The Phillips model of a random legislature was tested in the late 1980's.
Ernest Callenbach, my co-author on Citizen Legislature and I got 50 Berkeley California citizens to come to a Saturday meeting. The random sample came from the voter registration roles and the invitation was sent by the Berkeley Mayor Loni Hancock.
I just came across a photo of this historic event; in the upper right.
We broke the 50 Berkeleyites into 4 facilitator moderated sub groups to discuss issues of concern in Berkeley. We then convened two new groups to combine and synthesize the four sub groups; that is the group in the photo. The results were presented in the afternoon to three members of the Berkeley City Council and the Mayor.
What did we learn? Three important findings.
1) Ordinary citizens working in a group are extremely cooperative and fully comprehend the needs and operations of city government. They made short lists of needed improvements and agreed on them. They always found a source of revenue for any expense they agreed was needed.
2) All of the citizens had tried at some point to participate in their government and all found, with no exceptions, that the abusive language, personal vendettas and base level of discourse in political environments was off-putting. They all stopped trying to participate in local politics...the process was too demeaning and unpleasant to warrant further effort.