Joe Hall has this interesting post on some forthcoming research, indicating the central-count optical scan systems tend to “lose” more votes than precinct-count optical scan systems in precincts with relatively high concentrations of African American voters. That’s probably because precinct-count systems provide voters with notice of mistaken “overvotes.” This can be a problem with central-count systems as was documented by the media consortium study of Florida’s 2000 ballots, which found a high percentage of overvotes with central-count optical-scan systems.