How to make a bitcoin transaction and pay less than 1 sat/vByte

[Pmalek]

If your transaction is not signed with SIGHASH_ALL, then it matters, because other nodes can change it, and batch many transactions into one, or solve your "sighash puzzle", and broadcast it on-chain.

Do you know of cases where mining nodes have done that? I mean, for regular people like you and me and not with their own transactions or those from people connected to them in some ways. A mining node can include any valid transaction into a block they mine, but why would they do that with mine transaction? There are other transactions already paying higher fees waiting for confirmations, so why waste time to batch below 0.1 sat/vByte transactions into one?

[stwenhao]

Do you know of cases where mining nodes have done that?

Miners can include the final transaction, with high fees, that will be broadcasted by the last user. They don't have to participate in transaction batching, if they don't want to.

Then, it is all about regular, non-mining nodes, and their ability to find transactions, which could be open for modifications, and joining them, and then broadcasting a different version through full-RBF.

A mining node can include any valid transaction into a block they mine, but why would they do that with mine transaction?

Because by batching transactions, you can decrease their on-chain size. Which means, that you have exactly the same fees, but feerate can increase, if the final transaction will take less bytes than before.

why waste time to batch below 0.1 sat/vByte transactions into one?

Because then, users can pay for example 0.09 sat/vByte, 0.08 sat/vByte, or even lower fee rates, and after batching, they will see it as 0.1 sat/vByte transaction, for all participants. In general, I still think that going below 1 sat/vByte is a mistake in the long-term, because people should be focused on batching instead. Then, fees per user can be low, but fee per transaction can be higher, if enough people will join, and if enough things will be stripped (if you have Alice -> Bob -> Charlie, you don't have to put Bob on-chain, as long as coins are flying in the way he wanted).

[Pmalek]

Miners can include the final transaction, with high fees, that will be broadcasted by the last user. They don't have to participate in transaction batching, if they don't want to.

Then, it is all about regular, non-mining nodes, and their ability to find transactions, which could be open for modifications, and joining them, and then broadcasting a different version through full-RBF.

This all sounds like a theory based on the technical capabilities of what the Bitcoin network can do. But as I asked earlier, does this happen in practice? Do you have proof and examples of transactions where individuals engaged in batching and/or modifying transactions with the aim of decreasing their on-chain size? Are there forum threads, communities or blog posts where this approach is explained in more detail and where people have testified that it works?

[LoyceV]

examples of transactions where individuals engaged in batching

Isn't that what coinjoin does?

[Pmalek]

Isn't that what coinjoin does?

If you just mean the batching part, then perhaps. But the entire thing stwenhao talks about isn't a privacy-increasing way of transacting. The primary goal isn't to achieve a higher level of privacy. The idea is for the miners to include transactions that pay more fees in their blocks and save block space, while each individual participant pays less than what they would do on their own.

A coinjoin is also about batching inputs but for the purpose of masking which output belongs to whom. I don't know enough about coinjoins to be able to understand if what stwenhao talks about can be considered a type of coinjoin. Or if his batching and fee & block space preserving technique works in practice.