This makes no sense. Although you posted it a while ago so maybe you were confused at the time, and already by now understand the situation better.

The supply went from 3 million to 16 million precisely due to the pump that you claim is a good thing. No one "thought it was a good idea to print millions of SBD to counter a ... pump", the system did that entirely by itself automatically as a direct consequence of the pump and of the long-existing (since 2016) rules and economics.

The change after HF20 to increase the print rate cap made very little difference to this; well under a million SBD was printed. Printing, as currently set up (both before and after HF20) is an exceedingly slow process. It takes about a year to move the debt ratio about 2% through printing. Indeed HF20 had not been implemented, and if nothing else had changed (a somewhat implausible scenario, but this assumption maximizes the possible impact), the supply of SBD would still be about where it is now, which is to say something above 12 million (maybe right around 12 million instead of the current actual 12.5 million), the debt ratio would still be >10% and the haircut rule described and its side effects described in the post would still be happening.

The only way to imagine supply at (or anywhere near) 3 million as you suggested would be if the original pump never happened.

What you've done here is make a compelling argument how that sort of extreme pump in SBD are not a "good thing" and are quite disruptive. That was not fully appreciated in 2016 which is why the system doesn't handle it well and also why renewed emphasis has been put onto improving the ability of the system to respond to these sorts of pumps.