True, but if it overflows one year and then underflows the next, then the net shift is negligible. On the other hand, humans are adding a consistent, and growing, extra contribution on top of the natural fluctuations -- and, at the same time, they are removing quite a few of the natural "sinks".
So it *does* work, and it is quite easy to explain, and NJ's dismissal of the analogy is not unreasonable but incorrect. The simple fact remains, then, that the human's "4%" is very significant indeed.