It is sort of measurable. How many lines you output, how many bugs are in your code, how much time is spent fixing your bugs. Etc.
I mean its far from perfect and leadership roles are harder to measure (but can be measured by how the people under them think about them). But still I get your point.
I think the best we can do is have an expert familiar with the task and codebase evaluate contributions. Trouble is, this can be corrupted by all sorts of internal politicking.
We still do Agile Scrum, and our velocity has gone up significantly. Experienced team, mature code base, no real external forces to skew these figures.
Goodhart's Law does not apply, as we have done this for many years and are happy [1] with the way we do things.