Facebook Libra Is Architecturally Unsound

What's the profit potential for Facebook here, especially above using an existing cryptocurrency for payments on their platforms?

Having low friction payments on Facebook makes sense, it build value into the platform and Facebook can capture some of that. But can't they do that with an existing cryptocurrency?

It doesn't seem like Facebook will maintain full control of the currency due to the consensus algorithm. There is power and control if Facebook continues to control the fork of the code base that everyone uses, but presumably nodes could choose to switch away from Facebook's fork. So I'm not seeing "control a currency" as a long term benefit.

It makes sense that anyone running a Libra node would make money, but anyone else running a node would make similar profit.

The article mentions that a long term goal could be "act as a data broker and mediate consumers access to credit", although again, doesn't the decentralized nature permit any node from taking those steps? That doesn't seem to uniquely advantage Facebook.

As others mention, once a cryptocurrency is "too big to fail", regulators are locked out. Is Libra really an easier approach to getting a cryptocurrency to that point, versus adopting and accelerating the growth of an existing coin (like Bitcoin)?

Does Facebook just think they can build a better cryptocurrency? I don't doubt that they can hire good engineers, but with all the politics and marketing focus on the code now, development is probably getting stressful and chaotic.

This is simply not accurate.

I build financial market infrastructure and in the last twelve months I've seen companies in my industry presenting at AWS and Splunk conferences, and a Cassandra meetup.

There are lots of great technologists in the space that are just like the rest of us and love sharing ideas. There are definitely NDAs in place, but "financial infrastructure" generally refers to processing systems, not trading systems. The former is increasingly utilized, while the latter generates alpha (i.e. where you actually make money).

People like sharing ideas about processing infrastructure because everyone benefits when the rest of the market gets better at it also. Your efficiency is only as good as your counterparties' efficiency. If their system breaks, you still have a broken trade that costs you operational time and money no matter how good your infrastructure is.

There is a lot more wrong with what this author wrote, but in the interest of time I'll keep my answer to the question asked.

I've been thinking and it actually makes a lot of sense for new ways to be created to transfer money and pay for things (though I'm highly skeptical of having it operated by Facebook and of the ad-hoc pump-and-dump-prone and whitewashing-and-tax-avoidance-friendly properties of cryptocurrencies). But when you think about it, it's insane that transferring money is something credit cards and payment processors can regularly charge a fee of 1.5-3.0% on. In order to change some numbers in a couple databases... The marginal cost of this ought to be less than pennies. The only reason they can even partially justify this is because of fraud and credit, and the costs associated with dealing with that. But what if we had a way to transfer money between entities that mandated two factor authentication to prevent fraud, and sidestepped the whole credit thing by only working if you had the money? (or required that you took the loan elsewhere, so the account that actually transfers the money does not have to deal with credit at all). Verify that the person is who they claim to be, and that they have the money needed, then do the transaction - no more, no less. No chargebacks, no credit checks, no fraud checks. Why isn't this a thing?

There are generally 3 kinds of fraud

1. A scammer cheating someone out of their life savings through social engineering

2. A central bank "unjustly" inflating currency and giving the newly printed money to specific industry/people etc

3. A person paying for merchandise with a stolen credit card or refusing payment after services/good is delivered.

Fiat solves 1 and 3 (recovery) does not think 2 is a problem.

Crypto solves 2 and 3. People say it is meant for solving 1.

I disagree that blockchain has almost no good use cases.

All a blockchain is, is a Merkle tree with a third party that ensures there is only one “main line”. That’s the use case.

All the other stuff — such as having all computers in the network watch every transaction — that’s the wasteful part. There are other ways to have a set of third party validators, that is a subset of the network watching a given merkle tree, simply says which branch is correct. SAFE network uses a Kademlia DHT with various mechanisms to ensure the validators have no say in what they watch and they need to “earn” their way into having any say about anything over time.

There are tons of useful properties, some of which are used in Merkle trees like git:

Immutability

Quick verification of tree membership using Merkle branch

Ability to download different parts from different actors (Bittorent)

Consensus and rule enforcement (eg chess game or any other evolving document)

Smart contracts and autonomous code execution

And much much more. The main problem is when people think of blockchain they think of a giant monolithic chain of blocks each of which contains ALL TRANSACTIONS IN THE NETWORK. This is wasteful.

What’s even more wasteful is when you have divisibility of the tokens, leading to an exponential growth of UTXOs and unlimited storage requirements. And each full node needs to verify the entire history of every transaction because then they get intertwined. That’s the ridiculous part.

And proof of work is the most wasteful thing of all. People need to get off of that!