With a spring scale, you're measuring the weight of the object by balancing the force exerted by the springs vs the force exerted by the mass. Given a different local acceleration, the spring scale would register a different rate. Given an appropriate gradient, the balance scale should always give the right answer--you're comparing masses directly with each other, and although the medium is via their apparent weight, in a very real sense the result you're getting is the object's mass.
In other words, with a spring scale, given what you're measuring and what you're measuring it with (the object and the parameters of the spring), you can calculate the local force field. With a balance scale, you know nothing about the local environment other than there's some persistent direction of force.
Anyway, in real terms, so long as the system is self-consistent in which value it prefers, neither is inherently preferable; one might be useful for some use-cases over others, but that line of attack is like debating between mks and cgs units, which is not an argument any sane person wants to have. The idea that imperial being bad because it's hard to convert to an entirely different system assumes that the different system is better anyway, and that when doing your own calculations, you would always want to convert