A combination of a cryptographically-chained ledger such as this: https://docs.microsoft.com/en-us/sql/relational-databases/se...

... together with a trusted third party covers something like 90-95% of the supposed applications if blockchain. It's insufficient for typical mass-adopted, international currency, but for damned near everything else it is "more than good enough".

In the real world, before blockchain, what mutually distrusting corporations or people would do is set up a distinct third party organisation to keep the "ledger" (or whatever).

For example, banks use non-bank companies like Experian to do cross-bank fraud analysis. Similarly, many industries send their sales stats to a third-party non-competitor org that aggregates the data, and then sells back industry-wide anonymised sales stat reports. This lets companies that don't trust each other exchange data, etc...

This was a solved problem centuries ago, and technologies like Azure SQL Database Ledger make it nearly as robust as blockchain, with none of the downsides.

In other words, the NIST flow chart needs more steps towards the bottom pointing to both existing and new non-blockchain solutions.

In the last few steps, going from “Yes, you need trust-less updates” to “<tiny distraction>” to “Meh whatever, just use a regular database!” seems a bit lacking in short-term memory…

Well, if you need more proof that you don't always need a blockchain, Webcash ( https://webcash.org ) might be it because Webcash just uses a central database for double spending detection. A few of the early users were commenting that the project has put lots of altcoins to shame from an architectural standpoint.

This is wrong. Blockchains validate data. Their purpose isn't to store it. Mostly the answer to do you need a blockchain is no