Blockchain Could Help Us Reclaim Control of Our Personal Data | Hacker News Reader

Blockchain Could Help Us Reclaim Control of Our Personal Data

hbr.org

46 pointsdoener19 comments

I have never understood how blockchain will solve the "I lost my key" problem. I also don't understand how it will solve the "once I gave them access to my data, company X saved a record in a traditional database and now that database was compromised, thereby leaking all my information" problem.

This article does not answer either of these fundamental questions.

I've never understood how blockchain will solve anything it's popularly claimed to be able to solve. Every week for years now we've heard "we're betting big on blockchain" and "blockchain will solve everything" and it's done... what? It's enabled... what? Other than enabling cryptocurrencies and a specific type of accounting ledger, what good has or will come out of the massive hype around blockchain?

I bet that if I wrote a bot to post randomly generated comments to HN in the vein of "I wonder if blockchain would be a good fit for this application?", I would be bathing in karma.

1. If in your use-case it's important to let the user recover from lost keys, you can have some multisignature recovery (I designate n trusted parties like friends, family, governments etc, if I lose my key they are allowed to together designate some new address as me, and I trust them to verify me offline if I lose my key; we can do m-of-n multisignature, letting me trade of risk of them colluding to falsely verify someone vs risk of them not verifying me; using secret-sharing we can do this off-chain too).

2. If my use-case requires me to give someone access to my data in order to do their job (eg a hospital needs to see my medical records), then I think it's mathematically impossible to defend against what you said. (barring unlikely advances in homomorphic encryption and the like). Using public-key crypto allows us to eliminate having a intermediary party and the risk that the intermediary party gets hacked. Blockchain then allows us to credibly prove that our data existed at a certain point in time. Also there are use-cases where I maybe need not disclose all my data (eg maybe for some use-case my bank does not need to have access to my payslip, just a proof signed by my employer that it exists)

On #1: I hear you and I think this is a totally valid approach for a society where we don't have little old ladies and gents who lose the piece of paper on which they've written their key and all their friends are old and they urgently need their secure key so that the doctor can access their health records. Or the kid who comes into the hospital in a coma and you can't get their secure key. Or for the recent immigrant family with no friends and their house burns down with all their secure keys inside. Or for the gay kid who gets kicked out of his house without taking his key and the people that blockchain "trusts" are his parents who just threatened to kill him if he comes home.

The "use case" here is people's lives. People's lives are messy. Blockchain doesn't solve that mess, it pretty much just makes it worse.

We should focus on using blockchain to solve "neater" problems like who owns what and who is contractually obligated to do what, etc. Those problems don't have edge cases that involve a parent who loses their memory or a kid who gets thrown out of his house.

I'm intersted in knowing how the revocation of access to data at a later time could be possible. Also it seems likely that any organization you gave access to could simple store and sell the data.

The biggest problem I can see from the article is that once someone loses their key their identity is lost forever and they must rebuild the data. The first few times that happens the public will be turned off to the idea.

It's very difficult, arguably impossible, to revoke access to data that's already been accessed. You could try a DRM-type scheme but there's always ways to get around it.

What you could do, pretty easily, is revoke access to the latest data that hasn't already been shared, and to have an audit trail of exactly what data has been accessed and when it was accessed. That would still be a massive improvement on the status quo.

I wouldn't rule out the eventual possibility of something like Facebook migrating their Open Graph API to some type of encrypted blockchain. Although I would be completely shocked if it were more than essentially a publicity tactic that wasn't actually using any true decentralization.

For medical records - consider a secure reader tablet-like device. It has encrypted hardware which can be temporarily permissioned to have access to encrypted data based on blockchain timestamps. It is tamperproof - if you open it it has some self-destruct mechanism. Data is accessed in chunks, therefore the owner of the data can see whether data is being accessed in the way a human might use it, or if it is being downloaded en masse. The organisation with data access devices gets audited on an annual basis to see where all their access devices are and to ensure none of them are being tampered with.

Suddenly, we get to a situation where the security of the medical file begins to resemble that of the paper-based medical records department. If you want to copy records, you need to physically go to where they are, get access to each individual file, and go page-by-page photographing each.

This is actually a genius application. I hope more HN readers read it and this meme spreads.

The improvement here is this:

a secure reader tablet-like device. It has encrypted hardware which can be temporarily permissioned to have access to encrypted data

This part is how it works today and it has been like this for years:

Data is accessed in chunks, therefore the owner of the data can see whether data is being accessed in the way a human might use it, or if it is being downloaded en masse. The organisation with data access devices gets audited on an annual basis to see where all their access devices are and to ensure none of them are being tampered with.

No blockchain is needed to achieve this.

it is if you are working across multiple jurisdictions/regimes and you don't want a centralized point of failure.

There isn’t a single DRM scheme that cannot be technically broken even it its with pen and paper.

DRM is only as strong as the legal mechanism behind it.

Since the DMCA and other legislation protects DRM the actual strength of the DRM are irrelevant.

DVDs are still protected by the same broken key, BR are also broken every DRM system gets broken or is easily circumvented through side channels.

Maybe by ensuring all data access/decryption happens from the client side perhaps, so that the data never flows(unencrpyted) through corporate servers.

Like FB implemented as Signal on steroids, when I add a friend I exchange keys to open up a channel with him to enable him to see posts/photos etc from my life, but the org (say FB) itself cannot see any of this data.

Not sure how blockchain will fit in here though or enhance this

blockchain would allow: - trustless cloud-based indexed storage of encrypted data using protocols such as IPFS - logging of data access requests to a globally accessible database and ability to give and remove permissions based on timestamps - auditing of data accessors to ensure that their access patterns match their functions - eg. if a hospital suddenly starts trying to download data on a million patients instead of 300 then this would be visible on the blockchain and individuals could then question why this is occurring and prevent access.

All of these things can be done with trusted intermediaries, but when you start considering use cases such as - being in china and needing to allow a chinese hospital to access your records back in the USA, suddenly these things become important.

That’s a lot of buzzwords but it fails to answer how any of it would work, why you’d want that, or why it’s better than mature alternatives.

For example, how are you defining trustless? Personal data systems needs to be based on trust and reputation (otherwise you’ll have a bunch of 4chaners spamming claims about you) so are you simply thinking tamper-evidence and non-repudiation? That’d raise the questions of why this would be better than PKI, Merkel trees, etc.

All of the other claims hit similar problems: e.g. you can’t store the data in a blockchain and make any of those promises, and it’s not clear how useful a shared index or log would be when you can’t trust all of the participants.

Another fatal flaw is the lost of control and inability to correct mistakes. If there’s ever a vulnerability all of that data is irrecoverably public. What organization is going to turn control of something they get sued for over to an entity they don’t control? “They told us the software was perfect” doesn’t seem like a good HIPAA defense tactic to me.

Blockchain, solution looking for a real problem

So what they really mean is encryption could help us control our data

If your data is public you’ve lost control over it by defintion.

Some one can make a copy, game over.

anything that comes out is not personal anymore.

Reply on news.ycombinator.com