Hacker News Reader

Comment by psychlops | Hacker News Reader

The more one learns about how traditional finance works, the more horrifying it becomes.

On a technical basis yes. Finance rests upon a foundation of people emailing around spreadsheets and trust. Much better than new innovative cryptocurrencies that are just a trojan, wallet address typo, or exit scam away from losing your savings.

As trusting as the traditional finance system is, it seems they still experience exit scams, wire typos, and countless other scams that lose savings.

Crypto has a long way to go, but tradfi hasn't solved its own problems. Until then, I guess we'd all better hope our bank representatives have a lot of friends in the industry.

Crypto can't solve its main problem though; it's fundamentally based around not being able to reverse transactions, which means you can't reverse a mistaken or fraudulent transaction, which is the whole reason tradfi works. And I don't think the other stuff makes up for this, even if programmability and flash loans are interesting.

(I mean interesting mostly in a bad way.)

There's nothing in crypto that inherently prevents transaction reversals.

This can be implemented as a smart contract on any sufficiently advanced blockchain (definitely Ethereum, not sure about Bitcoin).

The way this works is that you put the money in an escrow contract first, designating a trusted third party as an arbiter. If you don't dispute the transaction for a few weeks, the money goes through irreversibly. If you do, the contract "locks up", and the third party can now either decide to send the money on to its destination or return it back to you. The contract is designed in such a way that those are the only things that third party can do with your money, they have no way of taking it for themselves. Either way, they get a small cut for the trouble of adjudicating the transaction.

If what you're worried about is wallet hacks, not disingenuous merchants, that's doable too. Instead of storing your money directly in your wallet, you'd have to store it in a smart contract, designed in such a way that your key is the only one that can move money out of it. However, limitations would be placed on the contract with regards to how much money can be moved daily. Such limits would likely be different for transfers to trusted escrow contracts and for irreversible transactions. If you needed to move more money, you could designate a trusted third party (let's say a traditional bank doing traditional bank things for identity verification) as being able to override these limits for you. Again, that would be their only responsibility, they would not have the right to move any of your money without your permission. This essentially relegates a bank to the role of an identity verifier who cannot take control of your money in any possible way except by colluding with a hacker who already has access to your wallet.

THe reason none of this exists is ecosystems and network effects, not technical limitations.

Or, just like tradfi, point out the mistake to the counterparty. Request your crypto back and the transaction is reversed.

In tradfi, the counterparty is typically a sophisticated financial institution that either is in your jurisdiction or at least cares somewhat about not annoying your government. If it isn't, it's probably a customer of said financial institution, and thus subject to their rules. In crypto, your counterparty might be literally anybody.

That just shifts the problem around: if there's a bug or mistake in the smart contract itself, then you face the problem that you can't reverse that smart contract.

Also, you can't patch bugs in the smart contract because people can spot the patch transactions and outbid you to exploit it instead.

Congratulations, you've just invented a bank.

I have emphasized the main difference between this and a bank multiple times, but if it's not sufficiently clear: A bank holds your money. It can do whatever it wants or is forced to with that money, including seizing it if you're Canadian and attend a protest[1]. In the system I outlined, a bank can only decide whether a transaction you dispute should be reversed. As long as you don't dispute any transactions, the bank has no ability to take your money.

[1] https://www.newsweek.com/banks-have-begun-freezing-accounts-...

But then I can't spend my money for several weeks - the feature that crypto can't implement is having the reversal window be wider than the lockout window.

> There's nothing in crypto that inherently prevents transaction reversals.

Irreversibility was a key design parameter of bitcoin.

You can kludge a reversal system on top, much as you can build an SQL-a-like atop Mongo. Or you could not use a system whose literal point is the opposite.

Of the many crypto problems I can think of, reversible transactions seems low on the list. Tradfi "works" because there is no other choice given so people make it functional and it is comfortable through familiarity.

People conveniently forget the rampant scams, identity theft, and numerous types of fraud that are clearly irreversible in tradfi. Then they present it as a new problem that crypto must solve as well as being the same as the old system.

Nothing is irreversible in tradfi, regulations and the court system will find someone to give it back to you.

That's what insurance is for, for instance.

Regulations and the court system also works for crypto.

Regulations don't because the bank regulations don't apply.

FWIW, "wallet address typo" should never be a thing.

There's a serious checksum in each of the BTC wallet formats that makes the chances of any random set of typos extremely unlikely to be valid.

A mis-pasted (but valid) wallet address could happen.

Not to minimize the myriad other ways to easily lose, or lose control of, cryptocurrency!

Bitcoin does, but Ethereum doesn’t, at least not originally.

There is some way to encode a checksum in the capitalization of the letters/digits A-F, but I’m not sure how ubiquitously supported that is. It seems like a pretty bad hack in any case.

Fascinating, I am unaware of the inner workings of Ethereum.

It boggles that they could make such a massive design error, especially with the Bitcoin examples staring them in the face.

I believe the idea was that “nobody will use raw low level addresses, so we don’t need to make them usable”.

But ENS is now prohibitively expensive in terms of fees and also not universally supported, so here we are.

One of my early lessons was realizing the spreadsheet folks have their own backup systems and they're fairly ingenious and orthogonal to my own concerns.

We certainly have the ability to implement transfer authorization, from the sender side, using good authentication technology. This doesn’t have to be cryptocurrency: it can just be the same techniques that we use to authenticate people logging into their bank account, eg passwords, MFA, passkeys, Face ID. We don’t do that for a bunch of business reasons given in the article (mostly that sending institutions are disincentivized from enabling this, and regulators have come up with a kludge that spreads the losses out in weird ways.) The result is that we use a system based on trust, reversal and insurance. This seems fine, but it almost certainly raises costs across the entire system in ways that we can’t obviously measure. Those costs presumably show up in places that typical consumers can’t really see, unless they’re very sophisticated: for example, higher management costs and worse spreads and weird fees.

Insomuch as this is a problem, I also don’t think it will change materially any time soon.

A robust financial system has to be reasonably accessible to such a wide range of people and institutions with varying capabilities that even building something that caters exclusively to the technically inclined is all but impossible.

I wrote on here before about how Vanguard used to offer a FIDO2-only MFA option. You could disable SMS MFA. That eventually got removed because the mobile app never supported it, so now they’ve enforced SMS as a mandatory MFA factor. And FIDO2 has been fairly easy to implement on mobile for years before Passkeys specifically were pushed.

I was going to make a dumb joke about how this was the plot to a horror story or movie or something. I paused for a few seconds (that's discouraged around here, so it better be a good joke)... and it started to sink in that it could well be the plot not only to a horror movie, but maybe has the potential to be a compelling one. "You'll be made whole, eventually" sounds great, until you think back to all those times in your life where "in a few weeks" would've ruined you. Even death's not out of the question if it pushes someone over the edge into suicide territory. The lunatics really do run the fucking asylum.

What’s horrifying about being able to quickly fix a honest mistake in mutual agreement? It’s not always possible, but when it is, it’s so much more efficient than the alternatives.

Trust isn’t a given and not always warranted, but when it is, it is a feature, not a bug.

So much of the world essentially runs on trust, it’s amazing that it works at all, let alone as well as it does.

Reply on news.ycombinator.com