The corridor system is the natural tendency of money

Dudley's Defense of the Fed's Floor System, Take 2

But that proviso is hardly the minor qualification Dudley makes it out to be. Instead, after more than a decade the Fed has yet to determine how many reserves are "sufficient." And despite what Dudley claims, it is far from obvious that adding more reserves to the system will prove capable of smoothing what he blandly describes as "a few bumps in the road" toward a floor system that works in practice the way it's supposed to work in theory. Instead, as Bill Nelson observed at a December Brookings event, the Fed could well end up discovering, as Norges Bank did a few years back, that under a floor system the demand for bank reserves tends eventually to catch-up to the supply, making the effort to satiate the banking system with reserves once and for all as futile as a dog's effort to catch its own tail.

Fixed uncertainty yields a corridor system. Not just for currency banking but for avalanches and black holes and tree trunks and chicken soup.  Fixed uncertainty results because agents minimize uncertainty until they reach market limits on certainty.

 So, we see that excess reserves track treasuries held, always since the floor system begin.

Fed should dispense with interest on reserves. Instead, they'd like to see the Fed move toward a standard "corridor" operating system, with interest on reserves determining the corridor's lower limit, and a standing lending facility—repo or otherwise—setting the upper limit. This is a tried-and-true system employed by many of the world's central banks. Yet the Fed itself has yet to try it.

Says George Selgin.  A more condense form:'You cannot fool mother Nature."