This could all be because of name recognition, it could all be because of insiders at the DNC deciding in advance who the candidates are going to be, or some such thing. But the variation among these four candidates does not correlate to their own levels of name recognition, and at least one of these candidates is very powerful in the DNC, so I’m thinking none of that is key. These are good people. They are impressive, and they impressed in the debates. These four candidates could provide the nominee, any one of them could rise up out of the very low numbers and become a key contender,the nominee, even the president. But for now, there is really only one thing to say about the polling numbers, using the same data set as described here, for Booker, Yang, Klobuchar, and Castro: Rounding errors.
That strange pattern you see there that looks like layers in a cross section of a pristine tropical rainforest, that’s rounding errors. All the internal structure of these data is from rounding errors. Even the ranking could be so affected by rounding that I don’t think we can say much about these candidates except to wish them well.