In my country, if a first preference isn't clearly marked on the ballot sheet, then the ballot is considered spoiled. This is a pencil and paper, counted by hand system. It looks like what counts as a defect in Alameda County is a perfectly working tallying machine in other parts of the world.
Having said that, ranked choice is harder for voters to understand, so it makes it easier for them to spoil their ballot. You can ameliorate that using electronic voting machines (which I disapprove of); and you can engage in various kinds of voter education. I'm from an FPTP country, and I wish we had ranked choice. But trying to read a clear intention to a ranking with no first choice is like reading tealeaves, and election officials shouldn't do it.
Officer: And remember it's not an X, it's 123 in order of preference.
Voter: But I only want to vote for 1.
Officer: Well you can do that but it has to be a 1 not an X.
Voter: Not an X?
Why? There's a wide spectrum of voting machine implementations that go between "ballot printer" (basically zero concerns of hacking) and "opaquely tallies votes with no audit trail". When it comes to complicated voting schemes like this, having a machine greatly improves the user experience by allowing the ballot to be validated before it's dropped off.
We do hand-counts here. I know pretty well how they work, and people trust them. I'm not familiar with all the varieties of electronic voting machine; but as far as I'm aware, none is certified by any federal agency, and most (all?) are proprietary. I'm not keen on having commercial businesses (with commercial secrets) involved at the heart of vote-counting. Instead, I'd have central government specify how voting machines must work, and certify that each proposed model meets specifications. Really, the innards of these machines should by law be exposed to scrutiny by the general public.
There's a lot of distrust in the USA voting procedures. Disenfranchisement is routine, and it's also routine for candidates to attack the voting system when they lose. That's partly because people don't trust electronic voting.
I'm not from USA, but I think the USA should enact voting laws that are uniform across the country. I think it's nuts that voting laws are state-by-state, even for federal elections. I don't understand why former slave-states should get to make their own voting laws for federal elections.
For instance, if you could select 8 people for the school board, if you selected 9 it would show an error.
A poll worker would then talk with the person and inform them about what mistake they made, and give them the option to have that section not count, or go fill out a new ballet.
You then put the ballet in a box for manual counting later.
A similar system could for for basically any somewhat "difficult" instructions for voters - you shouldn't assume what they want, you should have strict instructions but if they make a mistake that would spoil their ballet, they are informed of it and can fix it.
Would this not mean that the ballot isn't entirely secret? I suppose that's the compromise being made to ensure no accidental spoilage.
If a voter didn’t select a first choice, then the second choice should have been counted as the first choice in the first round. The same would occur in subsequent rounds moving lower choices up into the empty slot. Instead, the erroneous algorithm didn’t count any vote in a round if a space was blank.
So effectively your choices are ranked based on the ordinals you chose, not on the absolute values of the ordinals.
I wonder what happens to ballots that use non-standard symbols whose intent is clear. I would guess that a non-zero number of people write “A, B, C” or “first, second, third”. (Or even, in the context of this ranking-gaps bug, ballots that just have “second, third” on them.)
Protocol implementations that are “liberal in what [they] accept” certainly make life overly complicated. Every new implementation of a DNS server or HTML renderer has to have all sorts of awful quirks because of a lack of strictness in historical (and now widely deployed) client implementations.
One translation of this to voting is that if you don’t fill in all the boxes and use “1, 2, 3” then your vote just doesn’t count — it would make the tally algorithm much simpler — but that also goes against the principle that a tally algorithm should go out of its way to scry as much into the intent of a voter as possible, and reject as few ballots as it possibly can.