I'd be interested to learn which way this being implemented in practice for multi-question ballots.
Also, mathematically a question with > 2 choices isn’t ultimately that different from a ballot with multiple binary questions. You can imagine my scenario with credits allocated to a single non-binary question as well, just substitute “poisonous question” with “poisonous option/individual” (though it is harder for me to demonstrate with non-partisan or bi-partisan examples).
Edit: Some plausible examples for distributing credits across issues include multiple open seats and top x contenders get them (so n binary yes/no “do you want x to have a seat?” questions), a ballot to determine which candidate you’re most likely to align with based off their yes/no stance on multiple questions (which is a problem if anyone’s stance is boiled down to a simple yes/no but that’s unfortunately the state of politics today with no shades of grey), etc.
Imagine a ballot with one contentious question A. Opposition Y adds a question B they know will flop but is guaranteed to disproportionately attract all the attention of the other party to the ballot. Party X voters waste their credits on the red herring question B (which has greater than 50% support in the overall population anyway and would never fail them) while opposition Y advises its voters to spend the credits on the actual question A they care about.
As a completely contrived example, the real question is a controversial “add a carbon tax” that could go either way and one party either adds “ban all abortions in all cases for everyone” or “ban all firearms for all people” to the ballot to misdirect. Either of these is guaranteed to disproportionately attract all the attention of one of the parties more than the other despite neither having a remote chance of passing. Because of the phrasing and the topic chosen, even if you don’t educate your voters you can rely on the fact that their voters simply care a lot more about the topic than yours do.