1y ago

The Australian Securities Exchange CHESS blockchain: seven years of sunk cost fallacy

davidgerard.co.uk

13 comments

By this point, I think it's pretty obvious that blockchain doesn't have any good use cases beyond financial speculation and scams/fraud.
- YOU WOULD THINK THAT, BUT
- sanctions evasion and money laundering, and that's basically it
- It has one legitimate use case that I can think of. Immutable audit logs, even then there are better options.
  
  so in practice it actually doesn't, and enterprise "blockchain" systems tend to evolve:
  do the real work in the blockchain bit
  do the real work in a program with an SQL database attached, claim the blockchain is for audit logs
  every SQL DB can produce those anyway, remove the blockchain bit entirely
  don't bother removing the word "blockchain" from your marketing copy
  multiple such cases!
  
  Blockchain, the solution in search of a problem.
  
  Careful not to conflate things like hash trees with Blockchains. The former do get used for stuff like certificate transparency logs right now, because it is a sensible technology. Blockchains could do exactly the same thing (because they’re based on the same underlying principle), only with much more expense and waste, so there’s basically no point.
Impressive they could spend a quarter billion to fail building a new database.
- it was really pretty impressive
In capital markets infrastructure low latency and high throughput are the 2 core design principles of any system. Blockchain manages to violate both; how anyone could have thought this is “the future” is beyond me. I still hear all too often “ya blockchain (or cryptocurrency) doesn’t work now but I’m sure it’ll be great and useful eventually!”
Why oh why are people so attached to this proven failed tech idea? I’ll throw it the smallest of bones and say maybe a distributed ledger with cryptographic proofs has some very specific minor use case, but even if that maybe is true for one thing why is it shoved everywhere it doesn’t belong?
In enterprise computing, “smart contracts” are called “database triggers” or “stored procedures.” They’re a nightmare, because they’re very hard to reason about or maintain, and they’re prone to unexpected and spooky effects.
It occurs to me that the situation's even more dire than this single-node description. If everything's in one database, then yes, a smart contract is effectively a stored procedure. But it can be worse! Imagine e.g. an MMORPG where city centers or dungeons are disconnected from the regional map to prevent overload. A smart contract might need to synchronize data between two databases, e.g. a dungeon and a surrounding region, to maintain correctness.
- yeah, i'm talking about the general case of not separating the code from the data

13 comments