4 ms·

I have an extremely low opinion of blockchain/crypto, but in the interest of fairness, I did read an FT piece a few months ago about how banks are apparently ac

I have an extremely low opinion of blockchain/crypto, but in the interest of fairness, I did read an FT piece a few months ago about how banks are apparently actually using the blockchain for currency trading: "HSBC and Wells Fargo are cutting out a key part of the currency market’s infrastructure from some trades after the two banks agreed to settle transactions directly on blockchain technology. From Monday, they will use blockchain technology to reconcile and pay out on deals in dollars, sterling, euro and Canadian dollars between the two banks, using HSBC’s FX Everywhere platform. The agreement means they will bypass CLS, the nearly two-decades-old utility that central banks urge market participants to use to neutralise the risk of certain trade failures" https://www.ft.com/content/1a4dcaf5-2b4b-4f0b-8c58-a8fa173f24b3 https://www.ft.com/content/1a4dcaf5-2b4b-4f0b-8c58-a8fa173f2... https://archive.ph/qs8Cg https://archive.ph/qs8Cg non-paywalled link

simonw 4y ago
That sounds to me like someone used excitement about "blockchain" as an excuse to upgrade a two-decades old system to something a bit more modern. It looks like it's a private blockchain shared between two (at the moment) institutions, which means it's probably being used basically as an append-only log.
- civilized 4y ago
  Reminds me of the craze for using Hadoop on data that could easily be handled by basic R, Python, or even Excel https://www.chrisstucchio.com/blog/2013/hadoop_hatred.html https://www.chrisstucchio.com/blog/2013/hadoop_hatred.html In hindsight, the Big Data hype wave was really just a good excuse to bring in a wider set of analytics tools than Excel, Access and inflexible, IT-administered databases.
- matthewdgreen 4y ago
  A blockchain is an append-only log. Upgrading ancient technology is a huge use-case for blockchain tech. The advantage of this platform is that settlement is probably much faster and it will presumably be relatively easy to add more than two parties to the system, without running into gnarly trust issues.
- no_circuit 4y ago
  You'd be right if you thought that just using for an append-only log doesn't seem like a big difference. But the key thing to note is that they said it could replace CLS, the current settlement system which makes sure the net direction of monies is paid out. So I believe this to mean that instead of using CLS to settle up with currency transfers like USD or EUR, the banks agreed for certain subset of trades they are willing to agree on how to value Bitcoin, or some other crypto, versus the currencies they trade, and then settle the trades by periodic, usually daily, Bitcoin transfers. If Bitcoin's value isn't stable, then it would be risky for these banks to hold a lot of it in order to settle trades. Perhaps this is what would sink this usage of the blockchain, but probably only a quant would be able to answer that. I did find it somewhat amusing that the article claims a benefit of real-time visibility of settlement status. I believe some settlement systems still work by sending csv files over ftp (+ssh / sftp).
PragmaticPulp 4y ago
> how banks are apparently actually using the blockchain for currency trading: There isn’t a single blockchain that we can call “the blockchain” Technically, a private append-only log is a blockchain if it includes some hashing in each entry to verify the previous entry. This creates a chain of blocks that can be verified to have occurred in a certain order without tampering, though it wouldn’t have the features of a distributed blockchain like cryptocurrencies. This is actually fine for a lot of applications. It’s especially fine if you get a vague business direction to “use blockchain somewhere” and you need to check that box for PR/marketing/investor-pleasing purposes.
MBCook 4y ago
Why do two institutions need a blockchain for transacting only between themselves? They’re banks. They have lots of agreements and lawyers. They already trust eachother. And there’s no 3rd party involved. I’d be surprised if there was an actual blockchain involved instead of spinning a simple log as a blockchain for hype.
- matthewdgreen 4y ago
  Presumably whatever 1970s tech they were using previously was creaky and needed to be replaced. It probably had long settlement times. You’re correct that if two extremely trusted parties are going to handle all settlement, you can just have one party run a centralized database. But this gets more challenging when you expand from one party to dozens, and the trust relationships get less clear. Since you’re building a new platform anyway, you might as well build a database that’s replicated and decentralized and doesn’t require trusting any single party for availability and correctness, at which point you’ll either use a blockchain or something functionally equivalent.

simonw 4y ago

That sounds to me like someone used excitement about "blockchain" as an excuse to upgrade a two-decades old system to something a bit more modern. It looks like it's a private blockchain shared between two (at the moment) institutions, which means it's probably being used basically as an append-only log.

civilized 4y ago
Reminds me of the craze for using Hadoop on data that could easily be handled by basic R, Python, or even Excel https://www.chrisstucchio.com/blog/2013/hadoop_hatred.html https://www.chrisstucchio.com/blog/2013/hadoop_hatred.html In hindsight, the Big Data hype wave was really just a good excuse to bring in a wider set of analytics tools than Excel, Access and inflexible, IT-administered databases.
matthewdgreen 4y ago
A blockchain is an append-only log. Upgrading ancient technology is a huge use-case for blockchain tech. The advantage of this platform is that settlement is probably much faster and it will presumably be relatively easy to add more than two parties to the system, without running into gnarly trust issues.
no_circuit 4y ago
You'd be right if you thought that just using for an append-only log doesn't seem like a big difference. But the key thing to note is that they said it could replace CLS, the current settlement system which makes sure the net direction of monies is paid out. So I believe this to mean that instead of using CLS to settle up with currency transfers like USD or EUR, the banks agreed for certain subset of trades they are willing to agree on how to value Bitcoin, or some other crypto, versus the currencies they trade, and then settle the trades by periodic, usually daily, Bitcoin transfers. If Bitcoin's value isn't stable, then it would be risky for these banks to hold a lot of it in order to settle trades. Perhaps this is what would sink this usage of the blockchain, but probably only a quant would be able to answer that. I did find it somewhat amusing that the article claims a benefit of real-time visibility of settlement status. I believe some settlement systems still work by sending csv files over ftp (+ssh / sftp).

PragmaticPulp 4y ago

> how banks are apparently actually using the blockchain for currency trading: There isn’t a single blockchain that we can call “the blockchain” Technically, a private append-only log is a blockchain if it includes some hashing in each entry to verify the previous entry. This creates a chain of blocks that can be verified to have occurred in a certain order without tampering, though it wouldn’t have the features of a distributed blockchain like cryptocurrencies. This is actually fine for a lot of applications. It’s especially fine if you get a vague business direction to “use blockchain somewhere” and you need to check that box for PR/marketing/investor-pleasing purposes.

MBCook 4y ago

Why do two institutions need a blockchain for transacting only between themselves? They’re banks. They have lots of agreements and lawyers. They already trust eachother. And there’s no 3rd party involved. I’d be surprised if there was an actual blockchain involved instead of spinning a simple log as a blockchain for hype.

matthewdgreen 4y ago
Presumably whatever 1970s tech they were using previously was creaky and needed to be replaced. It probably had long settlement times. You’re correct that if two extremely trusted parties are going to handle all settlement, you can just have one party run a centralized database. But this gets more challenging when you expand from one party to dozens, and the trust relationships get less clear. Since you’re building a new platform anyway, you might as well build a database that’s replicated and decentralized and doesn’t require trusting any single party for availability and correctness, at which point you’ll either use a blockchain or something functionally equivalent.