[helen___keller]

> (and before someone yells "crypto fees", there are already solutions that can drive these well below $0.01 if we're just sending payments between 2 parties)

Can you specify what you're referring to? As someone who frequently yells "crypto fees" my understanding is most low-fee options either (a) require transfers within a centralized ecosystem, (b) aren't likely to scale effectively as a system (e.g. bitcoin's lightning network, or a shitcoin network of choice that's never experienced heavy usage) (c) are exciting but hypothetical future ideas, such as sharding on Ethereum

[xur17]

A lot of scaling solutions for Ethereum focus on being fully EVM compatible, and as such are forced to store every transaction on a widely distributed, permanent datastore, which in my mind is never going to be "cheap". Since we just need to move money around, payment channels [0] allow us to move the funds by just committing an entry and exit transaction to the widely distributed, permanent datastore, and once we "enter" this network, transactions can be very very cheap.

In terms of implementations, Bitcoin's lightning network is probably the biggest active version today, but this only allows you to move Bitcoin around, and hence is a non-starter for a widespread consumer adoption.

Which brings us to payment channels on Ethereum. It doesn't feel like there has been as much focus on the payments use case yet, and none of the solutions have as much adoption as Bitcoin's lightning network, but there are a few options I know of. Raiden network has been around for a while, visa is supposedly creating a "universal payments channel" [1], there's xdai [2] (not payment channels, and I'm not super familiar with this solution), and there's connext [3].

There's a long list here [4].

Basically it boils down to - payment channels are the solution, and it'll probably be a mix of self custodied wallets and custodial providers, all integrating with an open standard.

[0] https://hackernoon.com/how-we-built-the-largest-payment-chan...

[1] https://decrypt.co/82233/visa-universal-payment-channel-stab...

[2] https://www.xdaichain.com/

[3] https://connext.network/

[4] https://ethereum.org/en/developers/docs/scaling/state-channe...

[robcohen]

Payment/State channels exist today. Is there a difference between your (c) hypothetical future ideas and hypothetical future scaling problems? Lightning works today. Try Breez Wallet https://breez.technology/ or Strike https://strike.me/ and tell me what your experience is like.

[helen___keller]

> Is there a difference between your (c) hypothetical future ideas and hypothetical future scaling problems?

Hypothetical future ideas require development. Hypothetical future scaling problems block current development.

For example, suppose that Amazon Marketplace wanted, as soon as possible, to have all customers pay through bitcoin including through lightning network). The question is, would the average fee per transaction average 1c or less?

I claim the answer is no. Lightning channels require broadcasts (though much less frequently than 1 per merchant transaction) on the main network. Let's suppose someone does 100 transactions with amazon per 1 broadcast to the bitcoin network, which seems generous to me. A google search[0] suggests that the current average transaction fee on bitcoin main network is about 2.5 dollars, which then averages to 2.5 cents per lightning transaction with Amazon. This is close to the desired goal of 1 cent per transaction, under the listed assumptions.

However, Amazon adopting bitcoin for all transactions would drive a ridiculous amount of traffic to the bitcoin network. If we go to recent history, march-april 2021 saw about $20 average transaction cost (or more - peak at $60). Going back to our same calculations, $20 per broadcast puts us at a fee of 20c per amazon transaction, far beyond our goal of 1c per transaction. Furthermore historical data has not yet seen Bitcoin as a primary transaction network for a large commerce site like Amazon, so it's reasonable to speculate we could see per-broadcast fees average above $100.

In simpler terms, while Lightning network greatly decreases the fee per real-world transaction, it's not enough to handle large scale adoption unless ratio of real-world transactions to bitcoin broadcasts increase in direct proportion to the fees on the bitcoin network. Realistically, if a large amount of US commerce occurred on the bitcoin network, the most optimistic access pattern would have users broadcast to true-up their channel liquidity roughly in line with receiving paychecks - maybe twice a month.

The simple fact is that if you take millions of humans and have them broadcast twice a month on the bitcoin network, the bitcoin network would not be able to handle it (bitcoin volume is roughly 10 million transactions per month[1]). Forget even hoping to scale to billions of humans.

[0] https://ycharts.com/indicators/bitcoin_average_transaction_f...

[1] https://en.wikipedia.org/wiki/Bitcoin_scalability_problem#/m...