Look at the compromise of alternating slave and free states to prevent a rupture in U.S. politics in the mid 1800s. It satisfied neither side, and no compromise possibly could have.
So-called progressive issues and so-called conservative issues are just attention-seeking window displays designed to farm voter interest.
Policy is controlled by the economic leverage of donors and lobbyists, not by voters, and is enacted to protect the economic interests of donors and lobbyists, not of voters.
But if you apply a selection criteria to members of the crowd or the types of crowd recognized as such, you're likely going to get answers biased toward the selection criteria.
Likewise, to the extent members' expressed preferences reflect abstract arguments (i.e. ideology) rather than their real interests, then other dynamics will control. So, for example, if you go around asking the population detailed questions about their stances on abortion--when it should allowed, when, etc--you're going to get a ton of different answers and any consensus, to the extent there is one, will be sophisticated and nuanced. Ask those same people if they're pro-life or pro-choice and the majority consensus will simply reflect the ebb and flow of the dominate idealogical narratives and be as real and substantive as tallying the number of people who prefer cake vs pie.
In that sense I think Wisdom of the Crowds always had a sort of tautological character to it. And it reflected an optimism about the information age--that it would cut through rather than perpetuate all the complicated and complicating intermediate social layers.
in any case, representation itself could benefit a lot from experimentation and innovation. we now have the technological means to implement a direct democracy on the scale of billions of people, something practically impossible only a few decades ago.
the wisdom of the crowds covers this ground too--too much correlated thinking via information cascades often leads to bad decisions and outcomes. you want more independence, even if that increases the number of "wrong" opinions. having more wrong opinions is infinitely better than poor information cascades, as long as they're relatively independent and queried in a statistically sound manner for decision-making.