It would only "even out" if the different understandings of "acceptable" didn't have any correlation with people's political preferences. Given that we know th

It would only "even out" if the different understandings of "acceptable" didn't have any correlation with people's political preferences. Given that we know there are personality traits that are strongly indicative of party membership[0] (not to mention education level), I think the default assumption should be that different parties will be affected differently by voters interpreting the question differently. Also, all aspects of voting systems have to be considered in the context of an adversarial environment, and you have to imagine that some parties will be more unscrupulous than others at exploiting voters' confusion, or just better at using polling data or other strategies to instruct their voters how to make their votes more powerful. I suppose you could argue that we already accept that tactical voting is a reality under FPTP, and maybe in practice the situation under Approval wouldn't be any worse over all (although it might be better for some parties and worse for others), but even if it is better, you have the problem that any change adds to the "attack surface" of things that election losers can point at to cast the validity of the results into doubt. That's why any reform needs to ideally involve voter instructions and election outcomes that are more intuitive than FPTP, to counteract the potential for additional misinformation to be introduced. [0] https://practicalpie.com/personality-and-politics/ https://practicalpie.com/personality-and-politics/