Travis noted:

the 58x0 series just isn't matching up to anything we have.

But I have still some workunits which are marked as invalid by a quorum of inaccurate 58x0-clients. For these 58x0-clients the WUs are valid and granted with credits, but unusable for the project, because of the failing reliability compared to the other hosts.

But my question is:

Will all the users who were getting no credit by a false 58x0-quorum will sometimes get there credits?

What to do with all these inaccurate results? Does everything ever processed by a 58xx since the beginning of time have to be chucked and reissued?

I think that this is the worst case scenario. But when accuracy is such a high value, and no other way can be found, maybe it is the only way to get accurate science results.

Of course the milkyway is big, and so an error in a digit far right of the point becomes significant