I just don't think that the fraction of the healthy adult population with ZMP has risen much since the last time unemployment was 5%.
Here is where the term sustainable in patterns of sustainable specialization and trade comes in. Suppose that the last time we had 5% unemployment and PSST was fifteen years ago. Since then, we have had periods of 5% unemployment, but they have coincided with unsustainable bubbles.
It is indeed implausible that a large number of workers have become ZMP in just the past three years. But it is somewhat more plausible that over the past fifteen years the Internet has made a lot of workers ZMP. At least if you do not insist that they have exactly zero marginal product, but you merely take ZMP to mean "too little value to be worth the expense of hiring and the cost of supervising."
As an aside, I would say that critics of ZMP should be trying to respond to the Garett Jones and Daniel M. Rothschild work showing that many of the workers hired under the stimulus were already employed previously. If the unemployed are not ZMP, then one would expect to find the recipients of stimulus money fishing in that pool rather than in the pool of people who already have jobs.