Aye, this is an essential question. My analysis, and as far as I can tell the proposal’s original intent, was that funded parties who are among the 3 entities in the “2-out-of-3” multisig will participate in voting on their own proposal, and thus require just one more vote.
However, ECC Initial Assessment of Community Proposals asserts that
members of the panel should be required to recuse themselves from any decision in which they are also a potential recipient of funds
Under that, ECC and Zfnd proposals would effectively be subject to 2-out-of-2 rather than 2-out-of-3.
Moreover, it would decouple the technical aspect of locking funds under a multi-signature, from the intent. Technically, a rogue party holding one of the 3 keys can decide to use it to fund itself (with cooperation of just 1 more party), regardless of whether it’s supposed to recuse itself according to a social contract. Done blatantly, it would be the kind of egregious violation that may cause a hardfork. But there could be marginal cases, e.g., a party refusing to recuse itself from voting on funding someone who’s affiliated with them.
So under the latter interpretation, the crisp meaning of multisig becomes fuzzy and potentially contentious.
In any case, the proposal should explicitly say whether and when parties are supposed to recuse themselves or be excluded from funding decisions.