The traditional values are no better. The metre is traditionally defined as one ten-millionth part of the distance between the geographic north pole and the equator along the Paris meridian. The planet does, and will continue to, slightly change its shape for the same reason that the traditional second - the sixtieth part of the sixtieth part of the twenty-fourth part of a day - has gotten quite a bit longer. The gram is traditionally defined as the mass of a cubic centimeter (or millilitre) of water at a specific temperature and pressure (and let's hope that we don't see changes in the behaviour of water molecules over time as well).

There's a sort of mythos about the metric system being all "sciency", but it's just as arbitrary as any other system when you get right down to it. A metre is one ten-millionth part of an arbitrarily selected measureable distance because, well, it's "about yay big", close enough to the yard (an ad hoc body measurement) everyone had been using anyway. (The nearest familiar alternatives would have been either two rods or a hand, or the French equivalents.) It's just easier to do the math in a lot of cases because base ten. Anchoring the values to something anyone can, in principle, measure anywhere in any reference frame, at least makes sure that we're all talking about the same thing - even if that thing changes. That said, there's no evidence for the proposition that they do change, and quite a bit to suggest they don't.