We should be able to use the non Sybils (> 1 million EOAs over a few years) as non Sybils and then test and see what results we get.