[RFC] Canonical input and output ordering in transactions



Summary:

In an email conversation between Rusty Russell and Peter Todd on June 15, 2015, they discussed the possibility of hiding OP_RETURN transactions to make them look like something else. Russell suggested that Todd's best-effort language could be more practical in this scenario since a non-canonical ordering wouldn't distinguish such transactions. However, Todd raised a question about commitments that don't use op_return, specifically pay2contract commitments. They also discussed ways to order multi-party transactions without any consequences for other transaction types' privacy. One solution proposed was a deterministic method that depends on a random seed previously shared by all parties involved in the transaction. This way, the ordering would be deterministic for all parties, while the order would appear random or not canonical in a recognizable way to everyone else.


Updated on: 2023-06-09T22:18:43.631383+00:00