It does not make sense, practically. So they'll be using a balance with multiple moving parts made of multiple minerals that have to be precisely calibrated with margins of error adding up, instead of a simple platinum cylinder?
Defining your fundamental unit of mass in terms of a single arbitrary physical object costs a lot of theoretical purity, though it might not cause a lot of problems in practice. You mention a potential loss of precision. But we should gain accuracy by defining it in terms of fundamental constants. When the object itself changes, do we simply have a standard that drifts more than the physical constants of the universe? Do textbook publishers need to update all examples and problems using micrograms because the unit drift at that scale became significant? Should anyone without access to a reference object not get a real, accurate value for the kilogram? Do we keep using the mass value we know the reference object had, or does our kilogram really change with the object? Do we need to update any equations with a constant in them involving mass in any unit or derived unit in the equation once a year for high precision applications?
The problem with a physical standard like that is that you can't (easily) ship it to labs all over the world so that they can calibrate to it. At least when you use fundamental constants, each lab can set up their own equipment to produce the correct measurement.
Also, anything physical will shed atoms, which will affect the mass.
> Cumpson suspects that because the kilos living in national labs have been retrieved and handled more frequently than the international kilo, more carbon-containing contaminants have built up on them over time.
The simple platinum cylinder's margins of error are pretty much unbounded over time. If you come up with a good physical definition, our knowledge of the value will only get more precise over time as the means for measurement are improved.
It's the opposite: it makes sense practically, because it means a setup that can be reproduced "anywhere" based on pure knowledge, without depending on being able to ensure the continued integrity of a physical object that has been shown to change over time.
Most people can continue to depend on prototypes for calibration, so it does not cost us any complexity in that respect. What we're gaining is to be able to independently re-calibrate multiple prototypes.
I wouldn't say politically, but I'm sure that scientific aesthetics play a role. There are people who have devoted their careers to the development of fundamental standards. We search for the next digit of h, for the same reason why we search for the Higgs particle -- because we're curious.
A better kilogram might not have a practical use right away, but it might in the future, if for instance we want to do things like look for time dependent variations in the physical constants, tiny departures from existing theories of mechanics, etc.
Although, it makes sense politically