You are overlooking the disadvantage of not verifying the prior hash, which you can't do until you can verify hashMerkleRoot, which you can't do until you have collected the entire block. If you skip verification, you open yourself up to bad actors feeding you a bogus header - and you only figure that out after being drip fed a bunch of transactions that don't match the merkle root. Erasure coding doesn't fix that. In any case, proponents of the 'cram everything in the public ledger' method have already advocated 128MB and even completely uncapped block sizes... so the outcome is predictable.
Take a hint from every successful network that has ever networked since ever: hierarchy, extensibility, etc. Demanding that it all be crammed into a homogenous block is so obviously doomed to fail that you should be suspicious of the proponents' motives.
Take a hint from every successful network that has ever networked since ever: hierarchy, extensibility, etc. Demanding that it all be crammed into a homogenous block is so obviously doomed to fail that you should be suspicious of the proponents' motives.