> the rest can be done and is literally often done by email (sometimes even signed!) and spreadsheets Well, yes; signed email sent through anonymous remailers

> the rest can be done and is literally often done by email (sometimes even signed!) and spreadsheets Well, yes; signed email sent through anonymous remailers hosted by a network of peers (e.g. Mixmaster) is the direct ancestor of blockchains. Designed by the same group of people (the "cypherpunks"), even! Blockchains just add to Mixmaster-like networks a shared reputational-currency ledger with some inherent means of acquiring a balance on it, to facilitate the feature that spam-email-solution-proposers call "eStamps" — which prevents blockchain networks from being flooded with spam transactions, while also not requiring per-account identity verification. (Remailer networks had to deal with a lot of spam in their pursuit of anonymity.) Essentially, blockchains are a formalization of "signed emails and spreadsheets." However, there's a specific thing you get from blockchains' peer-to-peer distribution, that you don't get from "signed emails and spreadsheets" in this example use-case. The problem is reconciling the state when the person in the US doesn't trust any computer hosted/set up by the North Korean, and vice-versa. They can pick a third party, but how do they collaborate to set up the escrow system to enable the third party to reliably send their third signature; when the future two of them now disagree and are actively trying to trick one-another into thinking that e.g. the third party cannot be reached? Even if the third party hosts the escrow system themselves in some neutral country — the two of them probably would just DoS-attack that centralized system to prevent the third party from issuing their judgement. Presuming a big-enough blockchain, and anonymity for the arbitrator (enabled by the "eStamp" ledger property), you can't effectively silence the third party from publishing to it.