I probably avoid that necessity (if it is one) by only betting on meetings where it is unlikely that any "useless" jockeys are not likely to get a ride, so I trust the connections' ability to ensure that. This means that I can start from the assumption that all jockeys are equal or at least that any variance has a very small impact on overall chance and I if I'm wrong, it's because I don't know how to reliably assess that variance.
So, for example, I've been working on assessing Oz racing for a year or two now and expect to be betting there next year. To start with, I expect that will be limited to Sydney, Melbourne metro meetings where I can reasonably apply the above assumption, though the limitation is not solely for that reason.