I am massively sceptical of any claims that Blockchain can be useful. I even took the Certified Blockchain Professional course so I could better understand the batshittery.
At the risk of appearing on QTWTAIN, and seeking a papal indulgence from Betteridge, I think I might have stumbled on a retroactive use-case for Blockchain!
First up, what is LIBOR and what was the scandal which led to several bankers going to prison?
Here's the over-simplified explanation. The London Inter-Bank Offered Rate was a financial measure. Banks can borrow money from each other. They charge each other interest for this. These inter-bank interest rates are important - they reflect the financial health of the institutions and are used for all sorts of derived financial products.
Banks reported those LIBOR rates and, as a consequence, financial stuff happened. Those reports were sometimes lies.
The fraud investigation was long and complex. But it boils down to this - sometimes it was advantageous for the bank to lie about its rate. On at least 257 occasions, bankers would collude with each other to manipulate the rate.
For example, on 26 October 2006, an external trader made a request for a lower three month US dollar LIBOR submission. The external trader stated in an email to Trader G at Barclays “If it comes in unchanged I’m a dead man”. Trader G responded that he would “have a chat”.
Barclays’ submission on that day for three month US dollar LIBOR was half a basis point lower than the day before, rather than being unchanged. The external trader thanked Trader G for Barclays’ LIBOR submission later that day: “Dude. I owe you big time! Come over one day after work and I’m opening a bottle of Bollinger”.
So, would a Blockchain have prevented this sort of fraud? I'm going to go with a very cautious "maybe".
Let's take a look at the seminal Blockchain paper from NIST which contains this flowchart:
Taking the points in order:
- Do you need a shared, consistent data store?
- Yes. This data is read by multiple parties. It needs to be consistent - that is is Bank A lends at 1.23%, Bank B should record borrowing at 1.23%.
- Does more than one entity need to contribute data?
- Yes. Multiple parties will need to record their lending.
- Data records, once written, are never updated or deleted?
- Yes. These data are immutable. The caveat being that, as with all Blockchain projects, preventing fat-finger errors is difficult. But, with suitable UI work, it might be possible to prevent 1.23% being recorded as 123%.
- Sensitive identifiers WILL NOT be written to the data store?
- Yes. There's no GDPR for financial institutions. As these data are meant to be public, there should be nothing private or sensitive on there.
- Are the entities with write access having a hard time deciding who should be in control of the data store?
- No one institution should be in control of these data - that would be a recipe for corruption.
- Do you want a tamperproof log of all writes to the data store?
- Yes. Once borrowing is agreed, it should be recorded irrevocably. No manipulation or rewriting of events.
So, that looks pretty positive, right? A multi-stakeholder, decentralised ledger, which accurately records cross-party transaction, in an incorruptible and auditable manner.
The end of LIBOR came in 2024. It was replaced by the delightfully named "SONIA" (Sterling Overnight Index Average).
Did the financial institutions end up using a Blockchain? Of course not!
The way SONIA works is by having the data reported directly to the Bank of England. The Bank then checks the data for plausibility and errors, calculates the rate, and then publishes it.
I don't know how easy it would be for bankers to lie in their returns to the Bank of England - but off-chain fraud is also possible. Similarly, perhaps it is possible to bribe whoever collates and publishes the data.
But the point here is that this is close to textbook case of where a Blockchain could have been useful - and they rejected it.
The hunt for a real-world use-case for Blockchain continues in vain.
12 thoughts on “Would a Blockchain have prevented LIBOR fraud?”
Ian Mansfield
Equally sceptical, I did come across a situation where a blockchain-style database could be a useful interim product in countries with low trust in government.
https://www.economist.com/business/2017/06/01/governments-may-be-big-backers-of-the-blockchain
The idea being in situations where there is unreliable institutional information and poor governance, open blockchains might offer the necessary level of trust to make the data trustworthy.
Obvs, the ideal long term goal to build up the trust that it can eventually switch to a more conventional record keeping method.
The key problem with LIBOR was that it was not based on actual transactions, but hypotheticals (if you had to borrow overnight, what would you be charged?). Recording the submissions might have helped spot the fixing but wouldn’t have addressed the underlying issue. The replacement, SONIA, is based on actual transactions and administered by the Bank of England. They are acting as a trusted third party.
@Edent Considering that blockchain technologies seem to be constantly involved in fraud, I would expect the answer to be "No, but it might have changed how it was conducted"
| Reply to original comment on toot.cat
@Edent great memories of the time blockchain was the big hype are coming back, but apparently cryptocurrencies are still the only practical application, i always thought blockchain for software control would be used but apparently not (yet) , mdpi.com/2076-3417/13/5/3066
| Reply to original comment on mastodon.social
To all our great relief, I think this too is a QTWAIN.
As I understand it (which is limited), the problem was not that there was any actual lying about transactions which had taken place or any dispute about them.
| Reply to original comment on bsky.app
@ErikJonker @Edent There is not/cannot be a practical use case for blockchains beyond digital currencies.
Yes, there are use cases for tamperproof history. Transparency logs exist for that reason, even in productized form (e.g. immudb). Sometimes you need distributed consensus, there are algorithms for that, and then you put the results in the log.
Practically *never* do you need all of these: distributed consensus over history among non-cooperating nodes without an admissions authority.
| Reply to original comment on chaos.social
@ErikJonker @Edent The core value proposition would be automatic enforcement of rules (via consensus) among mutual distrusting nodes. BUT: By definition this application cannot interact with "the real world", because then you'd need a cyber/physical interface. Such as a node that is trusted to accurately report sensor readings.
The only application that can work *at all* is one that takes place purely within the blockchain itself. And you need an incentive. That pretty much defines a currency.
| Reply to original comment on chaos.social
@blog > Are the entities with write access having a hard time deciding who should be in control of the data store?
> No one institution should be in control of these data - that would be a recipe for corruption.
I think the BoE is a valid candidate here, though. Seems like everyone agreed (or didn’t have the clout to disagree strongly enough) on that, at least.
This doesn’t negate your article. I expect there’s nothing that will check all the boxes, and I really appreciate your breakdown.
| Reply to original comment on mastodon.functional.computer
Would Certificate Transparency Logs count as a practical use of blockchain?
@Edent it has the universal problem that someone has to take data from the real world and type it into a computer, and they could lie
| Reply to original comment on hachyderm.io
@Edent Yeah, I keep looking but the blockchain concept remains a solution in search of a problem. I keep trying to find novel problems that might be applicable but almost every time there's a better, easier solution. (Generally assuming you trust the current PKI infrastructure)
I keep coming back to one of the base issues is that so many ideas are trying to replicate something in the real world and that interface is never perfect. If you can lie at the input stage, you now have an immutable instance if that lie.
| Reply to original comment on mastodon.infrageeks.social
@Edent Speaking as someone who uses a blockchain every day I see a lot of entrenched opinions on both sides. As I understand it they can provide public storage that does not require you to trust someone to store it indefinitely whilst having traceability of transactions. Proving the input data is genuine is another issue. There are plenty of issues around who gets to validate transactions etc, but various chains have come up with different solutions. They are by no means a universal solution.
| Reply to original comment on mastodon.org.uk
More comments on Mastodon.