This rule is actually afaik one of the reasons why this discussion has started, as it became a matter of perspective if and how this barrier is applied and thus if and when it is trivial or not. After an incident some time ago, that has shown different interpretations of how/when to enforce this rule, the Join SIG agreed that distributing memberships in their SIG to users, whom they felt to be worth to vote, is ok. So, when someone wanted to vote they felt worth it, they make them member so that they can vote, and then remove them again from the FAS once they voted → so was their agreement, it happened only once I think, but this revealed a little also that different FAS groups had big differences in when and how they allocate memberships. This “provokes” asymmetries we cannot measure.
I am not against your points, and I agree that Fernando’s concerns are serious. As I wrote, I also see some entry barriers critical as they might exclude important groups, and that can be worse than including the “wrong” ones (I simplify to illustrate:). But if we keep the system the way it is, some alignment of when and how allocate FAS memberships should be done. It can be even something very generic, but with some barriers: this example is not about if the Join SIG was right or wrong, but that blurred unclear rules, and the resulting different interpretations, can lead to problematic incidents due to different perceptions (and thus conflicts, which happened some time ago leading to this).
However, another idea to solve this by technical means would be to just add one or two conditions, such as that only people who are for over 1 year member of an FAS group can vote or so. And the condition to just logging into FAS 3 or 4+ times in the recent year (with at least 2 weeks or so in between each eligible login or so?) might even exclude 99% of those who MIGHT think of logging into their “dead” account solely for the purpose of voting (I agree with you the “dead voters” are likely not relevant though). The latter unlikely would affect any “real” contributor, and the first might make people to get involve in Fedora and get an overview of what Fedora is and how it works before voting to FESCo → a year to “learn” could be desirable? Just some thoughts that might contain alternative compromises 
The example-conditions conflict with some thoughts of Fernando, but I am not sure if the average person only joining to vote for FESCo (not necessarily applicable to other boards) can be assumed to vote with predictability, which means to vote to get the result they actually want, which is not straightforward or reducible to a few sentences in the candidate introductions (a major issue in many ballot systems and origin of many contemporary issues in them). However, alternative conditions are possible too (that consider other stuff that is not necessarily bound to an FAS account or so?).
Just a thought: how would we know?