You make it sound like having an open system for timestamped records with support for highly complex permissions validation with a deep audit trail that is resistant to fraud and posthoc record tampering has no utility beyond buying drugs....

They criticized blockchain, but still said it was useful. They didn't act like it had no utility, that's a strawman. When I see IBM advertising blockchain in the middle of football games to an audience who has no fucking clue what that means, or how little it is used in production, I know it's hype.

The blockchain is a useful protocol, but what you are describing doesn't even require it, just a few trusted sources, which blockchains have to fall back to, anyway, with trusted oracles.

Their larger point might be that as much a blockchain advocates love it, there's not a single mainstream application outside of Bitcoin 8 years later. Someone yesterday compared it to the hype behind torrents changing everything, though torrents might be more practically useful.

Torrents are used by Blizzard, among others, to distribute updates for their games

We've had all of that for aeons.

The only really novel thing Bitcoin brought to the table was the consensus algorithm (which is in no way, shape, or form a complaint — it's precisely what it should be) — but that consensus algorithm only works for currencies (for a sufficiently loose definition of currency that allows for the current fees and delays). Nakamoto consensus solves the double-spend problem, but only if mining blocks gives you a reward that's denominated in the same currency as the transactions you're mining.

OP talked about "Blockchain" not bitcoin specifically (which has more limited use cases, but that's not relevant to this thread)

A git history is a "blockchain" in the strict sense (A Merkle tree with branching factor = 1), so it's not a very helpful term. Bitcoin, at least, has the common decency of having a very concrete innovation as part of its value proposition, so you can discuss its merits. "blockchain" as a term used in the industry at large means so many different things that it ends up not meaning anything at all.

But he's right - most of the properties of the blockchain are not novel. Distributed consensus is the novelty, which is part of the problem with people talking hyper-enthusiastically about applications that either don't require, or don't allow it.

Ethereum uses essentially the same consensus algorithm for lots of things besides currencies.

My definition of "doesn't work" here is: the security guarantees that the consensus protocol provides stop applying. Just because nobody has found it worthwhile to attack Ethereum "not a currency" stuff yet doesn't mean that it's safe.

Ethereum's consensus model makes the same security guarantees for contract operation that it does for the built-in currency; given the value of Ethereum-based tokens there's a very large bounty for anyone able to poke a hole in that.

Of course this is only a guarantee that contracts will run as written, not that they're written correctly. There's a lot of research on applying high-assurance techniques to contract development but it'll be a while before that stuff is ready for production.

You can buy drugs with git?

Given the complexity of git commands, I cannot rule it out.

Oh here it is, git-buy-drugs.

It would never be that straightforward.

git rebase --bong

Dammit, should've said:

git freebase

Indirectly. You get paid for committing code, after all.

git pull pull pass

I don't think I made it sound that way, but if I did, it was in error. I understand the implications of the technology, but my issue is with firms applying it to literally everything, even when it's not cogent to do so.

Just use git then? A number of auditable ledgers exist w/o decentralization.

with today's privacy and security concerns, for any new project, would starting from a position of anonymous trust despite the increased resource requirements really be a downside? I ask as an outsider to the industry. I do understand that blockchain tech doesn't need to be slapped onto everything, but from my perspective it seems the benefits generally outweigh the detriments - at least so far

The resource requirements are a massive downside, but they are only necessary for the "permissionless" model. In almost all cases it is equally good to use a permissioned model where the parties are all identified in advance, which completely avoids the need for proof of work (because there are no sybil attacks when everyone has a known and fixed identity).