[scintill76]

> What they should have done was waited an hour then done an O(n) scan of all transactions globally in history to find the transaction by inspecting for parameters which exactly matched the ones they provided.

Oh, you mean the scanning they have to do already, to verify "all transactions globally in history"? Inspecting all parameters on all transactions since the genesis, like you'd already have to do to verify they are not stealing or creating money from nothing? The inspection you have to do just to locate even the same transaction you submitted to the network yourself, to verify it was accepted? And you have to spend like 10 whole seconds of CPU time doing this, per ~10 minutes that a new block comes out, verifying the transactions from the last 10 minutes? Golly, that is sooooo much more onerous than just running the blockchain securely! /sarcasm

I'll agree that it's embarrassing, misleading, not documented well, and not gracefully handled by the community now that everyone points their fingers at each other. But you are deliberately making it sound worse by re-describing standard parts of the bitcoin protocol, as if they are new requirements in order to get a sane ID. Anyone writing financial software should be more than capable of quickly adding a few function hooks into the existing process to get a deterministic normalized ID, and the amount of extra computing resources is negligible compared to what you already have to do, just to use bitcoin safely.