1) A publisher accidentally uses a different, though very similar, URI for their id. They then apply the canonicalization rules and the error is erased. This will only work if they remember to apply the rules, if they accidentally chose a similar enough id for the rules to apply, and more bizarrely if they change the code that generates ids regularly, or fill out the id field themselves by hand from memory each time it is requested (the mark of a true blogger). This is not a problem that needs solving or that c14n solves.
2) An intermediary automatically c14nizes all URIs it processes. If URIs come pre-c14nized from the publisher, this won't do any damage. This is valid, but the problem is that these intermediaries are currently imaginary. A much better way to solve it would be to write, at the end of the comparison section:
For this reason, intermediaries should be careful not to canonicalize or otherwise modify Identity constructs they
process.
Problem solved, much more efficiently, much more reliably (since the current method breaks if one blogger don't remember to c14 their feed), and much less wordily.
Graham
smime.p7s
Description: S/MIME cryptographic signature
