LN tx's are actually bitcoin tx's. Micro ones! Proper real bitcoin segwittx's! Secured by the network. Its just they are not registered on chain until the payment channel is closed.
This is an oversimplification. LN transactions have much different properties than regular bitcoin transactions. Most notably, LN transactions require the user (or someone they trust) to be online in order to verify that they aren't being defrauded by someone broadcasting and earlier state of the payment channel.
You're right but to clarify, the user getting online once a week or so is enough. They don't have to continuously be online.
Nobody verifies the order, but if you try to broadcast an intermediate state without your counterparty's permission, you can't actually spend it for 1000 (?) blocks.
If your counterparty is able to present a proof that that was an intermediate state rather than the latest state then he gets to spend all of the money in the channel immediately.
So all you need to do is monitor for transactions that are spending old channel states, and take all of the money if your counterparty tries to do it. This is even out-sourceable, so you could have a third-party monitoring the blockchain for old channel states and spending the money to you if anything goes wrong.
There is an "off-chain" network that we are being assured is trustless, and you say nobody verifies the ordering of transactions. It just doesn't make sense. Either it requires trust or requires some alternative non-scalable consensus mechanism. There is no magic algorithm that doesn't have tradeoffs.
If bitcoin core agreed to ensure that opening a channel freezes both wallets until the channel is closed, then that would prevent one kind of fraud. But there's still other kinds of fraud, like double spending, or changing the ordering.
You say that a counterparty can "present a proof" but to whom? Isn't timestamping itself a hard problem, and one that it is said only the blockchain can solve in a trustless way?
It's a hash timelocked contract[0]. You "present the proof" by spending the UTXO. I was reading up on this quite a while ago so I'm fuzzy about the exact details, but the hashes are created as part of the LN transaction.
The transaction can be broadcast immediately by a malicious party, but won't be spendable until 1000 (?) blocks have passed.
If you can prove that there was a subsequent channel state signed by both parties (i.e. the tx that was maliciously broadcast was not the most recent state of the channel), then that can be used to spend the UTXO immediately instead of waiting for the 1000 blocks.
Like I said, do some research before asserting that it can't work.
[0] https://bitcoinmagazine.com/articles/understanding-the-light...
It sounds like this wouldn't be very useful for just one or two transactions, because to open a channel it requires a real bitcoin transaction, and then LN transaction fees on top of that.
So instead, it would only be useful if people plunk down a large amount of bitcoin in order to conduct a large amount of transactions. So you'd have a huge amount of money tied up in this network, and people would be spending it as if it were real bitcoin when it isn't, and the double spending problem is truly only solved when the channels are closed.
The only guarantee against double spending is the blockchain.
Note that when you have 1 LN channel open, that doesn't mean you can only transact with 1 person. You only need 1 channel open to get on the network, and then you can transact with everybody else on the network.
How is it not "real bitcoin"?
> the double spending problem is truly only solved when the channels are closed.
How are you going to double-spend on a LN channel? Your counterparty knows the state of the channel just as well as you do. You physically can't update the channel state without your counterparty's approval. Part of updating the channel state is revealing to your counterparty the secret hash that will allow them to steal the money from the previous state if you broadcast that previous state's HTLC instead of the updated state.
The double-spending problem is solved by it being possible to settle the channel on the blockchain. The blockchain is still required, but it's not like you can cheaply or easily double-spend a channel at any point in time before it's settled on the chain. If you try to do that, your counterparty will just take all of the money.