It’s very interesting as a matter of pure technology. I’m just not sure it’s that useful as a banking system.
---
Medical Records
Mid aughts, I was trying to figure out how to use tamper evident logs (rolling hashes) for our electronic medical records infrastructure. To answer questions like "what did you know and when did you know it?" and "where did this bad data come from?"
Coupled with the notions for Translucent Databases (all data at rest is encrypted) to protect privacy.
We hadn't even gotten to the consensus and trust issues.
---
Authenticity
I'm now most interested in verification of sources.
Where did this cotton or wheat or cobalt come from? Are these pills authentic?
Is the data cited by this paper legit?
Verify copyrights, perhaps even a realization of Ted Nelson Xanadu's notions of transcopyrights. (Disney's DragonChain and others have floated those use cases.)
Is this news report real or agitprop? Is this quote actually legit?
Did someone(s) silently update some published information?
Etc, etc.
---
Tursted Rules & Constraints Satisfaction
Since seeing a presentation on using ethereum style smart contracts for medical claims processing, I've been reading up.
I can barely wrap my head around this stuff. Computability, Turing complete, N vs NP, etc. But the domain experts I heard told a great narrative, so now I'm interested too.
And even if you manage to solve that problem, the blockchain still doesn't help with medical records, authenticity or anything else. You still have to trust the people entering those records. you still have to trust that the person who said "those pills are authentic" on the blockchain.
Smart contracts are also the biggest pile of hype. You can't have "code is law" because other wise you wind up with The DAO.... All code has bugs. All code has corner cases where assumptions fall apart. You can't code law.
Blockchain is pure, 100% hype. Almost every single use case falls apart under even a small amount of scrutiny. The only use the blockchain serves is as a platform for fraud. If you are a scammer, there is no better place to be than in crypto.
It's turtles all the way down, right?
None of my proposed use cases rely on PoW, bitcoin.
Perhaps I should have said it more clearly: I'm not interested in currency et al, because best as I can tell, bitcoin is just a better escrow, or worse a collectable (like baseball cards). And those things don't interest me.
--
Have you done accounting, bookkeeping, auditing? Another way to think about blockchains is an interoperable ledger. If your data doesn't leave your org, you can just using tamper evident logs (rolling hashes), no consensus stuff required.
I worked on election integrity issues for a decade. I'm the biggest technoskeptic. I pretty much oppose all things new until they're battle proven.
The biggest challenge for election administration is verifying and publishing the physical chain of custody of all the gear and materials. And no matter how hard you try, there's always some yahoos (flat earther types) that will demand ever greater proof.
Whaddya gonna do?
Well, based on my experience, applying some blockchain technologies to document and publish chain of custody would simplify A LOT of the public accountability and transparency issues.
At the end of the day, it's just double ledger accounting. Debits match credits. The parts of blockchain that interest me make it easier to "show your work".
https://www.cis.upenn.edu/~ahae/papers/peerreview-tr2.pdf
https://eprint.iacr.org/2005/002.pdf
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.625...
And here's you a blockchain-based solution to medical sharing since you said you were looking into such things.
Those papers are from about the same time frame, but I don't immediately recognize them. I do remember the implementation I modeled mine after was patent encumbered.
I may have adopted "tamper evident logging" because I was talking to a lot of non-geeks at the time and they were familiar with terms like "tamper evident seals".
That MeDShare paper is kinda weird. We were very fortunate to have domain experts (nurses, doctors, admins) on staff. So what we ended up with (after a few years) didn't look much like what MeDShare proposes. For example, access control is a non-starter, because in an emergency no one cares about permission. So the best we could do is log access. Also, demographic data at rest is plaintext, because in the USA there's no GUID, so you need access to all the fields to do record matching (linking) and mitigate poor data quality.
Repeating myself: my interest in tamper evident logging was for legal liability and improving our own QA/test.
I haven't thought much about all the other areas blockchains could be used in medicine, like prescriptions.
But if there's some meat on those bones, I wanna know about it. I think of it as a constraints satisfaction problem. If the "vocabulary" can be kept simple enough, where our current theorem provers are sufficient, that could be cool.