The argument in favor of microformats is that it’s an easy hack to add to your pages to make them a bit more semantic. But:

“Easy” doesn’t matter if you continue to pay that cost every time you want to modify a layout. It’s just fundamentally wrong to contort HTML to pretend it can also be a medium for such data.

“Easy” doesn’t matter if validating it and bugfixing is hard. Proponents may point to the tools now available for validation but now they’re contradicting themselves. If it’s an easy hack, why do I need fancy validation tools? The theory here is that microformats are for authors who don’t normally grok XML, but suddenly they do need to run parsers and validators?

We have a simple method available today for related content, or different versions of the same content: the LINK and A (anchor) elements. LINK works just fine for things like RSS. You never have to update your RSS when you want to change the visual layout. LINKed documents can signal what they are about with the REL attribute, and the actual document can use the standard HTTP headers for describing its format. The only flaw with LINK is that it applies to the whole document, rather than various DIVs and so on in the document. But very similar tricks could be done with the A element – it also can use the REL attribute. An empty anchor has no presentation in HTML anyway, so you don’t even have to update your CSS.

The microformat community could just focus on getting better browser support for that kind of linking and save about 90% of the effort.