I don't think they're 1:1 without making the XML very bloaty. That's to say, if, as many xml libraries for lisps do, you represent (test 1 2 3) as <test>1 2 3<

I don't think they're 1:1 without making the XML very bloaty. That's to say, if, as many xml libraries for lisps do, you represent (test 1 2 3) as <test>1 2 3</test> or <test><number>1</number> ... </test> then you lose the ability to represent the null list. The way around that would be to use something like <list><element value='test'/><number value='1'/> ... </list> which can then represent the null list as <list/> but you've now made the whole thing massively unwieldy. So I can't see how it's any progression over s-expressions. Unless I suppose you break the XML syntax itself and make XML able to do things it can't at present, but then you're just reinventing s-expressions.