ operations applied to matrix "b", the $n$ result cannot be matrix "a". But I'll let you write that out.