water_moon wrote:BUT, paper's not perfect (remember hang chads...)
That wasn't a problem with analogue paper. It was a problem with trying to digitise it. A proper system should already have been in place auto-sorting any papers that had not been readable, as any machine-readable vote, into a pile that could be assessed by human eyes. At leisure, but planned for. The machine-reading reject rate being an instantaneous guide to if more help needed drafting in, or a full-blown issue raised without having to go the whole way to the courts to bounce around some more.
For multi-question (single-choice per question) sheets simplest to dump all 'incomplete' ballots for checking,with false-positive on genuinely un-voted-for options, but still fairly simple to dedicate a reading machine to take 'any or all' rejects and assess Q1 into votes (tallied on Q1 only) or no-vote (human checker/tallier on this line only), feed back in for Q2 (ditto), etc.
If it was so totally failing with hanging and pregnant chads (falling upon whatever globally pre-agreed method was of establishing a voting intent*) that it overloaded even the worst-case scenario then this would have been noted early. As would inconsistencies in failure rates between districts (say those with a certain demographic bias had been issued punchers with slightly back-adjusted punching prongs..?) and a properly transparent process would have made that obvious.
That different systems were employed across the same state (because nobody thought there might be a planned state-wide ballot at some point, so they didn't have time to ramp up the resources to be consistent??), and that that
distribution was demographically biased, was also on the verge of suspicious, probably.
There's no problem with paper or
electronic voting that can't be out-problemed by a bad implementation atop of it, but right now we have better ideas about how to not badly implement paper. The only better system might be the ball-casting method**.
* I would have said a single clearly distorted punch-point (unrivalled by any other punch-point) should suffice. With perhaps the earlier-term 'pregnancies' perhaps being squirreled away/brought to the central count-point for consistent adjudication. But if there's a noted intention
, it should count, even if a machine is too dumb to note it.
** With suitable measures in place to prevent individuals using slight-of-hand. I suggest oversized tubs of coloured balls that can be fed singly, from the tub of one's choice, into an inspection chamber/optic. The optic has a lever to drop the choice into the ballot-bottle below (vote done, mechanism locked for that visitor), or a separate control to send a misdecision/error ball into a discard pot. The ballot-bottle can be sorted by any suitable means (optical sensor, differing manufactured buoyancies of the balls) into reseparated vats for totalling purposes, with mis-sorts by accident or nefarious design being visible, as can the contents of the discard pot to top up any feeders getting low. (There'd be enough balls of each colour to deal with every voter being possibly unanimous with the rest, maybe double or more, but just in case someone tried to empty an opposing feeder into the reject pot...) Feeders (not shown in their entirety to the voter) would be checked periodically by polling staff, topping up randomly, but not during a voter's residence within the booth itself. The optic-Ok lever might ding a small bell and the optic-reject mechanism might have a different audible signal just
to reinforce the oversight process. Now, how to guard against something like colour-changing substances under the influence of time, temperature, UV or similar? Hmmm…