[seanhunter]

You will probably consider these as "financial schemes", but having worked in finance for many years my impression is there are a number of problems in financial infrastructure that are extremely annoying and costly at the moment which could be much better addressed with blockchain based solutions. Blockchain is not the only way you could solve them, it just has the potential to be much better.

One example: Trade reconciliations. Lots of things (eg swaps contracts) that trade "over the counter" (not on an exchange) start off crazy bespoke but over time end up being largely standardized. So these days if you trade a swap, the contract is a standard contract called an "ISDA Master" with essentially a one-pager referencing the isda master filling in the blanks in the standard contract - what the legs and key numbers are, who is long vs short the legs etc. 10s of thousands of these are traded every day.

In order to settle these so can people actually start paying each other money, there needs to be a reconciliation (ie I need to agree with you on the contents of the one-pager so we both know exactly what we are both holding). Because the two sides don't trust each other (they are adversaries in the market), both will have entered the details of the trade in their own systems as they understood them and there will be errors and breaks and things are hard. Armies of people and systems are employed to solve this problem. If there was just a central database it would all be so much easier[1] and this whole reconciliation problem would go away, but how do things get into the central database in the first place and how do the participants trust that the record there is good? Well, you can see you would still have the same reconciliation problem.

Blockchain could resolve this by having a single definitive contract record stored in a trustless fashion. You could make it so if both sides put in the same details the trade is matched and otherwise both sides get the trade rejected until they can resolve. It becomes trivial because you have a database that can be shared without trusting any central party. There are trusted central parties (eg clearing houses) in these systems by the way, but they behave as another counterparty so they don't solve the matching and reconciliation problem described above. You still have to match the two potentially-conflicting versions of the real contract for the clearing house to novate (step into) it.

Now are there other ways to address this problem? Of course - they are doing that now. They just suck.

I'm sure other industries have similar issues around reconciliation between parties etc that would work the same - I'm just not personally familiar with them.

[1] It would also be far easier to report these trades to the CFTC btw so that's another pile of cost that could get taken out by such a solution.

[rojeee]

I worked for a blockchain company for 5 years and recently quit because I finally realised blockchain has no feasible real world uses.

What typically happens is that a company suggests they want blockchain to solve some problem, like trade reconciliations for example and gets together a groups of likeminded partners. After lots of deliberation they then build some blockchain thing. However it turns out that the blockchain solution is more expensive (hardware and devs), more difficult to understand (see upgrades comment below) and generally worse in most regards compared to a centralised solution.

Furthermore, there are difficult technical problems to overcome like versioning (eg flag days vs rolling upgrades, backward compatibility of protocol upgrades) which there are no good compromises to. Eg company A wants to upgrade but B doesn’t, so no one can upgrade. Because of this you can’t really use a single network so you need a federation of networks and this gets really complicated.

The other issue is around systems of record. Companies - of sufficient size - by law need to maintain financial and operation records and of course they manage the golden source for these records. Conventional wisdom is that blockchain can be the new golden source but it doesn’t work like that because what’s ends up needing to happen is that all the participants end up reconciling what’s in the blockchain to their own systems! The blockchain becomes another system! They need to do this because the blockchain can’t contain any private/enriched data for a particular company (eg company A reports using UK GAAP and company B reports using IFRS).

Some companies struggle on with the blockchain because there is a true believer sponsoring the project. They waste a lot of effort. I’ve seen companies Chuck away tens of millions and have nothing to show for it. Most companies end up ditching the blockchain and build a centralised solution using something like a clearing house model where all participants are owners and get a proportionate say on governance for the centralised entity.

Don’t get me wrong… There are uses for blockchain but they are typically infeasible to implement because the costs outweigh the benefits.

I was doing blockchain for 10 years before I gave up. I was really enthusiastic about it before but now I know that the enterprise space won’t go anywhere and the permissionless/web3 space will likely get regulated out of existence in due course.

[dragonwriter]

> but how do things get into the central database in the first place and how do the participants trust that the record there is good?

Either party enters them as a candidate record, including identifying the counterparty, and both acknowledge it.

Or one party creates and digitally signs a record, and sends it to the other, who digitally signs and records it in the central database that act also constituting acceptance.

Blockchain doesn't solve anything explicit in this scenario (it doesn't address lack of trust between parties better than a centralized system), what it arguably does is address lack of trust in the operator of the centralized system, if one can instead trust the aggregate of operators of the distributed system.

[seanhunter]

That's a good point that I didn't really address. Why can't you just have a centralised database?

1) If you had a centralised database, it would be extremely difficult to prevent the people operating the central database from exploiting the fact that they literally know how every bank in the financial system is positioned. It would be the juciest inside information ever.

You definitely need some kind of system where both sides can know they have agreed without anyone within the system being able to get access to all the facts globally in a timeframe where they would be able to take advantage of this knowledge.

2)Who would run it? The only feasible bodies that I could see would be the central clearing houses and they definitely haven't shown the level of operational competence that would be required given that if this thing went down pretty much all finance would stop until it was fixed. Banks use the swap rates to bootstrap their discount curves so if the swaps market disappeared for a few hours literally noone in the world would be able to accurately risk-manage any sort of trade that was dependent on future cashflows.

[jayd16]

So how does the chain verify ownership of the non-chain assets that are being traded?

Or does this not actually validate the transaction directly? Is this just about creating a trustless arbiter for market prices or something along those lines?

[seanhunter]

The chain doesn't, but that's another whole batch of problems associated with payment reconciliation (which is incredibly complex given netting, collateral posting etc) and you have to get the trade reconciliation problem solved first before you have any hope of getting payments reconciled. The chain here would just act as a record of the definitive facts of the agreement behind the trade making it really easy for both sides to verify whether or not the booking in their systems was correct.

What's being traded in a swap are just cashflows. An example of a swap would be for the next 5 years every month I pay you 4% interest on 1m USD and you pay me 3 month USD Libor + 250bps (interest on 1m USD). The facts that we need to agree are all of those things plus a couple of others that are a bit more technical but are important to make sure everything calculates correctly (various dates, the calendar basis to pro rate the payments correctly etc).