HN should just add this comment by default to every crypto discussion. It is parroted practically every time people get offended that others aren't convinced by their arguments.
Which side is not listening to other's arguments? I don't see any argument against crypto in general, except "I don't believe" or "I don't think it solves something I know of". These are not arguments, these are also not refutable.
As a practitioner of finance for 15 years, from front to back, vanilla to exotic, tech to investment, I've spent hours on HN to explain that yes, there are tremendous advantages to blockchain in finance.
Yet by the very definition of 1 user = 1 vote, it just takes a 22yo web developer to think "meh, no, it's just scams".
HN is good for a lot of stuff, but when it comes to something that is beyond the understanding of the average developer, there is not point for a few specialists trying to convince a mass of nay sayers.
Huh? I've been on HN for a long time and I don't remember it ever being pro-crypto. People got excited about BTC and ETH in the old days but HN was one of the few places on the internet where people were always sceptical of the crypto hype.
Also I remember people complaining about wash trading since forever.
Edit: I remember people posting literal ponzis a decade ago. Like sites titled “Bitcoin Ponzi” or similar. Or Satoshi Dice—can’t remember if that was posted here.
Interesting. This is an appeal to authority - because as a 'practitioner of finance' we should just trust you.
Two problems.
1) 'practitioner of finance' is a meaningless title. Are you an accountant? A financial advisor? A FP&A? A CFO?
2) Regardless of your title, you should be able to simply and succinctly provide an use-case for crypto. Instead, you suggest that after hours of work, even developers (who specialize in understanding abstract and complex problems, especially in regard to computation and its applications) cannot understand what you're selling.
Your current approach does not help install any confidence in crypto, instead it reinforces the notion that crypto supporters don't understand their own product.
It is, indeed, an appeal to authority, I have no issue admitting that.
> 1) 'practitioner of finance' is a meaningless title. Are you an accountant? A financial advisor? A FP&A? A CFO?
I worked on various fields of finance, at different levels of seniority.
I spent some years working on pricing models of vanilla and light exotic instruments, which gives me a good understanding of the challenges of defining, pricing, and assessing the risk of derivatives contracts.
I worked on realtime market feeds for execution algorithms, so I have a very good idea of the challenges of market impact, order flow, and the tradeoffs of various matching engines.
I worked in quantitative R&D in a major B$ hedge fund at the partner level. I was responsible for a big part of alpha creation, portfolio construction & optimization, operational and investment risk. I later had exposure to most aspects of the fund on both cash and derivatives, listed and OTC, on NA, EMEA and APAC, from back to front office.
I have multiple regulated responsibilities to the SFC which also gives me a good understanding of the compliance and regulatory framework around investment vehicles. I had a long exposure to implementing various European directives related to cash management, settlement, clearing and risk management under mifid 2.
Recently I have been a founding partner and director of a prop trading firm.
> 2) Regardless of your title, you should be able to simply and succinctly provide an use-case for crypto.
- Ever heard that "cache coherence & invalidation is the hardest thing in computer science"? Well in finance it's called "matching & settlement", and it's pretty much the same problem. I would estimate that around 50% of the resources of a financial actor is consumed in back/middle office tasks, which is basically making sure that everything is matched and settled properly, and if not, manually and painfully unwind the later transactions that depend on it. A blockchain is settled "by design", the whole settlement problem just vanishes. It's rather complicated to explain exactly what are the ins and out of front-middle-back office in this already long comment, but we can dive into that more specifically if you wish.
- Cash management & custody is a big thing as well. There's a whole industry of custodians whose job is just to hold collateral (cash or GBs) so that brokers can create a credit line for you. Blockchain can allow self custody, staking, locking and proof of reserve which makes custodians useless (except for 3rd party risk mgmt, but then it becomes an investor decision).
- There are thousands of thousands of derivative contracts in the wild. All major companies are constantly entering bespoke/OTC swaps to hedge against various factors (currency, commodity, interest rates, etc). Dealing, pricing and risk management of these contracts is a nightmare, they all have their own specificities, oddities, wordings, pricing, etc. Each OTC desk in investment banks have tons of quants and structurers dedicated to creating and maintaining these contracts. Being able to express these as smart contracts instead of paper legalities would drastically simplify the processing, issuance, management and payment of those.
- Transparency. A lot of things in finance is based on mandatory reporting by various entities (interbank interest rate, OIS rates, locate rates, trade reporting, etc). These reporting are always made in a bespoke manner, through ad-hoc channels and softwares. It's very hard to trust, corroborate and get a timely delivery of those things (not to mention all the scandals of forged reporting we had in the past). Blockchain could provide transparency and uniformity and trust on these.
I think these are the main aspects where blockchain could bring a tremendous impact of current finance, from my experience at least.
Now I also think blockchain has a lot to bring in other fields, where I'm no expert but have an opinion.
- Securitization: the overall concept is to allow for OTC (peer to peer) exchange of ownership, without a third party to bless the transaction, while still exposing a public record of ownership. Basically that's what NFTs do. Say I buy a photograph from an artist. Usually I will receive a proof of authenticity in the form of a signed paper by the artist. Whenever I want to sell that photograph, I will have to also give my proof of authenticity to the new buyer. I think this is inefficient in many ways. First, the artist will have to pay a lawyer/notary to draft that proof of authenticity, which (friends told me) is expensive and annoying. Second, I will have myself to keep that piece of paper and not loose it. Third, the new buyer will have to get this proof of authenticity verified by the artist or law firm/notary. NFTs are basically a solution to this. The artist can emit a token representing my photograph, and have me and him sign that token. Whenever I resell the photograph, I can unilaterally cede ownership of the token to someone else whenever I please. The signature of the artist ensures the buyers that my ownership is legit, no need for third party verification. This could be done with just cryptographic signatures of course, but the blockchain here allows for the whole life cycle and transfer of these tokens to happen in a public and auditable/verifiable manner.
- Removal of middle men: I think a lot of companies nowadays are glorified databases. Their only added value is that they perform control on modification of said database. Blockchain can eliminate these commission based middlemen.
> Instead, you suggest that after hours of work, even developers (who specialize in understanding abstract and complex problems, especially in regard to computation and its applications) cannot understand what you're selling.
It's not a problem of "cannot understand", it's more an issue of 1) attention span 2) lack of knowledge.
For 1) I think the typical HN reader just skims articles on the front page, up/downvote some comments, and forget about it once it's not on the front page. There is almost zero chance that an interesting discussion or answer, that takes time to write/read, will be able to catch-up the downvotes. This very response that took me 1h to write will probably be read by very few people, while the OP answer that got downvoted to hell won't budge.
As for 2), unfortunately, I'm of the opinion that we live in a generation of armchair Wikipedia experts. A lot of people just don't know the amount of things that they don't know, so they assume they understand finance because they've seen the big short, have an interactive brokers account, and watch Cofeezilla.
I think the OP message makes a lot of sense, those are all valid points, to me, as a practitioner.
How can I not be annoyed when I see that his post is downvoted to hell, and the 2 top answers to it are:
> Speculation is not a use case.
> The examples you gave are problems created by blockchains.
Clearly, to answer something like that, you would obviously have to not understand blockchain, and have no idea how finance is actually working.
I'm not sure blockchain is the right solution though.
For example, take NFT's where you suggest NFT's can remove the need for signed paper and proof of authenticity. Problem is, NFT's don't actually do this.
NFT's are not legally binding contracts - so you still need a lawyer to create a contract. NFT's don't prove ownership of anything beyond the NFT, so you still need that artist to prove that they created said NFT for said artwork. NFT's don't prevent fraud, or theft etc - so you still need experts to authenticate the art if you wish to do a transaction.
When you look closely, it's hard to see NFT's actually solving any problems. Instead, they just insert themselves into the picture to make their proponents money, while introducing an entire range of new problems that now additionally need to be solved.
This applies to more than NFT's though, one of your earlier statements caught my eye:
"which is basically making sure that everything is matched and settled properly, and if not, manually and painfully unwind the later transactions that depend on it."
You think that's bad today, wait until you need to do this on a public blockchain that has no built-in mechanism to unwind mistakes!
At the end of the day, when I look at the problems you raised this is what I see:
1) If all actors could agree upon a common set of API's. 2) And a trusted datasource. 3) And would modernize all their contracts. 4) And would use the above to automate many tasks.
Then things could be way simpler.
And you're probably right!
Question is - why use blockchain? Nothing here is technically difficult with existing technology. The problem is a business one - figuring out these specifications and getting the industry to adopt them. Blockchain can't help with that as it's purely technical.
Right but certificates of authenticity do not prevent fraud, theft, or illegal copies either.
I have limited knowledge of the "art world", so don't take my word for it, I'm just reporting discussions that I had with friends more knowledgeable on the subject.
My understanding is that nobody really cares if you have a copy of a photograph or digital art. What people care about is if you pretend to own it.
> you still need that artist to prove that they created said NFT for said artwork.
Well no, the NFT contains the signature of the artist that created it. That is, at a basic level, what an NFT is: a small token (often in the form of a json string) that is signed by a creator. This token can then be transfered by the current owner to others, and such transfers are on the blockchain.
Anyone can check that an NFT was issued by a legitimate artist by checking the signature of the artist on it. Anyone can see who is the current owner by following transfers of said token.
> You think that's bad today, wait until you need to do this on a public blockchain that has no built-in mechanism to unwind mistakes!
There is nothing to unwind on a blockchain, because you cannot create incoherent ledgers. That is why blockchains are perfect settlement systems.
Imagine the following typical simple settlement scenario:
1) You send USD to your brokerage account, willing to immediately spend it on some stocks. You do that through your credit card, which has a 3 day redemption possibility o such transfers.
2) Your broker does not want to prevent you to use that money for 3 days, because that would piss you off. So he "advances" the money to you, expecting that you won't cancel your credit card transaction.
3) You buy a share of Microsoft and see it appear in your portfolio. You sell it 2h later for a small profit. You probably don't realize it, but equity settlement will take at least 24h, so you did not really bought that Microsoft share. Also, you cannot really resell it immediately because you don't have it yet. Your broker will still allow you to do that "on margin", which basically means that he will again "advance" the cash and shares for you.
4) Since you made some gains by reselling your Microsoft shares, you can now afford a share of Amazon.
5) etc.
This is a very simplistic example. In real life these problems are amplified tenfold, because you have to add credit lines, derivatives, swaps, FX, currency settlement, etc.
But the overall idea is here: everything that seems "simple and fast" in finance is just because there are layers upon layers of local caches, that are settled together at varying time horizons, and generate mismatches that need to be handled.
A blockchain ledger is settled by construction. That is, if I want to buy Token1, sell it, use the profit to buy Token2. Then these very transactions need to be in the ledger, in this precise order, for the whole chain to be possible.
Blockchains _are_ settlement machines.
> Question is - why use blockchain? Nothing here is technically difficult with existing technology.
I think your question is backward. Blockchains do solve all the problems mentioned in my post. They solve it well, they allow any degree of customization that is possible, they solve it elegantly, and have been battle tested. Why _not_ use them?
Exactly, so what exactly is NFT trying to solve here?
> My understanding is that nobody really cares if you have a copy of a photograph or digital art. What people care about is if you pretend to own it.
It's hard to sell a painting without having possession of it, and I'd imagine, independent authentication in these high value markets.
> Well no, the NFT contains the signature of the artist that created it.
Which is absolutely worthless unless you know the signature of the artist. If you already know this, then you don't need the NFT.
Or in other words, a NFT cannot say 'this is an artwork by x' instead, x can say 'this NFT represents my artwork.'
> There is nothing to unwind on a blockchain, because you cannot create incoherent ledgers. That is why blockchains are perfect settlement systems.
What is it about blockchains that mean perfect ledgers, that cannot be solved more simply by using a ledger/database?
Either this is an easy problem - in which case this should be technically simple to solve, ormits a hard problem, in which case how are Blockchain solving it?
> Imagine the following typical simple settlement scenario:
I see and understand problem is how does Blockchain solve any of this?
That's the problem - yes it would be nice for certain use-case if this stuff was instantaneous, but that is a business problem not a technical one.
For example yes it would be nice if credit cards were instant. Problem is credit cards have chargebacks built into the design to protect the user. Making it instant and irreversible would break that promise, essentially making them a completely different offering.
> Blockchains _are_ settlement machines.
Sure, but they're aren't the only settlement machines. One can quickly replicate this settlement logic with a database and basic API - that would be far more scalable and affordable than your typical Blockchain.
Looking for technical solutions to non-technical problems doesn't work.
> Why _not_ use them?
Because they are incredibly expensive, and probably don't solve the problem you want to solve.
For example, take the settlement use-case earlier - your user wants to pay via credit card. What Blockchain has customer protections (e.g. chargebacks), incredible scale, privacy, etc... and yet fits within your ideal settlement model?
Does such a product exist? Can it exist... or are the requirements simply contradictory?
> It's hard to sell a painting without having possession of it, and I'd imagine, independent authentication in these high value markets
See that's actually not the target for NFTs as I see it.
Most people see the art world as very fancy paintings worth 6 figures. But I'm talking about the more down to earth kind of art. The kind that your average art-enthusiast buys at Art Basel, photographs or lithographies in the 100$ to 5000$ range. The bread and butter of the "art" scene.
There are no experts to certify these pieces, only for ultra high ends can you find such specialists.
Also these pieces are not worth enough to justify such certifications.
Still, it's very important for enthusiasts to own the pieces, it's part of why people buy them from the artist instead of just getting it printed. To an art enthusiast, it doesn't matter if 20000 people have that same art printed on a poster at home, or as wallpaper on their laptop. It matters to them that they have it, they bought it from the artist, and they have some physical or digital token to _feel_ their ownership.
> Which is absolutely worthless unless you know the signature of the artist. If you already know this, then you don't need the NFT.
I don't understand your point here. A typical certificate of ownership is just a paper with a sentence such as "I, Mr. X, hereby sell my photograph Y to Mr. Z". This paper is then signed by the artist, a notary, and the new owner.
An NFT contains exactly the same information, except the text is replaced by a Json containing a thumbnail of the art for reference, and signatures are digital. The fact that these tokens of ownership are on the blockchain allows you to verify them yourself, instead of going through calling the notary. And you can transfer these ownership (which ultimately link to the public key of the artist who created it) directly with a seller.
It's more auditable, cheaper for both artist and buyer, and overall simpler and faster than paper proofs of ownership.
> I see and understand problem is how does Blockchain solve any of this?
Let's get back to the example and consider that there are just 2 layers in this settlement: back office and front office.
The back office is the slow, golden record, of who owns what share. The back office is always correct, they will never allow credit or delay. If you bought a share but did not receive it yet, the back office will just consider that you don't have it, you cannot sell it.
Yet, you really want to be able to buy and sell a share in the same day. So you create a cache on top of the back office, called the front office. Technically it's quite similar to having a memcached or redis between an HTTP server and the filesystem.
This cache allows you to be faster, but the side effect is that there are now potential I coherency between what you have for real (data on the filesystem, or shares in your possession), and what you think you have (cached data in redis, or virtual ownership of shares in your front office system).
Note that the back office, or filesystem, is _never wrong_ here. The matching issues never happens at the bottom layer, only at the layers on top of it.
You memcached can be out of sync with your files if you changed them, but not the other way around.
Blockchains are like having a back office system that is fast enough to not require you to have a front office. It's like having SSDs that are as performant as your memcached. You just don't need these layers of cache anymore, and by removing them, you avoid the whole class of consistency, matching and settlement issues than happen with cache coherence.
So now let's discuss about why back office is that slow to begin with, and thus why blockchains cannot be replaced by just "having better APIs for a faster back office".
What makes back office slow is that there are always 2 parties in a trade, and none of these parties can ultimately trust the other.
You can never be sure that the party you're agreeing with to buy a share really has that share. Since back office needs to be perfect, by definition, it will have to wait until the shares are actually delivered for real to consider the trade done.
No amount of database or API can fix that _unless_ both parties of the trade share the same database of what they have and owe each other.
Such shared, centralized databases exist, at small scale, but they don't expand between more than a couple of banks or institutions.
Ultimately businesses cannot blindly trust each other, it's just not an acceptable counterparty risk when big money is at stake.
Blockchains are shared databases that allow the storage of transactions between entities that don't have to trust each other's records. As such, they are perfect settlement systems, and are fast enough to not require additional layers of caches and coherence issues.
> Because they are incredibly expensive, and probably don't solve the problem you want to solve.
You have to realize the massive scale of what back and middle office is for the financial industry. I'm not joking when I say easily half of the resources of any bank, investment company or broker are dedicated to just that.
To summarize, I think you've identified area's where there are potential problems that could be improved. However, as an engineer, blockchain wouldn't be my first (or second) tool that I'd use to solve these problems.
Using NFT's to do ownership of art is interesting, and maybe in a niche area or two people will use it, but ultimately runs into issues since it doesn't address theft, loss of wallet or other practical concerns. NFT ownership only ever proves ownership of the NFT, never of anything associated with NFT.
For trading, the main claim you make is these banks don't trust each other. Yet clearly they must trust each other, or someone, for the trade to finally finish.
All blockchain does is add in something else you must trust - but this something else is wildly complicated and complex. For example, chains can become worthless or attacked and taken over. Users can decide which chain to follow, sometimes leading to splits where assets are essentially duplicated on each side of the split.
So yes, I don't doubt that there are some serious problems that can be fixed. I just find that NFT's/Blockchain often make claims that are not unique to them but introduce problems that are hidden and glossed over.