Hacker News Reader

Comment by ak_111 | Hacker News Reader

I do agree that tamper proof log is very useful, but I don't believe that it is something you can't achieve using traditional database tech.

I have a question for you that can short-circuit our discussion. You claim that the value is "so obvious" to you, why then one decade in there isn't one serious application of blockchain in the real world (not even in the dark web with early adpaters)? I mean a workflow that has been entirely transformed by it, despite all the investment and smart people working in the space? The lack of examples is the point that I find most baffling at all.

View on HN

You can only create "tamper proofness" aka security with social coordination.

Your bankaccount is somewhat secure due to layers and layers of agreements and interdependencies of people. Bitcoin works the same way, but in a much more modern way and achieves much better security.

The git repo you mention elsewhere is something completely different. There is no security baked into git. A git repo is just a bunch of files.

No workflows have been transformed in a decade because workflows do not change that fast. TCP/IP and and the rest of the internet protocol suite was developed the 60s. It started to disrupted workflows 40 years later.

> Bitcoin works the same way, but in a much more modern way and achieves much better security.

Is this true in practice? Haven't there been multiple massive bitcoin thefts? If these happened to a bank, you'd have a chance to get your money back. With Bitcoin -- it's just gone.

That is not the type of log level security we are discussing.

"Security" here means that a piece of data that was written to the log stays there unaltered.

"Theft" is on a much higher abstraction level. A "Bitcoin theft" is that someone steals your key and then appends a message to the log "Boplicity pays 17 BTC to Joe. Signed: Boblicity". Stealing your private key does not take place on the blockchain. And all previous entries to the log are still there.

And those goalposts just fly by!

These are the unmoved goalposts from this comment chain.

Please provide an example of any database that is tamper resistant. As a DevOps engineer I have setup and configured many different databases, I have yet to come across any that someone with admin credentials (which you always need at one least of), didn’t have the ultimate ability to tamper and do anything they will to database.

even simpler than database technology. A git repository is tamper resistant.

No it's not. You can rewrite history.

Not under SHA-1 it isn't.

I agree that databases on their own are never tamper resistant. But you can certainly set up a process around the database that guarantees tamper resistance. You'll still have to trust said process, and with blockchain you still have to trust the code. You'll never get away from having to trust _something_, but at least it is in theory feasible to have tamper resistance both with a centralised database and with a blockchain.

in practice, there's a difference between "our SREs and management promise to not change anything" and "it would cost x billion dollars to replace this record with a different one, based on previous attacks we've observed"

> "our SREs and management promise to not change anything"

But you can lock a database down to the point where only external auditors have access to the admin keys. It would make debugging challenging for sure, but that's anyway a tricky part of distributed systems such as bitcoin.

Selling illicit goods seems to have been transformed by Bitcoin.

A bit of contentious point but I want to distinguish between bitcoin blockchain and blockchain tech.

Even as a bitcion skeptic, its market cap went from $0 to $1T and you can't argue with that. But circa 2012 people started marketing blockchain as an independent technology that can solve all sort of problems. This is the bit I have issues with, there hasn't been a single successful use case sense.

This can be a good thing. Just because a government prohibits something doesn't mean it is bad.

(Yes, now we can devolve into a discussion about what "bad" means.)

Reply on news.ycombinator.com