That has so many wrong assumptions. First off the assumption that there is a group trusted in aggregate. If I trusted them they would have been party to it in the first place!
Second the logistics of it are "clever dick" as opposed to a real solution - like suggesting using one off Hoffman encoding to compress a large file to only 1 bit.
Third - blockchain to protect against leakage? That is the exact opposite of its job. Buzzwords aren't magic spells.
Your #1: Your trust is irrelevant when discussing whether mnw21cam's list is complete, as your trust was not part of the criteria.
(That said, on trust "If I trusted them they would have been party to it in the first place!" doesn't make sense. There's a huge difference between wanting to share all your content with a group of people all the time, versus trusting those people to make a collective decision to release your documents, for example when they agree that you have died. People are seriously examining this sort of mechanism now because it's relevant to modern life. For example releasing your password and accounts store to family or trusted friends upon your death or incapacitation, using some kind of distributed dead man's switch that needs human judgement to confirm.)
The fact is, mnw21cam's statement that there are only two possible branches because of maths, depends on the assumption that keys "will be leaked" being inevitable.
A sibling commenter believes it is inevitable they will be leaked no matter how sophisticated an aggregation mechanism is used. That is a reasonable argument, though one I disagree with.
If you have a threat model strong enough to break the distributed consensus mechanism of things like Ethereum, then you have a threat model that invalidates branch 1 in mnw21cam's list as well as branch 2, so you cannot win: Under that model, you cannot have "secure legal communication that no-one can break" because your own device is vulnerable to compromise as well. You should find a way to talk without a recording device.
Your #2: What I've said is a real logical possibility, and in fact is what we might actually end up with in a number of areas of life. It is not as 'clever dick nonsense' as you think. It might be an undesirable idea, but it is a technically possible one.
Your #3: Yes blockchains store and publicise. They also implement strong distributed consensus, and on top of those other things are layered, such as Ethereum-style smart contracts, and zero-knowledge calculations. If you think those cannot be used to control the release of fragmented or encrypted secrets driven by measurements of human decisions, you haven't understood them yet. I really recommend you look at ZKProof.org, and have a think about how privacy-maintaining blockchain coins like Zcash and Monero are able to use a public blockchain to exchange secret transactions.
Note: I have no stake in this, I'm not a cryptocurrency or blockchain fan particularly. But I do understand how they work, and I'm not fooled by the buzzwords (at all, I'm quite skeptical).
Second the logistics of it are "clever dick" as opposed to a real solution - like suggesting using one off Hoffman encoding to compress a large file to only 1 bit.
Third - blockchain to protect against leakage? That is the exact opposite of its job. Buzzwords aren't magic spells.