All of these things except "without intermediaries" are already solved with intermediaries. Intermediaries are desirable: they solve problems so that the end users don't have to do it. It's fundamental to programming, in fact: you don't handwrite machine code, you use a language with libraries that gets interpreted or compiled to run on an operating system that provides lots of useful facilities and abstractions so that you don't have to care whether the machine is connected via a 3Com 905b or a Lucent Orinoco, and the same code that worked over those obsolete interfaces still works over a 10gig fiber NIC.
In particular, intermediaries allow people to fix mistakes. And people make mistakes all the time.
People like banking; it's individual banks that they hate. Better regulation fixes that. Unregulated transactions are terrible: the entire history of finance proves that people will lie, cheat and defraud each other if given half a chance.
Financial safety comes from the ability of a trustable third party to adjudicate and correct mistakes and disputes.
If you want to compile a credit rating, you need an accurate history of transactions. Blockchains don't give you accurate histories, they give you timestamped signed logs of entries that are really really difficult to amend. "This landlord reported that I was late on rent but I've never lived in that state" is a complaint that a credit agency must accept and evaluate.
Eliminating notaries: the purpose of a notary is not just to say "this event happened at this time" but to say "I witnessed this event happening at this time". A blockchain can't do that. You need to trust the notary as well as the notary's log, and random people adding entries doesn't make them trustable.
Record of private property and deeds: you literally want a single authoritative database here, where every write operation is done by a trusted person.
Each of these cases is not "we need a blockchain" but "it's good to have a signed, difficult-to-forge journal that can be inspected and verified".
Your last point is basically "we don't know what cryptocurrencies are good for."
