From blindfolds to naked emperors: swing and fair electoral boundaries

Electoral districts contain communities with established voting patterns. Australian parliaments have given the task of redrawing electoral district boundaries to independent commissions, but the commissions understand their independence to mean they cannot look at voting patterns. Working blindfolded, they can unwittingly build partisan bias and poor responsiveness into electoral boundaries, and sometimes they have. Even after one-vote-one-value, unfair election outcomes still occur, calling into question the legitimacy of governments so formed. For 20 years South Australia's redistribution authority has drawn electoral districts with voting data in mind, and the parties agree that election outcomes have been fair. What would it take for this model to be useful elsewhere? It assumes uniform swings: have swings been uniform enough in other jurisdictions that any set of new boundaries could be judged on the basis of past voting data? (Parties and analysts do this now, but would it be accurate enough for a redistribution authority?) How much leeway could the parties accept - would a party winning 51% of the vote accept losing government? If swings have not been uniform enough, how complex a modelling exercise could we ask a redistribution authority to undertake? When the South Australian redistribution methodology was changed, the commission relied heavily on political scientists as well as the parties. Political scientists must be prepared to engage with this area, if we are to have a redistribution methodology generating electoral districts that produce fair election outcomes.