> If you want the same results from every run ...

That's not how "bureaucracies" seem to actually function in the real world though, regardless of the theory. ;)

We only hear when it goes wrong

But it goes wrong very often actually (meaning the outcome is non repeatable). Most of the time you don’t find out, but if there’s two or more people doing the exactly same process (application for getting something out of say a government institution). You just need the forms to end up with different clerks to get two different outcomes.

This is why in Germany the common wisdom is for say claiming disability, you will always have to reject the first ruling, no matter what. You will get a better result the second run through (maybe even reject that, too).

Imho it’s the curse of complexity. But there is little incentive for bureaucratic institutions to „refactor“ their processes, as it would mean less headcount.

I am not convinced it goes wrong a significant percentage of the time, compared to the volume. I don't have the data one way or the other though.

Regarding headcount, I don't contend the issue, but then keeping more people employed is not that bad of a goal in itself.

It is very frustrating that different clerks can result in different outcomes. IMO it's the lack of clearly defined procedures. However, I don't follow this bit:

> You just need the forms to end up with different clerks to get two different outcomes...You will get a better result the second run through (maybe even reject that, too)

What if you get a worse outcome and you have just rejected the friendlier clerk's work?