Evolution (as proposed) only allows random mutations at random times. Some live - and some of those mutations may get passed on to the next generation. The rest die.
His premise requires a long series of changes - unless you allow that the DNA/genes are this “storage device” - but then you still need to (1) cause a second random mutation to “stop the first (beneficial) mutation, then (2) “store the stopped mutation sufficiently accurately so that it can be re-started, but NOT be active for many generations; then (3) have ANOTHER random mutation to “unstop” the second change - but NOT destroy the first random mutation; then (4) and a fourth random mutation NOT happen to stop the now-needed third random mutation from getting “turned off” too early.
I supposed all that could happen. Randomly.
8<)