Hacker News Reader

Comment by rglullis | Hacker News Reader

As another pointed out: supply chain does make sense.

There are plenty of cases where it's better to not have to trust on any particular entity. Examples:

- Marketing and advertisement campaigns. How much ad fraud exist and how much money is spent by advertisers without knowing if they are actually reaching their customers. Do you trust Google to actually mitigate these problems?

- Health care: remember Google Health and Microsoft Health Vault? Nowadays Apple/Google/Samsung wants to collect that data. Wouldn't you rather have it in a way that only doctors had access to it?

View on HN

I don't understand the supply chain use case. You can't trust the third party to enter "shipping container #123 has 10,000 bananas" into a centralized database, but you can trust them to enter "shipping container #123 has 10,000 bananas" into a blockchain? They can still lie about what they put into the blockchain, it isn't magic.

Of course they can lie, but with blockchain it is much easier to detect the lie. If the banana producer has a receipt from the shipping company that says "10,000 bananas in the container for Jasper" and the shipping company only has 8,000 to put at your door, you can be sure it who was malicious.

Without a blockchain, if the shipping company and banana producers were colluding to skim you of 2,000 bananas, it's much easier for them to do.

If the shipping company and producer are colluding, blockchain does nothing to stop that.

No, but if the producer has a record saying they put 10 bananas and the shipping company only delivers 8, it will be undoubtedly on the shipping company to bear the costs of the difference.

OK. But what if the producer said they put 10 bananas, but they lied and actually put 8? When it finally gets to the recipient, nobody knows whether the shipping company stole them or whether the producer lied. The blockchain did not help.

Come on, it's not that hard...

I mention about the producer saying they have a receipt indicating 10 bananas. Whoever produced that receipt signed a message saying "I got 10 bananas from the producer" is now on record responsible for it.

So why wouldn't a central database work, with the producer entering how much they sent to the shipping company, and the shipping company entering how much they sent?

What in the world does the use of blockchain enable in this case?

Who controls the "central database"? What if the manager of the central database is colluding with the shipping company, or the producer?

What if the manager of the central database wants to block one of the producers to favor another one?

"What does the blockchain enable in this case?"

To get rid of the central database and all the potential sources of conflict?

Imagine like you trust edges but not middleman.

It doesn't eliminate that. But it gathers evidence. Where did they acquire the 10,000 bananas? Multiple parties must collude -- and then that collusion is on record.

I'm interested in if that data is useful to anyone?

Also, why not trust a central entity with that data?

>Also, why not trust a central entity with that data?

I.E. blockchain does nothing to solve this potential issue

> Wouldn't you rather have it in a way that only doctors had access to it?

How is a blockchain the answer here? The "distribute all data" and "keep history forever" attributes seem actively bad for healthcare use cases.

That is a good question. The answer is that you don't put the data in the blockchain, you just keep the log of who is interacting with the patient and who needs access to the records. ;)

Even which doctor someone visited is confidential data one ideally would not preserve when not needed anymore (and "ideally" might very well be law). Health data is messy, and I have trouble imagining a solution that puts it on a blockchain but allows only private access to everything and has a chance of fixing instances of leaked credentials while still getting value from the fact that there is a blockchain. (If there is prior work on this that actually goes into detail I'd be happy about pointers - sadly a lot of this stuff seems to end at the stage of whitepapers when it comes to public information)

>Even which doctor someone visited is confidential data one ideally would not preserve when not needed anymore.

Indeed, for example "John Smith went to see a HIV specialist on Monday" is protected information and would be a HIPAA violation if leaked. Its much simpler to throw it in a relational database and only let the HIV specialist and staff see the appointment.

https://brave.com/themis/ should be a good pointer about how to get an advertisement network that can avoid fraud and report views accurately while preserving privacy. The same principles could be applied to a permissioned blockchain that could be used by physicians and healthcare providers.

A public immutable and distributed log saying somebody sees various psychology specialists, or disease specialists over the course of a few years... this metadata is almost as personal as the raw data in many cases.

I am almost surprised that companies don't get into more trouble over location tracking. Your smartphone knows exactly when and how often you see a doctor.

All the doctors websites around here, including for specialists, use google fonts and google maps etc. I assume there is some metadata involved there too

I guess HN's pedantism will make me write a whole paper about what was supposed to be a simple example...

Anyway, both the example of advertisement and health care data need to also provide privacy. In both cases the data should be considered sensitive and not public. In both cases, the data can be protected through blind signatures.

[0]: https://en.wikipedia.org/wiki/Blind_signature

Reply on news.ycombinator.com