{"id":4927,"date":"2018-02-10T22:40:22","date_gmt":"2018-02-10T22:40:22","guid":{"rendered":"https:\/\/davidgerard.co.uk\/blockchain\/?p=4927"},"modified":"2018-06-06T20:27:46","modified_gmt":"2018-06-06T20:27:46","slug":"do-you-need-a-blockchain-probably-less-than-wust-and-gervais-think-you-do","status":"publish","type":"post","link":"https:\/\/davidgerard.co.uk\/blockchain\/2018\/02\/10\/do-you-need-a-blockchain-probably-less-than-wust-and-gervais-think-you-do\/","title":{"rendered":"Do you need a Blockchain? Probably less than W\u00fcst and Gervais think you do"},"content":{"rendered":"

The general answer to “do you need a blockchain?” is “no.” As I detail in chapter 11<\/a> of the book,<\/a> append-only transaction ledgers with cryptographic tamper-proofing are good and deserve wider use \u2014 but the “distributed” bit is largely superfluous.<\/p>\n

You’ll basically never need a so-called “permissioned blockchain.” If you have a trusted third party, you don’t need any sort of blockchain \u2014 and in real-world use cases, you’ll almost always have a trusted third party, or central administrator of some sort.<\/p>\n

When Satoshi Nakamoto created Bitcoin, he took the append-only ledger \u2014 a robust construct that had been around for decades \u2014 and bolted on a torturously convoluted<\/a> and stupidly wasteful<\/a> consensus mechanism, to determine who got to write the next block, in order to avoid having any central control whatsoever.<\/p>\n

This solved a problem only the economically delusional<\/a> thought they had, and even then it only solved it temporarily \u2014 Bitcoin had recentralised by 2014. It was an extremely clever and interesting hack \u2014 but that’s not the same as being useful for any other purpose.<\/p>\n

Most business usage that thinks it needs a blockchain just wants the append-only ledger. And we’ve had those for decades.<\/p>\n

The best-known example is probably Git<\/a> from 2005, four years before Bitcoin. This is a program by Linus Torvalds, the creator of Linux<\/a> \u2014 it’s his second big hit. He needed something to store the Linux computer code in, that would let you readily look through the full history of the code,<\/a> back to the beginning.<\/p>\n

Git has cryptographic tamper-proofing, as part of how it ensures the data is consistent\u00a0\u2014 if a single bit changes, all the hashes change. Software developers routinely copy entire full-history Git repositories around. Git is append-only ledgers on steroids.<\/p>\n

As I note in the book, your actual problem is usually cleaning up your data and its formats. If you really want an append-only ledger, you may just need Git, or a simplified version of Git.<\/a> Some actually-useful “blockchain” products can reasonably be described as a simplified Git.<\/p>\n

But the question “do I need a blockchain?” needs addressing in detail. Karl W\u00fcst and Arthur Gervais from the ETH Z\u00fcrich Department of Computer Science have written a paper on the subject: Do you need a Blockchain?<\/a><\/p>\n

This paper is worth your time. They explain the jargon at length, and discuss many commonly-advocated blockchain use cases\u00a0\u2014 it’s a useful survey of the area\u00a0\u2014 even as the authors are huge Bitcoin and blockchain advocates, and somewhat more optimistic for applying blockchains than is really warranted.<\/p>\n

The main takeaway from the paper is this flowchart:<\/p>\n

 <\/p>\n

\"\"<\/a><\/p>\n

To put that chart into text form:<\/p>\n